michaelc0des's picture
Upload checkpoint-12462
56a009e verified
Raw
History Blame Contribute Delete
94 kB
{
"best_global_step": 12462,
"best_metric": 2.547119379043579,
"best_model_checkpoint": "outputs/chess-explainer/checkpoint-12462",
"epoch": 6.0,
"eval_steps": 500,
"global_step": 12462,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00048146364949446316,
"grad_norm": 650.5328369140625,
"learning_rate": 0.0,
"loss": 18.217763900756836,
"step": 1
},
{
"epoch": 0.01203659123736158,
"grad_norm": 88.18751525878906,
"learning_rate": 7.692307692307694e-06,
"loss": 14.24288813273112,
"step": 25
},
{
"epoch": 0.02407318247472316,
"grad_norm": 22.323110580444336,
"learning_rate": 1.5705128205128205e-05,
"loss": 7.188988037109375,
"step": 50
},
{
"epoch": 0.036109773712084736,
"grad_norm": 14.299266815185547,
"learning_rate": 2.3717948717948718e-05,
"loss": 4.387466735839844,
"step": 75
},
{
"epoch": 0.04814636494944632,
"grad_norm": 13.159931182861328,
"learning_rate": 3.1730769230769234e-05,
"loss": 3.8534796142578127,
"step": 100
},
{
"epoch": 0.0601829561868079,
"grad_norm": 16.505720138549805,
"learning_rate": 3.974358974358974e-05,
"loss": 3.6811083984375,
"step": 125
},
{
"epoch": 0.07221954742416947,
"grad_norm": 14.32520866394043,
"learning_rate": 4.775641025641026e-05,
"loss": 3.5446063232421876,
"step": 150
},
{
"epoch": 0.08425613866153106,
"grad_norm": 15.38774299621582,
"learning_rate": 5.576923076923077e-05,
"loss": 3.47775390625,
"step": 175
},
{
"epoch": 0.09629272989889263,
"grad_norm": 11.126094818115234,
"learning_rate": 6.378205128205128e-05,
"loss": 3.4696426391601562,
"step": 200
},
{
"epoch": 0.10832932113625421,
"grad_norm": 10.559870719909668,
"learning_rate": 7.17948717948718e-05,
"loss": 3.4579669189453126,
"step": 225
},
{
"epoch": 0.1203659123736158,
"grad_norm": 11.133081436157227,
"learning_rate": 7.980769230769231e-05,
"loss": 3.359324951171875,
"step": 250
},
{
"epoch": 0.13240250361097738,
"grad_norm": 10.544161796569824,
"learning_rate": 8.782051282051283e-05,
"loss": 3.298221435546875,
"step": 275
},
{
"epoch": 0.14443909484833894,
"grad_norm": 8.630677223205566,
"learning_rate": 9.583333333333334e-05,
"loss": 3.361368408203125,
"step": 300
},
{
"epoch": 0.15647568608570053,
"grad_norm": 8.548908233642578,
"learning_rate": 0.00010384615384615386,
"loss": 3.345328674316406,
"step": 325
},
{
"epoch": 0.16851227732306212,
"grad_norm": 8.20425796508789,
"learning_rate": 0.00011185897435897436,
"loss": 3.280472717285156,
"step": 350
},
{
"epoch": 0.18054886856042368,
"grad_norm": 9.912631034851074,
"learning_rate": 0.00011987179487179487,
"loss": 3.3183026123046875,
"step": 375
},
{
"epoch": 0.19258545979778527,
"grad_norm": 9.256693840026855,
"learning_rate": 0.00012788461538461537,
"loss": 3.2713458251953127,
"step": 400
},
{
"epoch": 0.20462205103514686,
"grad_norm": 8.893656730651855,
"learning_rate": 0.0001358974358974359,
"loss": 3.2640109252929688,
"step": 425
},
{
"epoch": 0.21665864227250842,
"grad_norm": 8.918048858642578,
"learning_rate": 0.0001439102564102564,
"loss": 3.2969732666015625,
"step": 450
},
{
"epoch": 0.22869523350987,
"grad_norm": 8.13305377960205,
"learning_rate": 0.00015192307692307692,
"loss": 3.25054443359375,
"step": 475
},
{
"epoch": 0.2407318247472316,
"grad_norm": 7.971743583679199,
"learning_rate": 0.00015993589743589745,
"loss": 3.234297790527344,
"step": 500
},
{
"epoch": 0.25276841598459315,
"grad_norm": 7.267224311828613,
"learning_rate": 0.00016794871794871796,
"loss": 3.2479425048828126,
"step": 525
},
{
"epoch": 0.26480500722195477,
"grad_norm": 8.255003929138184,
"learning_rate": 0.00017596153846153846,
"loss": 3.2425189208984375,
"step": 550
},
{
"epoch": 0.27684159845931633,
"grad_norm": 7.8051042556762695,
"learning_rate": 0.00018397435897435897,
"loss": 3.2422137451171875,
"step": 575
},
{
"epoch": 0.2888781896966779,
"grad_norm": 8.21413516998291,
"learning_rate": 0.0001919871794871795,
"loss": 3.2551699829101564,
"step": 600
},
{
"epoch": 0.3009147809340395,
"grad_norm": 7.534178733825684,
"learning_rate": 0.0002,
"loss": 3.2088958740234377,
"step": 625
},
{
"epoch": 0.31295137217140107,
"grad_norm": 7.962012767791748,
"learning_rate": 0.00019999924007355532,
"loss": 3.2368560791015626,
"step": 650
},
{
"epoch": 0.3249879634087626,
"grad_norm": 7.770707130432129,
"learning_rate": 0.00019999696030577107,
"loss": 3.2101596069335936,
"step": 675
},
{
"epoch": 0.33702455464612424,
"grad_norm": 7.733644008636475,
"learning_rate": 0.00019999316073129634,
"loss": 3.2300433349609374,
"step": 700
},
{
"epoch": 0.3490611458834858,
"grad_norm": 7.780461311340332,
"learning_rate": 0.00019998784140787908,
"loss": 3.2330572509765627,
"step": 725
},
{
"epoch": 0.36109773712084736,
"grad_norm": 6.770570278167725,
"learning_rate": 0.00019998100241636516,
"loss": 3.2057620239257814,
"step": 750
},
{
"epoch": 0.373134328358209,
"grad_norm": 5.945966720581055,
"learning_rate": 0.0001999726438606972,
"loss": 3.2183990478515625,
"step": 775
},
{
"epoch": 0.38517091959557054,
"grad_norm": 7.591160774230957,
"learning_rate": 0.000199962765867913,
"loss": 3.191751708984375,
"step": 800
},
{
"epoch": 0.3972075108329321,
"grad_norm": 7.020476341247559,
"learning_rate": 0.00019995136858814346,
"loss": 3.181463623046875,
"step": 825
},
{
"epoch": 0.4092441020702937,
"grad_norm": 7.284232139587402,
"learning_rate": 0.00019993845219461046,
"loss": 3.1981622314453126,
"step": 850
},
{
"epoch": 0.4212806933076553,
"grad_norm": 6.910461902618408,
"learning_rate": 0.00019992401688362422,
"loss": 3.14352294921875,
"step": 875
},
{
"epoch": 0.43331728454501683,
"grad_norm": 6.141090393066406,
"learning_rate": 0.00019990806287458024,
"loss": 3.1020236206054688,
"step": 900
},
{
"epoch": 0.44535387578237845,
"grad_norm": 6.360198497772217,
"learning_rate": 0.00019989059040995594,
"loss": 3.233653564453125,
"step": 925
},
{
"epoch": 0.45739046701974,
"grad_norm": 6.173579692840576,
"learning_rate": 0.00019987159975530712,
"loss": 3.1653778076171877,
"step": 950
},
{
"epoch": 0.46942705825710157,
"grad_norm": 6.959299087524414,
"learning_rate": 0.00019985109119926377,
"loss": 3.1667141723632812,
"step": 975
},
{
"epoch": 0.4814636494944632,
"grad_norm": 6.9440107345581055,
"learning_rate": 0.00019982906505352577,
"loss": 3.1662142944335936,
"step": 1000
},
{
"epoch": 0.49350024073182475,
"grad_norm": 7.002237319946289,
"learning_rate": 0.00019980552165285815,
"loss": 3.1565814208984375,
"step": 1025
},
{
"epoch": 0.5055368319691863,
"grad_norm": 6.859927177429199,
"learning_rate": 0.00019978046135508596,
"loss": 3.177081298828125,
"step": 1050
},
{
"epoch": 0.5175734232065479,
"grad_norm": 6.390033721923828,
"learning_rate": 0.00019975388454108884,
"loss": 3.1408489990234374,
"step": 1075
},
{
"epoch": 0.5296100144439095,
"grad_norm": 6.612856864929199,
"learning_rate": 0.00019972579161479524,
"loss": 3.125029296875,
"step": 1100
},
{
"epoch": 0.5416466056812711,
"grad_norm": 6.74856424331665,
"learning_rate": 0.0001996961830031764,
"loss": 3.1270068359375,
"step": 1125
},
{
"epoch": 0.5536831969186327,
"grad_norm": 5.670539379119873,
"learning_rate": 0.00019966505915623963,
"loss": 3.079051208496094,
"step": 1150
},
{
"epoch": 0.5657197881559942,
"grad_norm": 6.464207649230957,
"learning_rate": 0.00019963242054702157,
"loss": 3.1350054931640625,
"step": 1175
},
{
"epoch": 0.5777563793933558,
"grad_norm": 5.980476379394531,
"learning_rate": 0.0001995982676715811,
"loss": 3.1213824462890627,
"step": 1200
},
{
"epoch": 0.5897929706307173,
"grad_norm": 6.767940044403076,
"learning_rate": 0.00019956260104899166,
"loss": 3.107749328613281,
"step": 1225
},
{
"epoch": 0.601829561868079,
"grad_norm": 6.736208915710449,
"learning_rate": 0.00019952542122133346,
"loss": 3.1330535888671873,
"step": 1250
},
{
"epoch": 0.6138661531054406,
"grad_norm": 6.237167835235596,
"learning_rate": 0.00019948672875368524,
"loss": 3.090823974609375,
"step": 1275
},
{
"epoch": 0.6259027443428021,
"grad_norm": 5.9626288414001465,
"learning_rate": 0.00019944652423411547,
"loss": 3.068702392578125,
"step": 1300
},
{
"epoch": 0.6379393355801637,
"grad_norm": 6.667422294616699,
"learning_rate": 0.00019940480827367377,
"loss": 3.1129641723632813,
"step": 1325
},
{
"epoch": 0.6499759268175253,
"grad_norm": 6.018889427185059,
"learning_rate": 0.0001993615815063814,
"loss": 3.0886264038085938,
"step": 1350
},
{
"epoch": 0.6620125180548868,
"grad_norm": 5.504490375518799,
"learning_rate": 0.00019931684458922159,
"loss": 3.0925244140625,
"step": 1375
},
{
"epoch": 0.6740491092922485,
"grad_norm": 6.392239093780518,
"learning_rate": 0.00019927059820212969,
"loss": 3.067706298828125,
"step": 1400
},
{
"epoch": 0.68608570052961,
"grad_norm": 5.7260637283325195,
"learning_rate": 0.0001992228430479827,
"loss": 3.1082818603515623,
"step": 1425
},
{
"epoch": 0.6981222917669716,
"grad_norm": 5.574620723724365,
"learning_rate": 0.00019917357985258877,
"loss": 3.0852099609375,
"step": 1450
},
{
"epoch": 0.7101588830043332,
"grad_norm": 5.973651885986328,
"learning_rate": 0.00019912280936467598,
"loss": 3.032557373046875,
"step": 1475
},
{
"epoch": 0.7221954742416947,
"grad_norm": 6.408794403076172,
"learning_rate": 0.00019907053235588103,
"loss": 3.0586279296875,
"step": 1500
},
{
"epoch": 0.7342320654790563,
"grad_norm": 5.924404144287109,
"learning_rate": 0.00019901674962073757,
"loss": 3.0331820678710937,
"step": 1525
},
{
"epoch": 0.746268656716418,
"grad_norm": 6.371328830718994,
"learning_rate": 0.00019896146197666408,
"loss": 3.044822998046875,
"step": 1550
},
{
"epoch": 0.7583052479537795,
"grad_norm": 5.798267841339111,
"learning_rate": 0.00019890467026395137,
"loss": 3.086248779296875,
"step": 1575
},
{
"epoch": 0.7703418391911411,
"grad_norm": 6.200769424438477,
"learning_rate": 0.00019884637534574995,
"loss": 3.0701373291015623,
"step": 1600
},
{
"epoch": 0.7823784304285026,
"grad_norm": 5.729808807373047,
"learning_rate": 0.00019878657810805682,
"loss": 3.033974914550781,
"step": 1625
},
{
"epoch": 0.7944150216658642,
"grad_norm": 5.3061699867248535,
"learning_rate": 0.00019872527945970206,
"loss": 3.023597412109375,
"step": 1650
},
{
"epoch": 0.8064516129032258,
"grad_norm": 6.059074878692627,
"learning_rate": 0.00019866248033233486,
"loss": 3.038387451171875,
"step": 1675
},
{
"epoch": 0.8184882041405874,
"grad_norm": 5.904861927032471,
"learning_rate": 0.00019859818168040962,
"loss": 3.0605734252929686,
"step": 1700
},
{
"epoch": 0.830524795377949,
"grad_norm": 7.235841751098633,
"learning_rate": 0.00019853238448117128,
"loss": 3.0388153076171873,
"step": 1725
},
{
"epoch": 0.8425613866153105,
"grad_norm": 5.9548869132995605,
"learning_rate": 0.00019846508973464047,
"loss": 3.052736511230469,
"step": 1750
},
{
"epoch": 0.8545979778526721,
"grad_norm": 6.491075038909912,
"learning_rate": 0.00019839629846359828,
"loss": 3.04111572265625,
"step": 1775
},
{
"epoch": 0.8666345690900337,
"grad_norm": 6.274312496185303,
"learning_rate": 0.0001983260117135709,
"loss": 3.0903143310546874,
"step": 1800
},
{
"epoch": 0.8786711603273952,
"grad_norm": 6.714303016662598,
"learning_rate": 0.0001982542305528135,
"loss": 3.0477191162109376,
"step": 1825
},
{
"epoch": 0.8907077515647569,
"grad_norm": 6.037519931793213,
"learning_rate": 0.00019818095607229412,
"loss": 3.0680242919921876,
"step": 1850
},
{
"epoch": 0.9027443428021185,
"grad_norm": 6.358810901641846,
"learning_rate": 0.00019810618938567712,
"loss": 3.0358065795898437,
"step": 1875
},
{
"epoch": 0.91478093403948,
"grad_norm": 5.641435623168945,
"learning_rate": 0.00019802993162930607,
"loss": 3.0244366455078127,
"step": 1900
},
{
"epoch": 0.9268175252768416,
"grad_norm": 6.10072135925293,
"learning_rate": 0.00019795218396218673,
"loss": 3.0294232177734375,
"step": 1925
},
{
"epoch": 0.9388541165142031,
"grad_norm": 5.697592735290527,
"learning_rate": 0.00019787294756596927,
"loss": 2.9873196411132814,
"step": 1950
},
{
"epoch": 0.9508907077515648,
"grad_norm": 5.733670711517334,
"learning_rate": 0.0001977922236449303,
"loss": 3.004986572265625,
"step": 1975
},
{
"epoch": 0.9629272989889264,
"grad_norm": 6.309650897979736,
"learning_rate": 0.00019771001342595471,
"loss": 3.00642578125,
"step": 2000
},
{
"epoch": 0.9749638902262879,
"grad_norm": 6.235594749450684,
"learning_rate": 0.0001976263181585169,
"loss": 2.9961761474609374,
"step": 2025
},
{
"epoch": 0.9870004814636495,
"grad_norm": 6.950772285461426,
"learning_rate": 0.00019754113911466176,
"loss": 3.034189453125,
"step": 2050
},
{
"epoch": 0.999037072701011,
"grad_norm": 5.884391784667969,
"learning_rate": 0.0001974544775889855,
"loss": 3.075783996582031,
"step": 2075
},
{
"epoch": 1.0,
"eval_loss": 2.9775943756103516,
"eval_runtime": 9.4192,
"eval_samples_per_second": 159.249,
"eval_steps_per_second": 4.99,
"step": 2077
},
{
"epoch": 1.0110736639383726,
"grad_norm": 5.676021575927734,
"learning_rate": 0.0001973663348986158,
"loss": 2.86690673828125,
"step": 2100
},
{
"epoch": 1.0231102551757343,
"grad_norm": 6.018618106842041,
"learning_rate": 0.00019727671238319188,
"loss": 2.8802978515625,
"step": 2125
},
{
"epoch": 1.0351468464130957,
"grad_norm": 6.222283840179443,
"learning_rate": 0.00019718561140484413,
"loss": 2.8613705444335937,
"step": 2150
},
{
"epoch": 1.0471834376504574,
"grad_norm": 5.6597208976745605,
"learning_rate": 0.0001970930333481734,
"loss": 2.856763916015625,
"step": 2175
},
{
"epoch": 1.059220028887819,
"grad_norm": 6.7447662353515625,
"learning_rate": 0.00019699897962022997,
"loss": 2.9224319458007812,
"step": 2200
},
{
"epoch": 1.0712566201251805,
"grad_norm": 7.006275653839111,
"learning_rate": 0.0001969034516504922,
"loss": 2.93245361328125,
"step": 2225
},
{
"epoch": 1.0832932113625422,
"grad_norm": 5.889551162719727,
"learning_rate": 0.00019680645089084454,
"loss": 2.875589599609375,
"step": 2250
},
{
"epoch": 1.0953298025999036,
"grad_norm": 6.063162326812744,
"learning_rate": 0.000196707978815556,
"loss": 2.874587097167969,
"step": 2275
},
{
"epoch": 1.1073663938372653,
"grad_norm": 6.2372145652771,
"learning_rate": 0.00019660803692125715,
"loss": 2.8780206298828124,
"step": 2300
},
{
"epoch": 1.1194029850746268,
"grad_norm": 6.315883159637451,
"learning_rate": 0.00019650662672691783,
"loss": 2.8887982177734375,
"step": 2325
},
{
"epoch": 1.1314395763119884,
"grad_norm": 6.411891937255859,
"learning_rate": 0.0001964037497738238,
"loss": 2.9513558959960937,
"step": 2350
},
{
"epoch": 1.14347616754935,
"grad_norm": 6.806637763977051,
"learning_rate": 0.00019629940762555336,
"loss": 2.8823919677734375,
"step": 2375
},
{
"epoch": 1.1555127587867116,
"grad_norm": 5.565702438354492,
"learning_rate": 0.0001961936018679537,
"loss": 2.8928656005859374,
"step": 2400
},
{
"epoch": 1.1675493500240732,
"grad_norm": 5.8602294921875,
"learning_rate": 0.00019608633410911667,
"loss": 2.9296820068359377,
"step": 2425
},
{
"epoch": 1.1795859412614347,
"grad_norm": 7.057725429534912,
"learning_rate": 0.0001959776059793544,
"loss": 2.926005859375,
"step": 2450
},
{
"epoch": 1.1916225324987963,
"grad_norm": 6.5474629402160645,
"learning_rate": 0.00019586741913117458,
"loss": 2.8745428466796876,
"step": 2475
},
{
"epoch": 1.2036591237361578,
"grad_norm": 5.752487659454346,
"learning_rate": 0.0001957557752392551,
"loss": 2.9059454345703126,
"step": 2500
},
{
"epoch": 1.2156957149735195,
"grad_norm": 6.054665565490723,
"learning_rate": 0.0001956426760004189,
"loss": 2.9109234619140625,
"step": 2525
},
{
"epoch": 1.2277323062108811,
"grad_norm": 5.9463276863098145,
"learning_rate": 0.0001955281231336081,
"loss": 2.8662490844726562,
"step": 2550
},
{
"epoch": 1.2397688974482426,
"grad_norm": 7.36589241027832,
"learning_rate": 0.00019541211837985768,
"loss": 2.9000286865234375,
"step": 2575
},
{
"epoch": 1.2518054886856043,
"grad_norm": 6.027646064758301,
"learning_rate": 0.0001952946635022693,
"loss": 2.8943441772460936,
"step": 2600
},
{
"epoch": 1.263842079922966,
"grad_norm": 5.916275978088379,
"learning_rate": 0.00019517576028598425,
"loss": 2.899263916015625,
"step": 2625
},
{
"epoch": 1.2758786711603274,
"grad_norm": 6.0202789306640625,
"learning_rate": 0.00019505541053815653,
"loss": 2.925803527832031,
"step": 2650
},
{
"epoch": 1.287915262397689,
"grad_norm": 6.8593339920043945,
"learning_rate": 0.00019493361608792528,
"loss": 2.9056402587890626,
"step": 2675
},
{
"epoch": 1.2999518536350505,
"grad_norm": 5.5273661613464355,
"learning_rate": 0.00019481037878638695,
"loss": 2.877239990234375,
"step": 2700
},
{
"epoch": 1.3119884448724122,
"grad_norm": 7.119853973388672,
"learning_rate": 0.0001946857005065672,
"loss": 2.85099853515625,
"step": 2725
},
{
"epoch": 1.3240250361097736,
"grad_norm": 5.379928112030029,
"learning_rate": 0.0001945595831433925,
"loss": 2.9064642333984376,
"step": 2750
},
{
"epoch": 1.3360616273471353,
"grad_norm": 6.6639604568481445,
"learning_rate": 0.00019443202861366124,
"loss": 2.882684326171875,
"step": 2775
},
{
"epoch": 1.348098218584497,
"grad_norm": 6.571842193603516,
"learning_rate": 0.00019430303885601465,
"loss": 2.933927307128906,
"step": 2800
},
{
"epoch": 1.3601348098218584,
"grad_norm": 6.048442363739014,
"learning_rate": 0.00019417261583090723,
"loss": 2.8845724487304687,
"step": 2825
},
{
"epoch": 1.37217140105922,
"grad_norm": 6.913212776184082,
"learning_rate": 0.00019404076152057713,
"loss": 2.92847412109375,
"step": 2850
},
{
"epoch": 1.3842079922965815,
"grad_norm": 6.1456122398376465,
"learning_rate": 0.0001939074779290159,
"loss": 2.864238586425781,
"step": 2875
},
{
"epoch": 1.3962445835339432,
"grad_norm": 6.314334869384766,
"learning_rate": 0.00019377276708193802,
"loss": 2.93041015625,
"step": 2900
},
{
"epoch": 1.4082811747713047,
"grad_norm": 6.124015808105469,
"learning_rate": 0.00019363663102675024,
"loss": 2.8582913208007814,
"step": 2925
},
{
"epoch": 1.4203177660086663,
"grad_norm": 5.999044895172119,
"learning_rate": 0.0001934990718325203,
"loss": 2.9177670288085937,
"step": 2950
},
{
"epoch": 1.432354357246028,
"grad_norm": 5.440136432647705,
"learning_rate": 0.00019336009158994562,
"loss": 2.942555236816406,
"step": 2975
},
{
"epoch": 1.4443909484833894,
"grad_norm": 6.6183929443359375,
"learning_rate": 0.00019321969241132138,
"loss": 2.93353515625,
"step": 3000
},
{
"epoch": 1.4564275397207511,
"grad_norm": 6.443948268890381,
"learning_rate": 0.00019307787643050857,
"loss": 2.914281311035156,
"step": 3025
},
{
"epoch": 1.4684641309581128,
"grad_norm": 6.57510232925415,
"learning_rate": 0.0001929346458029015,
"loss": 2.919800109863281,
"step": 3050
},
{
"epoch": 1.4805007221954742,
"grad_norm": 5.723294734954834,
"learning_rate": 0.00019279000270539497,
"loss": 2.875799255371094,
"step": 3075
},
{
"epoch": 1.4925373134328357,
"grad_norm": 6.246444225311279,
"learning_rate": 0.0001926439493363513,
"loss": 2.853829040527344,
"step": 3100
},
{
"epoch": 1.5045739046701974,
"grad_norm": 5.6569743156433105,
"learning_rate": 0.00019249648791556684,
"loss": 2.845904541015625,
"step": 3125
},
{
"epoch": 1.516610495907559,
"grad_norm": 5.917844295501709,
"learning_rate": 0.00019234762068423824,
"loss": 2.8845635986328126,
"step": 3150
},
{
"epoch": 1.5286470871449205,
"grad_norm": 7.2379302978515625,
"learning_rate": 0.00019219734990492843,
"loss": 2.8919659423828126,
"step": 3175
},
{
"epoch": 1.5406836783822822,
"grad_norm": 6.368435859680176,
"learning_rate": 0.00019204567786153213,
"loss": 2.8638519287109374,
"step": 3200
},
{
"epoch": 1.5527202696196438,
"grad_norm": 6.432679176330566,
"learning_rate": 0.0001918926068592414,
"loss": 2.8769232177734376,
"step": 3225
},
{
"epoch": 1.5647568608570053,
"grad_norm": 6.268744945526123,
"learning_rate": 0.00019173813922451015,
"loss": 2.88654296875,
"step": 3250
},
{
"epoch": 1.5767934520943667,
"grad_norm": 5.5019683837890625,
"learning_rate": 0.00019158227730501932,
"loss": 2.8765948486328123,
"step": 3275
},
{
"epoch": 1.5888300433317286,
"grad_norm": 5.808835983276367,
"learning_rate": 0.00019142502346964068,
"loss": 2.8841793823242186,
"step": 3300
},
{
"epoch": 1.60086663456909,
"grad_norm": 5.601189613342285,
"learning_rate": 0.00019126638010840126,
"loss": 2.90692138671875,
"step": 3325
},
{
"epoch": 1.6129032258064515,
"grad_norm": 6.727347373962402,
"learning_rate": 0.00019110634963244677,
"loss": 2.8471392822265624,
"step": 3350
},
{
"epoch": 1.6249398170438132,
"grad_norm": 5.520786762237549,
"learning_rate": 0.000190944934474005,
"loss": 2.825626220703125,
"step": 3375
},
{
"epoch": 1.6369764082811749,
"grad_norm": 5.6855244636535645,
"learning_rate": 0.0001907821370863489,
"loss": 2.85347900390625,
"step": 3400
},
{
"epoch": 1.6490129995185363,
"grad_norm": 6.275842189788818,
"learning_rate": 0.0001906179599437593,
"loss": 2.8564202880859373,
"step": 3425
},
{
"epoch": 1.661049590755898,
"grad_norm": 6.627660751342773,
"learning_rate": 0.0001904524055414872,
"loss": 2.88417236328125,
"step": 3450
},
{
"epoch": 1.6730861819932596,
"grad_norm": 5.4954609870910645,
"learning_rate": 0.00019028547639571598,
"loss": 2.883042297363281,
"step": 3475
},
{
"epoch": 1.685122773230621,
"grad_norm": 6.2831034660339355,
"learning_rate": 0.00019011717504352312,
"loss": 2.89915771484375,
"step": 3500
},
{
"epoch": 1.6971593644679825,
"grad_norm": 6.845497131347656,
"learning_rate": 0.00018994750404284155,
"loss": 2.81878173828125,
"step": 3525
},
{
"epoch": 1.7091959557053442,
"grad_norm": 7.301944732666016,
"learning_rate": 0.00018977646597242087,
"loss": 2.8858172607421877,
"step": 3550
},
{
"epoch": 1.7212325469427059,
"grad_norm": 7.100738525390625,
"learning_rate": 0.00018960406343178815,
"loss": 2.8665240478515623,
"step": 3575
},
{
"epoch": 1.7332691381800673,
"grad_norm": 6.5114593505859375,
"learning_rate": 0.00018943029904120836,
"loss": 2.8382354736328126,
"step": 3600
},
{
"epoch": 1.745305729417429,
"grad_norm": 5.751381874084473,
"learning_rate": 0.00018925517544164468,
"loss": 2.868603515625,
"step": 3625
},
{
"epoch": 1.7573423206547907,
"grad_norm": 5.941408634185791,
"learning_rate": 0.00018907869529471814,
"loss": 2.8781854248046876,
"step": 3650
},
{
"epoch": 1.7693789118921521,
"grad_norm": 5.040097236633301,
"learning_rate": 0.00018890086128266735,
"loss": 2.90862060546875,
"step": 3675
},
{
"epoch": 1.7814155031295136,
"grad_norm": 7.4268341064453125,
"learning_rate": 0.0001887216761083077,
"loss": 2.870086364746094,
"step": 3700
},
{
"epoch": 1.7934520943668752,
"grad_norm": 5.456490993499756,
"learning_rate": 0.00018854114249499025,
"loss": 2.8436822509765625,
"step": 3725
},
{
"epoch": 1.805488685604237,
"grad_norm": 6.014543533325195,
"learning_rate": 0.00018835926318656033,
"loss": 2.8642138671875,
"step": 3750
},
{
"epoch": 1.8175252768415984,
"grad_norm": 6.103621482849121,
"learning_rate": 0.00018817604094731582,
"loss": 2.8625579833984376,
"step": 3775
},
{
"epoch": 1.82956186807896,
"grad_norm": 6.694746494293213,
"learning_rate": 0.0001879914785619653,
"loss": 2.8301632690429686,
"step": 3800
},
{
"epoch": 1.8415984593163217,
"grad_norm": 5.783210754394531,
"learning_rate": 0.00018780557883558543,
"loss": 2.8412396240234377,
"step": 3825
},
{
"epoch": 1.8536350505536832,
"grad_norm": 6.31929349899292,
"learning_rate": 0.00018761834459357863,
"loss": 2.8531256103515625,
"step": 3850
},
{
"epoch": 1.8656716417910446,
"grad_norm": 5.855931282043457,
"learning_rate": 0.00018742977868162992,
"loss": 2.8694882202148437,
"step": 3875
},
{
"epoch": 1.8777082330284065,
"grad_norm": 5.388266563415527,
"learning_rate": 0.00018723988396566376,
"loss": 2.848956298828125,
"step": 3900
},
{
"epoch": 1.889744824265768,
"grad_norm": 7.263085842132568,
"learning_rate": 0.0001870486633318005,
"loss": 2.8653179931640627,
"step": 3925
},
{
"epoch": 1.9017814155031294,
"grad_norm": 5.926449298858643,
"learning_rate": 0.00018685611968631243,
"loss": 2.8881048583984374,
"step": 3950
},
{
"epoch": 1.913818006740491,
"grad_norm": 6.0646820068359375,
"learning_rate": 0.0001866622559555797,
"loss": 2.8977716064453123,
"step": 3975
},
{
"epoch": 1.9258545979778527,
"grad_norm": 5.905919075012207,
"learning_rate": 0.0001864670750860459,
"loss": 2.901802673339844,
"step": 4000
},
{
"epoch": 1.9378911892152142,
"grad_norm": 5.82496976852417,
"learning_rate": 0.00018627058004417303,
"loss": 2.8932440185546877,
"step": 4025
},
{
"epoch": 1.9499277804525759,
"grad_norm": 7.004024028778076,
"learning_rate": 0.00018607277381639672,
"loss": 2.8420306396484376,
"step": 4050
},
{
"epoch": 1.9619643716899375,
"grad_norm": 7.782563209533691,
"learning_rate": 0.0001858736594090806,
"loss": 2.901356201171875,
"step": 4075
},
{
"epoch": 1.974000962927299,
"grad_norm": 6.040978908538818,
"learning_rate": 0.00018567323984847075,
"loss": 2.8719561767578123,
"step": 4100
},
{
"epoch": 1.9860375541646604,
"grad_norm": 5.953182220458984,
"learning_rate": 0.00018547151818064966,
"loss": 2.86795166015625,
"step": 4125
},
{
"epoch": 1.998074145402022,
"grad_norm": 5.744591236114502,
"learning_rate": 0.00018526849747148996,
"loss": 2.860693359375,
"step": 4150
},
{
"epoch": 2.0,
"eval_loss": 2.8608288764953613,
"eval_runtime": 5.1874,
"eval_samples_per_second": 289.161,
"eval_steps_per_second": 9.06,
"step": 4154
},
{
"epoch": 2.0101107366393838,
"grad_norm": 6.642997741699219,
"learning_rate": 0.0001850641808066077,
"loss": 2.6906396484375,
"step": 4175
},
{
"epoch": 2.0221473278767452,
"grad_norm": 6.480685234069824,
"learning_rate": 0.00018485857129131564,
"loss": 2.722727966308594,
"step": 4200
},
{
"epoch": 2.0341839191141067,
"grad_norm": 6.283753395080566,
"learning_rate": 0.00018465167205057597,
"loss": 2.683138732910156,
"step": 4225
},
{
"epoch": 2.0462205103514686,
"grad_norm": 5.976901531219482,
"learning_rate": 0.00018444348622895275,
"loss": 2.7119317626953126,
"step": 4250
},
{
"epoch": 2.05825710158883,
"grad_norm": 5.369650840759277,
"learning_rate": 0.00018423401699056423,
"loss": 2.71682861328125,
"step": 4275
},
{
"epoch": 2.0702936928261915,
"grad_norm": 5.941702842712402,
"learning_rate": 0.0001840232675190346,
"loss": 2.7161077880859374,
"step": 4300
},
{
"epoch": 2.0823302840635534,
"grad_norm": 6.431283950805664,
"learning_rate": 0.0001838112410174459,
"loss": 2.7145794677734374,
"step": 4325
},
{
"epoch": 2.094366875300915,
"grad_norm": 5.800447940826416,
"learning_rate": 0.000183597940708289,
"loss": 2.675470886230469,
"step": 4350
},
{
"epoch": 2.1064034665382763,
"grad_norm": 5.919703960418701,
"learning_rate": 0.00018338336983341475,
"loss": 2.70038818359375,
"step": 4375
},
{
"epoch": 2.118440057775638,
"grad_norm": 6.257649898529053,
"learning_rate": 0.0001831675316539849,
"loss": 2.69438720703125,
"step": 4400
},
{
"epoch": 2.1304766490129996,
"grad_norm": 5.656405925750732,
"learning_rate": 0.00018295042945042217,
"loss": 2.7394085693359376,
"step": 4425
},
{
"epoch": 2.142513240250361,
"grad_norm": 6.51097297668457,
"learning_rate": 0.0001827320665223607,
"loss": 2.763361511230469,
"step": 4450
},
{
"epoch": 2.1545498314877225,
"grad_norm": 6.375082969665527,
"learning_rate": 0.00018251244618859583,
"loss": 2.730120849609375,
"step": 4475
},
{
"epoch": 2.1665864227250844,
"grad_norm": 6.981193542480469,
"learning_rate": 0.00018229157178703342,
"loss": 2.71025146484375,
"step": 4500
},
{
"epoch": 2.178623013962446,
"grad_norm": 6.799978733062744,
"learning_rate": 0.00018206944667463953,
"loss": 2.68072265625,
"step": 4525
},
{
"epoch": 2.1906596051998073,
"grad_norm": 6.913100242614746,
"learning_rate": 0.0001818460742273891,
"loss": 2.73613037109375,
"step": 4550
},
{
"epoch": 2.202696196437169,
"grad_norm": 6.257125377655029,
"learning_rate": 0.00018162145784021466,
"loss": 2.756729736328125,
"step": 4575
},
{
"epoch": 2.2147327876745306,
"grad_norm": 5.430022239685059,
"learning_rate": 0.00018139560092695495,
"loss": 2.709376220703125,
"step": 4600
},
{
"epoch": 2.226769378911892,
"grad_norm": 6.211202621459961,
"learning_rate": 0.00018116850692030272,
"loss": 2.7708380126953127,
"step": 4625
},
{
"epoch": 2.2388059701492535,
"grad_norm": 6.597946643829346,
"learning_rate": 0.0001809401792717528,
"loss": 2.701777038574219,
"step": 4650
},
{
"epoch": 2.2508425613866154,
"grad_norm": 5.942161560058594,
"learning_rate": 0.0001807106214515496,
"loss": 2.6990762329101563,
"step": 4675
},
{
"epoch": 2.262879152623977,
"grad_norm": 7.107296466827393,
"learning_rate": 0.00018047983694863425,
"loss": 2.741151123046875,
"step": 4700
},
{
"epoch": 2.2749157438613383,
"grad_norm": 6.462876319885254,
"learning_rate": 0.00018024782927059167,
"loss": 2.7986865234375,
"step": 4725
},
{
"epoch": 2.2869523350987,
"grad_norm": 5.901287078857422,
"learning_rate": 0.0001800146019435973,
"loss": 2.723853759765625,
"step": 4750
},
{
"epoch": 2.2989889263360617,
"grad_norm": 6.557262420654297,
"learning_rate": 0.00017978015851236338,
"loss": 2.731398620605469,
"step": 4775
},
{
"epoch": 2.311025517573423,
"grad_norm": 6.787270545959473,
"learning_rate": 0.00017954450254008516,
"loss": 2.7483465576171877,
"step": 4800
},
{
"epoch": 2.3230621088107846,
"grad_norm": 7.6440887451171875,
"learning_rate": 0.0001793076376083868,
"loss": 2.7426837158203123,
"step": 4825
},
{
"epoch": 2.3350987000481465,
"grad_norm": 5.517909049987793,
"learning_rate": 0.00017906956731726675,
"loss": 2.718466796875,
"step": 4850
},
{
"epoch": 2.347135291285508,
"grad_norm": 5.788064956665039,
"learning_rate": 0.00017883029528504324,
"loss": 2.699304504394531,
"step": 4875
},
{
"epoch": 2.3591718825228694,
"grad_norm": 6.685894012451172,
"learning_rate": 0.00017858982514829913,
"loss": 2.743288269042969,
"step": 4900
},
{
"epoch": 2.3712084737602313,
"grad_norm": 5.696547031402588,
"learning_rate": 0.00017834816056182677,
"loss": 2.7296527099609373,
"step": 4925
},
{
"epoch": 2.3832450649975927,
"grad_norm": 6.552695274353027,
"learning_rate": 0.00017810530519857235,
"loss": 2.7500262451171875,
"step": 4950
},
{
"epoch": 2.395281656234954,
"grad_norm": 5.272420883178711,
"learning_rate": 0.00017786126274958015,
"loss": 2.7251507568359377,
"step": 4975
},
{
"epoch": 2.4073182474723156,
"grad_norm": 5.337221145629883,
"learning_rate": 0.00017761603692393636,
"loss": 2.794324951171875,
"step": 5000
},
{
"epoch": 2.4193548387096775,
"grad_norm": 6.288032531738281,
"learning_rate": 0.00017736963144871276,
"loss": 2.73385498046875,
"step": 5025
},
{
"epoch": 2.431391429947039,
"grad_norm": 6.103067398071289,
"learning_rate": 0.00017712205006891013,
"loss": 2.7030935668945313,
"step": 5050
},
{
"epoch": 2.4434280211844004,
"grad_norm": 5.142453670501709,
"learning_rate": 0.0001768732965474012,
"loss": 2.7068914794921874,
"step": 5075
},
{
"epoch": 2.4554646124217623,
"grad_norm": 6.099581241607666,
"learning_rate": 0.00017662337466487355,
"loss": 2.7641656494140623,
"step": 5100
},
{
"epoch": 2.4675012036591237,
"grad_norm": 6.829136848449707,
"learning_rate": 0.00017637228821977213,
"loss": 2.7043777465820313,
"step": 5125
},
{
"epoch": 2.479537794896485,
"grad_norm": 5.465605735778809,
"learning_rate": 0.00017612004102824153,
"loss": 2.7184182739257814,
"step": 5150
},
{
"epoch": 2.491574386133847,
"grad_norm": 6.038458347320557,
"learning_rate": 0.00017586663692406798,
"loss": 2.7635272216796873,
"step": 5175
},
{
"epoch": 2.5036109773712085,
"grad_norm": 6.614681720733643,
"learning_rate": 0.0001756120797586211,
"loss": 2.769054260253906,
"step": 5200
},
{
"epoch": 2.51564756860857,
"grad_norm": 5.841181755065918,
"learning_rate": 0.0001753563734007953,
"loss": 2.7224261474609377,
"step": 5225
},
{
"epoch": 2.527684159845932,
"grad_norm": 6.294864177703857,
"learning_rate": 0.00017509952173695107,
"loss": 2.7511309814453124,
"step": 5250
},
{
"epoch": 2.5397207510832933,
"grad_norm": 5.583828449249268,
"learning_rate": 0.00017484152867085583,
"loss": 2.7023577880859375,
"step": 5275
},
{
"epoch": 2.5517573423206548,
"grad_norm": 5.762092590332031,
"learning_rate": 0.00017458239812362462,
"loss": 2.7184487915039064,
"step": 5300
},
{
"epoch": 2.563793933558016,
"grad_norm": 7.2237677574157715,
"learning_rate": 0.0001743221340336606,
"loss": 2.746933898925781,
"step": 5325
},
{
"epoch": 2.575830524795378,
"grad_norm": 6.3399786949157715,
"learning_rate": 0.00017406074035659506,
"loss": 2.769549560546875,
"step": 5350
},
{
"epoch": 2.5878671160327396,
"grad_norm": 5.478464603424072,
"learning_rate": 0.0001737982210652273,
"loss": 2.738487548828125,
"step": 5375
},
{
"epoch": 2.599903707270101,
"grad_norm": 6.867932319641113,
"learning_rate": 0.00017353458014946442,
"loss": 2.7589501953125,
"step": 5400
},
{
"epoch": 2.611940298507463,
"grad_norm": 6.149633407592773,
"learning_rate": 0.00017326982161626044,
"loss": 2.7048358154296874,
"step": 5425
},
{
"epoch": 2.6239768897448243,
"grad_norm": 5.812656402587891,
"learning_rate": 0.00017300394948955558,
"loss": 2.7163870239257815,
"step": 5450
},
{
"epoch": 2.636013480982186,
"grad_norm": 8.48040771484375,
"learning_rate": 0.0001727369678102151,
"loss": 2.7666748046875,
"step": 5475
},
{
"epoch": 2.6480500722195472,
"grad_norm": 6.530597686767578,
"learning_rate": 0.00017246888063596774,
"loss": 2.683943176269531,
"step": 5500
},
{
"epoch": 2.660086663456909,
"grad_norm": 5.324703216552734,
"learning_rate": 0.00017219969204134412,
"loss": 2.756099853515625,
"step": 5525
},
{
"epoch": 2.6721232546942706,
"grad_norm": 5.394314289093018,
"learning_rate": 0.0001719294061176149,
"loss": 2.7569219970703127,
"step": 5550
},
{
"epoch": 2.684159845931632,
"grad_norm": 6.482044219970703,
"learning_rate": 0.00017165802697272854,
"loss": 2.7562155151367187,
"step": 5575
},
{
"epoch": 2.696196437168994,
"grad_norm": 5.999021530151367,
"learning_rate": 0.00017138555873124874,
"loss": 2.7514157104492187,
"step": 5600
},
{
"epoch": 2.7082330284063554,
"grad_norm": 6.616397380828857,
"learning_rate": 0.000171112005534292,
"loss": 2.709639892578125,
"step": 5625
},
{
"epoch": 2.720269619643717,
"grad_norm": 6.24782133102417,
"learning_rate": 0.0001708373715394645,
"loss": 2.7355551147460937,
"step": 5650
},
{
"epoch": 2.7323062108810783,
"grad_norm": 6.01416015625,
"learning_rate": 0.00017056166092079887,
"loss": 2.681227722167969,
"step": 5675
},
{
"epoch": 2.74434280211844,
"grad_norm": 5.480840682983398,
"learning_rate": 0.00017028487786869097,
"loss": 2.77478759765625,
"step": 5700
},
{
"epoch": 2.7563793933558016,
"grad_norm": 6.248979568481445,
"learning_rate": 0.00017000702658983603,
"loss": 2.71619140625,
"step": 5725
},
{
"epoch": 2.768415984593163,
"grad_norm": 6.709351062774658,
"learning_rate": 0.0001697281113071647,
"loss": 2.6932601928710938,
"step": 5750
},
{
"epoch": 2.780452575830525,
"grad_norm": 5.883296966552734,
"learning_rate": 0.00016944813625977895,
"loss": 2.7240725708007814,
"step": 5775
},
{
"epoch": 2.7924891670678864,
"grad_norm": 6.043994903564453,
"learning_rate": 0.00016916710570288767,
"loss": 2.730412902832031,
"step": 5800
},
{
"epoch": 2.804525758305248,
"grad_norm": 6.184457302093506,
"learning_rate": 0.0001688850239077419,
"loss": 2.7380224609375,
"step": 5825
},
{
"epoch": 2.8165623495426093,
"grad_norm": 6.543738842010498,
"learning_rate": 0.00016860189516156992,
"loss": 2.7130126953125,
"step": 5850
},
{
"epoch": 2.828598940779971,
"grad_norm": 6.47238826751709,
"learning_rate": 0.0001683177237675122,
"loss": 2.7353460693359377,
"step": 5875
},
{
"epoch": 2.8406355320173327,
"grad_norm": 5.968029975891113,
"learning_rate": 0.00016803251404455582,
"loss": 2.7125958251953124,
"step": 5900
},
{
"epoch": 2.8526721232546945,
"grad_norm": 6.4303717613220215,
"learning_rate": 0.00016774627032746904,
"loss": 2.710166015625,
"step": 5925
},
{
"epoch": 2.864708714492056,
"grad_norm": 5.7404093742370605,
"learning_rate": 0.0001674589969667353,
"loss": 2.7399676513671873,
"step": 5950
},
{
"epoch": 2.8767453057294174,
"grad_norm": 6.183196544647217,
"learning_rate": 0.00016717069832848698,
"loss": 2.7283358764648438,
"step": 5975
},
{
"epoch": 2.888781896966779,
"grad_norm": 5.792095184326172,
"learning_rate": 0.00016688137879443935,
"loss": 2.70381591796875,
"step": 6000
},
{
"epoch": 2.9008184882041403,
"grad_norm": 5.859752655029297,
"learning_rate": 0.00016659104276182365,
"loss": 2.7537738037109376,
"step": 6025
},
{
"epoch": 2.9128550794415022,
"grad_norm": 6.3950371742248535,
"learning_rate": 0.0001662996946433205,
"loss": 2.703232421875,
"step": 6050
},
{
"epoch": 2.9248916706788637,
"grad_norm": 6.948364734649658,
"learning_rate": 0.00016600733886699268,
"loss": 2.7639871215820313,
"step": 6075
},
{
"epoch": 2.9369282619162256,
"grad_norm": 6.1982197761535645,
"learning_rate": 0.0001657139798762179,
"loss": 2.733709411621094,
"step": 6100
},
{
"epoch": 2.948964853153587,
"grad_norm": 6.387999534606934,
"learning_rate": 0.00016541962212962127,
"loss": 2.7400689697265626,
"step": 6125
},
{
"epoch": 2.9610014443909485,
"grad_norm": 5.882162570953369,
"learning_rate": 0.00016512427010100748,
"loss": 2.7062359619140626,
"step": 6150
},
{
"epoch": 2.97303803562831,
"grad_norm": 7.070481777191162,
"learning_rate": 0.0001648279282792929,
"loss": 2.6777212524414065,
"step": 6175
},
{
"epoch": 2.9850746268656714,
"grad_norm": 5.267948150634766,
"learning_rate": 0.0001645306011684373,
"loss": 2.7145626831054686,
"step": 6200
},
{
"epoch": 2.9971112181030333,
"grad_norm": 6.219203948974609,
"learning_rate": 0.00016423229328737526,
"loss": 2.75112060546875,
"step": 6225
},
{
"epoch": 3.0,
"eval_loss": 2.7634668350219727,
"eval_runtime": 4.3083,
"eval_samples_per_second": 348.166,
"eval_steps_per_second": 10.909,
"step": 6231
},
{
"epoch": 3.0091478093403947,
"grad_norm": 7.532505035400391,
"learning_rate": 0.0001639330091699478,
"loss": 2.617950439453125,
"step": 6250
},
{
"epoch": 3.021184400577756,
"grad_norm": 7.172801971435547,
"learning_rate": 0.00016363275336483328,
"loss": 2.505116271972656,
"step": 6275
},
{
"epoch": 3.033220991815118,
"grad_norm": 6.818567276000977,
"learning_rate": 0.0001633315304354781,
"loss": 2.5599952697753907,
"step": 6300
},
{
"epoch": 3.0452575830524795,
"grad_norm": 6.314502716064453,
"learning_rate": 0.00016302934496002775,
"loss": 2.5814093017578124,
"step": 6325
},
{
"epoch": 3.057294174289841,
"grad_norm": 7.728026866912842,
"learning_rate": 0.0001627262015312569,
"loss": 2.5842514038085938,
"step": 6350
},
{
"epoch": 3.069330765527203,
"grad_norm": 6.728147506713867,
"learning_rate": 0.00016242210475649968,
"loss": 2.587703857421875,
"step": 6375
},
{
"epoch": 3.0813673567645643,
"grad_norm": 5.884317398071289,
"learning_rate": 0.00016211705925757975,
"loss": 2.538614044189453,
"step": 6400
},
{
"epoch": 3.0934039480019258,
"grad_norm": 6.846307277679443,
"learning_rate": 0.00016181106967073987,
"loss": 2.541700592041016,
"step": 6425
},
{
"epoch": 3.1054405392392876,
"grad_norm": 7.39595365524292,
"learning_rate": 0.00016150414064657168,
"loss": 2.56431884765625,
"step": 6450
},
{
"epoch": 3.117477130476649,
"grad_norm": 7.125731468200684,
"learning_rate": 0.00016119627684994478,
"loss": 2.5778378295898436,
"step": 6475
},
{
"epoch": 3.1295137217140105,
"grad_norm": 6.258044242858887,
"learning_rate": 0.00016088748295993603,
"loss": 2.5168409729003907,
"step": 6500
},
{
"epoch": 3.141550312951372,
"grad_norm": 7.049948215484619,
"learning_rate": 0.00016057776366975825,
"loss": 2.556155548095703,
"step": 6525
},
{
"epoch": 3.153586904188734,
"grad_norm": 6.056478500366211,
"learning_rate": 0.00016026712368668903,
"loss": 2.5735009765625,
"step": 6550
},
{
"epoch": 3.1656234954260953,
"grad_norm": 6.195144176483154,
"learning_rate": 0.00015995556773199912,
"loss": 2.563138427734375,
"step": 6575
},
{
"epoch": 3.177660086663457,
"grad_norm": 7.984062194824219,
"learning_rate": 0.0001596431005408807,
"loss": 2.5712921142578127,
"step": 6600
},
{
"epoch": 3.1896966779008187,
"grad_norm": 6.834407806396484,
"learning_rate": 0.00015932972686237543,
"loss": 2.524141845703125,
"step": 6625
},
{
"epoch": 3.20173326913818,
"grad_norm": 7.851695537567139,
"learning_rate": 0.0001590154514593022,
"loss": 2.527924041748047,
"step": 6650
},
{
"epoch": 3.2137698603755416,
"grad_norm": 6.800895690917969,
"learning_rate": 0.00015870027910818476,
"loss": 2.559481658935547,
"step": 6675
},
{
"epoch": 3.225806451612903,
"grad_norm": 6.966695785522461,
"learning_rate": 0.00015838421459917927,
"loss": 2.643104248046875,
"step": 6700
},
{
"epoch": 3.237843042850265,
"grad_norm": 6.1939005851745605,
"learning_rate": 0.00015806726273600127,
"loss": 2.6130157470703126,
"step": 6725
},
{
"epoch": 3.2498796340876264,
"grad_norm": 6.8425612449646,
"learning_rate": 0.0001577494283358528,
"loss": 2.6474554443359377,
"step": 6750
},
{
"epoch": 3.261916225324988,
"grad_norm": 6.54850435256958,
"learning_rate": 0.00015743071622934917,
"loss": 2.5912841796875,
"step": 6775
},
{
"epoch": 3.2739528165623497,
"grad_norm": 6.886445045471191,
"learning_rate": 0.00015711113126044555,
"loss": 2.5946051025390626,
"step": 6800
},
{
"epoch": 3.285989407799711,
"grad_norm": 7.529403209686279,
"learning_rate": 0.00015679067828636335,
"loss": 2.5568299865722657,
"step": 6825
},
{
"epoch": 3.2980259990370726,
"grad_norm": 7.362818717956543,
"learning_rate": 0.0001564693621775163,
"loss": 2.6015423583984374,
"step": 6850
},
{
"epoch": 3.310062590274434,
"grad_norm": 6.586751937866211,
"learning_rate": 0.00015614718781743663,
"loss": 2.6284942626953125,
"step": 6875
},
{
"epoch": 3.322099181511796,
"grad_norm": 6.70015811920166,
"learning_rate": 0.00015582416010270062,
"loss": 2.587824401855469,
"step": 6900
},
{
"epoch": 3.3341357727491574,
"grad_norm": 6.754818916320801,
"learning_rate": 0.00015550028394285432,
"loss": 2.632406005859375,
"step": 6925
},
{
"epoch": 3.346172363986519,
"grad_norm": 7.001147747039795,
"learning_rate": 0.0001551755642603389,
"loss": 2.567904052734375,
"step": 6950
},
{
"epoch": 3.3582089552238807,
"grad_norm": 7.025686264038086,
"learning_rate": 0.00015485000599041587,
"loss": 2.5608477783203125,
"step": 6975
},
{
"epoch": 3.370245546461242,
"grad_norm": 5.955574035644531,
"learning_rate": 0.000154523614081092,
"loss": 2.579781494140625,
"step": 7000
},
{
"epoch": 3.3822821376986036,
"grad_norm": 8.050995826721191,
"learning_rate": 0.0001541963934930441,
"loss": 2.60483154296875,
"step": 7025
},
{
"epoch": 3.394318728935965,
"grad_norm": 6.99261999130249,
"learning_rate": 0.00015386834919954386,
"loss": 2.603854064941406,
"step": 7050
},
{
"epoch": 3.406355320173327,
"grad_norm": 6.021421432495117,
"learning_rate": 0.00015353948618638177,
"loss": 2.6255908203125,
"step": 7075
},
{
"epoch": 3.4183919114106884,
"grad_norm": 6.664973258972168,
"learning_rate": 0.00015320980945179196,
"loss": 2.589926452636719,
"step": 7100
},
{
"epoch": 3.43042850264805,
"grad_norm": 7.45606803894043,
"learning_rate": 0.00015287932400637576,
"loss": 2.5651092529296875,
"step": 7125
},
{
"epoch": 3.4424650938854118,
"grad_norm": 7.181027412414551,
"learning_rate": 0.0001525480348730258,
"loss": 2.617786865234375,
"step": 7150
},
{
"epoch": 3.4545016851227732,
"grad_norm": 6.788301944732666,
"learning_rate": 0.00015221594708684952,
"loss": 2.6190371704101563,
"step": 7175
},
{
"epoch": 3.4665382763601347,
"grad_norm": 7.416632175445557,
"learning_rate": 0.00015188306569509274,
"loss": 2.5922671508789064,
"step": 7200
},
{
"epoch": 3.4785748675974966,
"grad_norm": 6.855120658874512,
"learning_rate": 0.0001515493957570629,
"loss": 2.5730856323242186,
"step": 7225
},
{
"epoch": 3.490611458834858,
"grad_norm": 5.933901786804199,
"learning_rate": 0.00015121494234405218,
"loss": 2.5321611022949218,
"step": 7250
},
{
"epoch": 3.5026480500722195,
"grad_norm": 6.569029808044434,
"learning_rate": 0.00015087971053926052,
"loss": 2.56937744140625,
"step": 7275
},
{
"epoch": 3.5146846413095814,
"grad_norm": 5.820389747619629,
"learning_rate": 0.00015054370543771809,
"loss": 2.5830810546875,
"step": 7300
},
{
"epoch": 3.526721232546943,
"grad_norm": 7.1573004722595215,
"learning_rate": 0.0001502069321462082,
"loss": 2.5898391723632814,
"step": 7325
},
{
"epoch": 3.5387578237843043,
"grad_norm": 7.135001182556152,
"learning_rate": 0.0001498693957831894,
"loss": 2.631561279296875,
"step": 7350
},
{
"epoch": 3.5507944150216657,
"grad_norm": 7.952722549438477,
"learning_rate": 0.00014953110147871793,
"loss": 2.5712603759765624,
"step": 7375
},
{
"epoch": 3.5628310062590276,
"grad_norm": 8.231966972351074,
"learning_rate": 0.00014919205437436946,
"loss": 2.610478820800781,
"step": 7400
},
{
"epoch": 3.574867597496389,
"grad_norm": 6.280411720275879,
"learning_rate": 0.00014885225962316128,
"loss": 2.529931945800781,
"step": 7425
},
{
"epoch": 3.5869041887337505,
"grad_norm": 5.856441497802734,
"learning_rate": 0.00014851172238947372,
"loss": 2.5853460693359374,
"step": 7450
},
{
"epoch": 3.5989407799711124,
"grad_norm": 6.418439865112305,
"learning_rate": 0.00014817044784897173,
"loss": 2.6548699951171875,
"step": 7475
},
{
"epoch": 3.610977371208474,
"grad_norm": 5.8023295402526855,
"learning_rate": 0.00014782844118852627,
"loss": 2.5802447509765627,
"step": 7500
},
{
"epoch": 3.6230139624458353,
"grad_norm": 5.9953813552856445,
"learning_rate": 0.00014748570760613549,
"loss": 2.6488763427734376,
"step": 7525
},
{
"epoch": 3.6350505536831967,
"grad_norm": 6.7495832443237305,
"learning_rate": 0.00014714225231084562,
"loss": 2.645798645019531,
"step": 7550
},
{
"epoch": 3.6470871449205586,
"grad_norm": 6.393189430236816,
"learning_rate": 0.0001467980805226719,
"loss": 2.5896685791015623,
"step": 7575
},
{
"epoch": 3.65912373615792,
"grad_norm": 6.28262996673584,
"learning_rate": 0.00014645319747251921,
"loss": 2.6326702880859374,
"step": 7600
},
{
"epoch": 3.6711603273952815,
"grad_norm": 6.2060065269470215,
"learning_rate": 0.00014610760840210254,
"loss": 2.6136151123046876,
"step": 7625
},
{
"epoch": 3.6831969186326434,
"grad_norm": 5.923848628997803,
"learning_rate": 0.00014576131856386735,
"loss": 2.587392578125,
"step": 7650
},
{
"epoch": 3.695233509870005,
"grad_norm": 5.914072036743164,
"learning_rate": 0.00014541433322090973,
"loss": 2.6291851806640625,
"step": 7675
},
{
"epoch": 3.7072701011073663,
"grad_norm": 7.405152320861816,
"learning_rate": 0.0001450666576468965,
"loss": 2.5964593505859375,
"step": 7700
},
{
"epoch": 3.7193066923447278,
"grad_norm": 6.541108131408691,
"learning_rate": 0.0001447182971259849,
"loss": 2.565693054199219,
"step": 7725
},
{
"epoch": 3.7313432835820897,
"grad_norm": 5.909659385681152,
"learning_rate": 0.00014436925695274235,
"loss": 2.5886358642578124,
"step": 7750
},
{
"epoch": 3.743379874819451,
"grad_norm": 6.0169219970703125,
"learning_rate": 0.000144019542432066,
"loss": 2.610029296875,
"step": 7775
},
{
"epoch": 3.7554164660568126,
"grad_norm": 6.487315654754639,
"learning_rate": 0.00014366915887910214,
"loss": 2.6469882202148436,
"step": 7800
},
{
"epoch": 3.7674530572941745,
"grad_norm": 5.895727157592773,
"learning_rate": 0.0001433181116191653,
"loss": 2.5247569274902344,
"step": 7825
},
{
"epoch": 3.779489648531536,
"grad_norm": 6.585092067718506,
"learning_rate": 0.0001429664059876574,
"loss": 2.5999505615234373,
"step": 7850
},
{
"epoch": 3.7915262397688974,
"grad_norm": 7.017783164978027,
"learning_rate": 0.00014261404732998664,
"loss": 2.633411865234375,
"step": 7875
},
{
"epoch": 3.803562831006259,
"grad_norm": 6.384684085845947,
"learning_rate": 0.00014226104100148628,
"loss": 2.5933465576171875,
"step": 7900
},
{
"epoch": 3.8155994222436207,
"grad_norm": 7.031004905700684,
"learning_rate": 0.00014190739236733316,
"loss": 2.595792541503906,
"step": 7925
},
{
"epoch": 3.827636013480982,
"grad_norm": 6.241200923919678,
"learning_rate": 0.0001415531068024663,
"loss": 2.5923992919921877,
"step": 7950
},
{
"epoch": 3.839672604718344,
"grad_norm": 6.395936965942383,
"learning_rate": 0.0001411981896915051,
"loss": 2.613453369140625,
"step": 7975
},
{
"epoch": 3.8517091959557055,
"grad_norm": 6.0687689781188965,
"learning_rate": 0.0001408426464286675,
"loss": 2.5782708740234375,
"step": 8000
},
{
"epoch": 3.863745787193067,
"grad_norm": 7.165973663330078,
"learning_rate": 0.00014048648241768807,
"loss": 2.5483168029785155,
"step": 8025
},
{
"epoch": 3.8757823784304284,
"grad_norm": 6.542165279388428,
"learning_rate": 0.0001401297030717358,
"loss": 2.636019287109375,
"step": 8050
},
{
"epoch": 3.88781896966779,
"grad_norm": 6.252635478973389,
"learning_rate": 0.00013977231381333193,
"loss": 2.6014303588867187,
"step": 8075
},
{
"epoch": 3.8998555609051517,
"grad_norm": 6.700534343719482,
"learning_rate": 0.0001394143200742674,
"loss": 2.5424969482421873,
"step": 8100
},
{
"epoch": 3.911892152142513,
"grad_norm": 6.896066188812256,
"learning_rate": 0.0001390557272955204,
"loss": 2.5399298095703124,
"step": 8125
},
{
"epoch": 3.923928743379875,
"grad_norm": 6.549251079559326,
"learning_rate": 0.00013869654092717363,
"loss": 2.5947125244140623,
"step": 8150
},
{
"epoch": 3.9359653346172365,
"grad_norm": 6.1599836349487305,
"learning_rate": 0.00013833676642833152,
"loss": 2.61697021484375,
"step": 8175
},
{
"epoch": 3.948001925854598,
"grad_norm": 6.616363525390625,
"learning_rate": 0.00013797640926703717,
"loss": 2.5735784912109376,
"step": 8200
},
{
"epoch": 3.9600385170919594,
"grad_norm": 6.72858190536499,
"learning_rate": 0.00013761547492018928,
"loss": 2.5907177734375,
"step": 8225
},
{
"epoch": 3.972075108329321,
"grad_norm": 6.385076522827148,
"learning_rate": 0.00013725396887345897,
"loss": 2.6179083251953124,
"step": 8250
},
{
"epoch": 3.9841116995666828,
"grad_norm": 7.102210521697998,
"learning_rate": 0.00013689189662120633,
"loss": 2.61026611328125,
"step": 8275
},
{
"epoch": 3.996148290804044,
"grad_norm": 6.699630260467529,
"learning_rate": 0.000136529263666397,
"loss": 2.623244323730469,
"step": 8300
},
{
"epoch": 4.0,
"eval_loss": 2.6820578575134277,
"eval_runtime": 4.4599,
"eval_samples_per_second": 336.333,
"eval_steps_per_second": 10.538,
"step": 8308
},
{
"epoch": 4.008184882041406,
"grad_norm": 5.906957626342773,
"learning_rate": 0.00013616607552051834,
"loss": 2.4383436584472657,
"step": 8325
},
{
"epoch": 4.0202214732787676,
"grad_norm": 6.453551292419434,
"learning_rate": 0.00013580233770349592,
"loss": 2.347547607421875,
"step": 8350
},
{
"epoch": 4.032258064516129,
"grad_norm": 6.723680019378662,
"learning_rate": 0.00013543805574360948,
"loss": 2.373088073730469,
"step": 8375
},
{
"epoch": 4.0442946557534905,
"grad_norm": 7.363197326660156,
"learning_rate": 0.00013507323517740884,
"loss": 2.4030003356933594,
"step": 8400
},
{
"epoch": 4.056331246990852,
"grad_norm": 6.1376471519470215,
"learning_rate": 0.00013470788154963,
"loss": 2.389628448486328,
"step": 8425
},
{
"epoch": 4.068367838228213,
"grad_norm": 7.293612003326416,
"learning_rate": 0.00013434200041311057,
"loss": 2.391862640380859,
"step": 8450
},
{
"epoch": 4.080404429465576,
"grad_norm": 7.916410446166992,
"learning_rate": 0.00013397559732870564,
"loss": 2.357501678466797,
"step": 8475
},
{
"epoch": 4.092441020702937,
"grad_norm": 6.08396577835083,
"learning_rate": 0.000133608677865203,
"loss": 2.4158206176757813,
"step": 8500
},
{
"epoch": 4.104477611940299,
"grad_norm": 7.022041320800781,
"learning_rate": 0.0001332412475992388,
"loss": 2.438868408203125,
"step": 8525
},
{
"epoch": 4.11651420317766,
"grad_norm": 7.004306793212891,
"learning_rate": 0.00013287331211521246,
"loss": 2.3929733276367187,
"step": 8550
},
{
"epoch": 4.1285507944150215,
"grad_norm": 6.981446743011475,
"learning_rate": 0.0001325048770052021,
"loss": 2.431510467529297,
"step": 8575
},
{
"epoch": 4.140587385652383,
"grad_norm": 6.265749931335449,
"learning_rate": 0.00013213594786887944,
"loss": 2.4273892211914063,
"step": 8600
},
{
"epoch": 4.152623976889744,
"grad_norm": 7.472740173339844,
"learning_rate": 0.00013176653031342456,
"loss": 2.409165344238281,
"step": 8625
},
{
"epoch": 4.164660568127107,
"grad_norm": 7.548570156097412,
"learning_rate": 0.00013139662995344085,
"loss": 2.4531121826171876,
"step": 8650
},
{
"epoch": 4.176697159364468,
"grad_norm": 6.898479461669922,
"learning_rate": 0.00013102625241086965,
"loss": 2.432555847167969,
"step": 8675
},
{
"epoch": 4.18873375060183,
"grad_norm": 6.945794105529785,
"learning_rate": 0.00013065540331490472,
"loss": 2.455061798095703,
"step": 8700
},
{
"epoch": 4.200770341839191,
"grad_norm": 6.814357280731201,
"learning_rate": 0.00013028408830190674,
"loss": 2.4712957763671874,
"step": 8725
},
{
"epoch": 4.2128069330765525,
"grad_norm": 7.292290687561035,
"learning_rate": 0.00012991231301531773,
"loss": 2.4582168579101564,
"step": 8750
},
{
"epoch": 4.224843524313914,
"grad_norm": 7.236823558807373,
"learning_rate": 0.00012954008310557506,
"loss": 2.446655578613281,
"step": 8775
},
{
"epoch": 4.236880115551276,
"grad_norm": 6.882739543914795,
"learning_rate": 0.0001291674042300258,
"loss": 2.405202178955078,
"step": 8800
},
{
"epoch": 4.248916706788638,
"grad_norm": 7.120377063751221,
"learning_rate": 0.00012879428205284058,
"loss": 2.400958251953125,
"step": 8825
},
{
"epoch": 4.260953298025999,
"grad_norm": 7.028608322143555,
"learning_rate": 0.00012842072224492762,
"loss": 2.4500877380371096,
"step": 8850
},
{
"epoch": 4.272989889263361,
"grad_norm": 7.8143744468688965,
"learning_rate": 0.00012804673048384647,
"loss": 2.452120056152344,
"step": 8875
},
{
"epoch": 4.285026480500722,
"grad_norm": 7.579928398132324,
"learning_rate": 0.00012767231245372168,
"loss": 2.4451675415039062,
"step": 8900
},
{
"epoch": 4.2970630717380836,
"grad_norm": 7.399620056152344,
"learning_rate": 0.00012729747384515648,
"loss": 2.492531433105469,
"step": 8925
},
{
"epoch": 4.309099662975445,
"grad_norm": 8.107259750366211,
"learning_rate": 0.0001269222203551464,
"loss": 2.432616424560547,
"step": 8950
},
{
"epoch": 4.321136254212807,
"grad_norm": 6.925912857055664,
"learning_rate": 0.0001265465576869923,
"loss": 2.385529022216797,
"step": 8975
},
{
"epoch": 4.333172845450169,
"grad_norm": 6.463615894317627,
"learning_rate": 0.00012617049155021418,
"loss": 2.434081726074219,
"step": 9000
},
{
"epoch": 4.34520943668753,
"grad_norm": 7.031606674194336,
"learning_rate": 0.0001257940276604641,
"loss": 2.464780731201172,
"step": 9025
},
{
"epoch": 4.357246027924892,
"grad_norm": 6.946998596191406,
"learning_rate": 0.00012541717173943936,
"loss": 2.4792620849609377,
"step": 9050
},
{
"epoch": 4.369282619162253,
"grad_norm": 6.719104290008545,
"learning_rate": 0.0001250399295147955,
"loss": 2.485868225097656,
"step": 9075
},
{
"epoch": 4.381319210399615,
"grad_norm": 6.994076728820801,
"learning_rate": 0.00012466230672005945,
"loss": 2.398624267578125,
"step": 9100
},
{
"epoch": 4.393355801636976,
"grad_norm": 6.991820335388184,
"learning_rate": 0.0001242843090945421,
"loss": 2.4889877319335936,
"step": 9125
},
{
"epoch": 4.405392392874338,
"grad_norm": 6.849164009094238,
"learning_rate": 0.00012390594238325132,
"loss": 2.4707565307617188,
"step": 9150
},
{
"epoch": 4.4174289841117,
"grad_norm": 7.551220417022705,
"learning_rate": 0.00012352721233680454,
"loss": 2.47287109375,
"step": 9175
},
{
"epoch": 4.429465575349061,
"grad_norm": 7.383901119232178,
"learning_rate": 0.00012314812471134125,
"loss": 2.460268707275391,
"step": 9200
},
{
"epoch": 4.441502166586423,
"grad_norm": 6.343733310699463,
"learning_rate": 0.00012276868526843573,
"loss": 2.451211242675781,
"step": 9225
},
{
"epoch": 4.453538757823784,
"grad_norm": 7.342288970947266,
"learning_rate": 0.00012238889977500924,
"loss": 2.4492242431640623,
"step": 9250
},
{
"epoch": 4.465575349061146,
"grad_norm": 7.359269618988037,
"learning_rate": 0.00012200877400324267,
"loss": 2.4454722595214844,
"step": 9275
},
{
"epoch": 4.477611940298507,
"grad_norm": 6.943546772003174,
"learning_rate": 0.00012162831373048848,
"loss": 2.4957781982421876,
"step": 9300
},
{
"epoch": 4.489648531535869,
"grad_norm": 8.258399963378906,
"learning_rate": 0.00012124752473918315,
"loss": 2.491903839111328,
"step": 9325
},
{
"epoch": 4.501685122773231,
"grad_norm": 6.876328945159912,
"learning_rate": 0.00012086641281675912,
"loss": 2.4897991943359377,
"step": 9350
},
{
"epoch": 4.513721714010592,
"grad_norm": 5.609645843505859,
"learning_rate": 0.00012048498375555702,
"loss": 2.4758555603027346,
"step": 9375
},
{
"epoch": 4.525758305247954,
"grad_norm": 7.3538126945495605,
"learning_rate": 0.00012010324335273744,
"loss": 2.4127471923828123,
"step": 9400
},
{
"epoch": 4.537794896485315,
"grad_norm": 7.781580924987793,
"learning_rate": 0.00011972119741019288,
"loss": 2.491829528808594,
"step": 9425
},
{
"epoch": 4.549831487722677,
"grad_norm": 7.821054935455322,
"learning_rate": 0.00011933885173445968,
"loss": 2.4405499267578126,
"step": 9450
},
{
"epoch": 4.561868078960039,
"grad_norm": 6.082571983337402,
"learning_rate": 0.00011895621213662962,
"loss": 2.4237486267089845,
"step": 9475
},
{
"epoch": 4.5739046701974,
"grad_norm": 6.463493347167969,
"learning_rate": 0.0001185732844322617,
"loss": 2.4664483642578126,
"step": 9500
},
{
"epoch": 4.585941261434762,
"grad_norm": 6.795047760009766,
"learning_rate": 0.00011819007444129368,
"loss": 2.4464166259765623,
"step": 9525
},
{
"epoch": 4.597977852672123,
"grad_norm": 6.385959625244141,
"learning_rate": 0.0001178065879879537,
"loss": 2.4959071350097655,
"step": 9550
},
{
"epoch": 4.610014443909485,
"grad_norm": 6.93854284286499,
"learning_rate": 0.0001174228309006717,
"loss": 2.5116204833984375,
"step": 9575
},
{
"epoch": 4.622051035146846,
"grad_norm": 5.849823474884033,
"learning_rate": 0.00011703880901199081,
"loss": 2.5280581665039064,
"step": 9600
},
{
"epoch": 4.634087626384208,
"grad_norm": 7.291132926940918,
"learning_rate": 0.00011665452815847886,
"loss": 2.4106024169921874,
"step": 9625
},
{
"epoch": 4.646124217621569,
"grad_norm": 7.537511348724365,
"learning_rate": 0.00011626999418063951,
"loss": 2.471666107177734,
"step": 9650
},
{
"epoch": 4.6581608088589315,
"grad_norm": 7.474704265594482,
"learning_rate": 0.00011588521292282346,
"loss": 2.4104833984375,
"step": 9675
},
{
"epoch": 4.670197400096293,
"grad_norm": 6.489062786102295,
"learning_rate": 0.00011550019023313987,
"loss": 2.4799649047851564,
"step": 9700
},
{
"epoch": 4.682233991333654,
"grad_norm": 6.90573263168335,
"learning_rate": 0.00011511493196336714,
"loss": 2.4340060424804686,
"step": 9725
},
{
"epoch": 4.694270582571016,
"grad_norm": 6.583920001983643,
"learning_rate": 0.00011472944396886423,
"loss": 2.4326211547851564,
"step": 9750
},
{
"epoch": 4.706307173808377,
"grad_norm": 6.703094482421875,
"learning_rate": 0.0001143437321084816,
"loss": 2.415018310546875,
"step": 9775
},
{
"epoch": 4.718343765045739,
"grad_norm": 6.776236057281494,
"learning_rate": 0.00011395780224447203,
"loss": 2.453302307128906,
"step": 9800
},
{
"epoch": 4.730380356283101,
"grad_norm": 7.194311141967773,
"learning_rate": 0.0001135716602424018,
"loss": 2.426635437011719,
"step": 9825
},
{
"epoch": 4.7424169475204625,
"grad_norm": 6.533903121948242,
"learning_rate": 0.00011318531197106126,
"loss": 2.5510267639160156,
"step": 9850
},
{
"epoch": 4.754453538757824,
"grad_norm": 6.399111270904541,
"learning_rate": 0.00011279876330237573,
"loss": 2.452451477050781,
"step": 9875
},
{
"epoch": 4.766490129995185,
"grad_norm": 6.134057998657227,
"learning_rate": 0.00011241202011131633,
"loss": 2.4819575500488282,
"step": 9900
},
{
"epoch": 4.778526721232547,
"grad_norm": 6.859342575073242,
"learning_rate": 0.00011202508827581066,
"loss": 2.471083221435547,
"step": 9925
},
{
"epoch": 4.790563312469908,
"grad_norm": 6.8716206550598145,
"learning_rate": 0.00011163797367665338,
"loss": 2.4077735900878907,
"step": 9950
},
{
"epoch": 4.80259990370727,
"grad_norm": 6.526734352111816,
"learning_rate": 0.00011125068219741695,
"loss": 2.4468711853027343,
"step": 9975
},
{
"epoch": 4.814636494944631,
"grad_norm": 6.951447486877441,
"learning_rate": 0.00011086321972436202,
"loss": 2.467903594970703,
"step": 10000
},
{
"epoch": 4.8266730861819935,
"grad_norm": 6.659648418426514,
"learning_rate": 0.00011047559214634823,
"loss": 2.462569580078125,
"step": 10025
},
{
"epoch": 4.838709677419355,
"grad_norm": 7.650249481201172,
"learning_rate": 0.00011008780535474457,
"loss": 2.4519964599609376,
"step": 10050
},
{
"epoch": 4.850746268656716,
"grad_norm": 7.839598655700684,
"learning_rate": 0.00010969986524333973,
"loss": 2.425300750732422,
"step": 10075
},
{
"epoch": 4.862782859894078,
"grad_norm": 5.824965476989746,
"learning_rate": 0.00010931177770825272,
"loss": 2.4256991577148437,
"step": 10100
},
{
"epoch": 4.874819451131439,
"grad_norm": 6.790541648864746,
"learning_rate": 0.00010892354864784318,
"loss": 2.487823028564453,
"step": 10125
},
{
"epoch": 4.886856042368801,
"grad_norm": 7.4350996017456055,
"learning_rate": 0.00010853518396262166,
"loss": 2.4774606323242185,
"step": 10150
},
{
"epoch": 4.898892633606163,
"grad_norm": 6.516840934753418,
"learning_rate": 0.00010814668955516007,
"loss": 2.44733154296875,
"step": 10175
},
{
"epoch": 4.910929224843525,
"grad_norm": 6.86664342880249,
"learning_rate": 0.0001077580713300019,
"loss": 2.3821670532226564,
"step": 10200
},
{
"epoch": 4.922965816080886,
"grad_norm": 6.734835147857666,
"learning_rate": 0.00010736933519357248,
"loss": 2.4690182495117186,
"step": 10225
},
{
"epoch": 4.9350024073182475,
"grad_norm": 7.759254455566406,
"learning_rate": 0.00010698048705408918,
"loss": 2.461112060546875,
"step": 10250
},
{
"epoch": 4.947038998555609,
"grad_norm": 6.959079265594482,
"learning_rate": 0.00010659153282147167,
"loss": 2.455868835449219,
"step": 10275
},
{
"epoch": 4.95907558979297,
"grad_norm": 6.43532133102417,
"learning_rate": 0.00010620247840725217,
"loss": 2.4868319702148436,
"step": 10300
},
{
"epoch": 4.971112181030332,
"grad_norm": 6.879856109619141,
"learning_rate": 0.00010581332972448534,
"loss": 2.4183367919921874,
"step": 10325
},
{
"epoch": 4.983148772267694,
"grad_norm": 6.395660877227783,
"learning_rate": 0.00010542409268765876,
"loss": 2.403534393310547,
"step": 10350
},
{
"epoch": 4.995185363505056,
"grad_norm": 5.965160846710205,
"learning_rate": 0.00010503477321260273,
"loss": 2.441262054443359,
"step": 10375
},
{
"epoch": 5.0,
"eval_loss": 2.610805034637451,
"eval_runtime": 4.3088,
"eval_samples_per_second": 348.121,
"eval_steps_per_second": 10.908,
"step": 10385
},
{
"epoch": 5.007221954742417,
"grad_norm": 7.076479434967041,
"learning_rate": 0.00010464537721640056,
"loss": 2.3097071838378906,
"step": 10400
},
{
"epoch": 5.0192585459797785,
"grad_norm": 7.681911468505859,
"learning_rate": 0.00010425591061729856,
"loss": 2.3071243286132814,
"step": 10425
},
{
"epoch": 5.03129513721714,
"grad_norm": 7.441799640655518,
"learning_rate": 0.00010386637933461603,
"loss": 2.2487457275390623,
"step": 10450
},
{
"epoch": 5.043331728454501,
"grad_norm": 7.052056312561035,
"learning_rate": 0.0001034767892886555,
"loss": 2.2547097778320313,
"step": 10475
},
{
"epoch": 5.055368319691863,
"grad_norm": 6.836119651794434,
"learning_rate": 0.00010308714640061252,
"loss": 2.280919189453125,
"step": 10500
},
{
"epoch": 5.067404910929225,
"grad_norm": 8.23390007019043,
"learning_rate": 0.00010269745659248575,
"loss": 2.189468994140625,
"step": 10525
},
{
"epoch": 5.079441502166587,
"grad_norm": 7.598567962646484,
"learning_rate": 0.00010230772578698704,
"loss": 2.2594345092773436,
"step": 10550
},
{
"epoch": 5.091478093403948,
"grad_norm": 6.551168441772461,
"learning_rate": 0.00010191795990745125,
"loss": 2.32159912109375,
"step": 10575
},
{
"epoch": 5.1035146846413095,
"grad_norm": 7.032190322875977,
"learning_rate": 0.0001015281648777464,
"loss": 2.2451826477050782,
"step": 10600
},
{
"epoch": 5.115551275878671,
"grad_norm": 8.39832878112793,
"learning_rate": 0.00010113834662218353,
"loss": 2.2589088439941407,
"step": 10625
},
{
"epoch": 5.127587867116032,
"grad_norm": 7.9065351486206055,
"learning_rate": 0.00010074851106542662,
"loss": 2.293925476074219,
"step": 10650
},
{
"epoch": 5.139624458353394,
"grad_norm": 7.675061225891113,
"learning_rate": 0.00010035866413240264,
"loss": 2.323810729980469,
"step": 10675
},
{
"epoch": 5.151661049590756,
"grad_norm": 6.749380111694336,
"learning_rate": 9.996881174821147e-05,
"loss": 2.256572570800781,
"step": 10700
},
{
"epoch": 5.163697640828118,
"grad_norm": 7.127290725708008,
"learning_rate": 9.957895983803586e-05,
"loss": 2.270260925292969,
"step": 10725
},
{
"epoch": 5.175734232065479,
"grad_norm": 7.5138020515441895,
"learning_rate": 9.918911432705131e-05,
"loss": 2.293995666503906,
"step": 10750
},
{
"epoch": 5.187770823302841,
"grad_norm": 7.370467185974121,
"learning_rate": 9.879928114033611e-05,
"loss": 2.273690490722656,
"step": 10775
},
{
"epoch": 5.199807414540202,
"grad_norm": 7.263876914978027,
"learning_rate": 9.840946620278117e-05,
"loss": 2.3219985961914062,
"step": 10800
},
{
"epoch": 5.2118440057775635,
"grad_norm": 7.593104362487793,
"learning_rate": 9.801967543900015e-05,
"loss": 2.263868713378906,
"step": 10825
},
{
"epoch": 5.223880597014926,
"grad_norm": 6.794946193695068,
"learning_rate": 9.762991477323916e-05,
"loss": 2.3010122680664065,
"step": 10850
},
{
"epoch": 5.235917188252287,
"grad_norm": 7.785322189331055,
"learning_rate": 9.724019012928693e-05,
"loss": 2.30885986328125,
"step": 10875
},
{
"epoch": 5.247953779489649,
"grad_norm": 7.871342658996582,
"learning_rate": 9.68505074303848e-05,
"loss": 2.3073556518554685,
"step": 10900
},
{
"epoch": 5.25999037072701,
"grad_norm": 7.8656768798828125,
"learning_rate": 9.646087259913647e-05,
"loss": 2.360163269042969,
"step": 10925
},
{
"epoch": 5.272026961964372,
"grad_norm": 8.353021621704102,
"learning_rate": 9.607129155741822e-05,
"loss": 2.3517471313476563,
"step": 10950
},
{
"epoch": 5.284063553201733,
"grad_norm": 6.3557047843933105,
"learning_rate": 9.568177022628874e-05,
"loss": 2.3101840209960938,
"step": 10975
},
{
"epoch": 5.2961001444390945,
"grad_norm": 8.154989242553711,
"learning_rate": 9.529231452589923e-05,
"loss": 2.2766015625,
"step": 11000
},
{
"epoch": 5.308136735676457,
"grad_norm": 8.35250473022461,
"learning_rate": 9.490293037540345e-05,
"loss": 2.2737637329101563,
"step": 11025
},
{
"epoch": 5.320173326913818,
"grad_norm": 6.666666030883789,
"learning_rate": 9.451362369286761e-05,
"loss": 2.3025851440429688,
"step": 11050
},
{
"epoch": 5.33220991815118,
"grad_norm": 6.981100559234619,
"learning_rate": 9.412440039518057e-05,
"loss": 2.3344212341308594,
"step": 11075
},
{
"epoch": 5.344246509388541,
"grad_norm": 8.032634735107422,
"learning_rate": 9.373526639796396e-05,
"loss": 2.2642364501953125,
"step": 11100
},
{
"epoch": 5.356283100625903,
"grad_norm": 7.3971638679504395,
"learning_rate": 9.334622761548197e-05,
"loss": 2.3268336486816406,
"step": 11125
},
{
"epoch": 5.368319691863264,
"grad_norm": 6.635739326477051,
"learning_rate": 9.295728996055182e-05,
"loss": 2.2535565185546873,
"step": 11150
},
{
"epoch": 5.3803562831006255,
"grad_norm": 6.337807655334473,
"learning_rate": 9.256845934445368e-05,
"loss": 2.308728790283203,
"step": 11175
},
{
"epoch": 5.392392874337988,
"grad_norm": 8.622179985046387,
"learning_rate": 9.217974167684094e-05,
"loss": 2.2889668273925783,
"step": 11200
},
{
"epoch": 5.404429465575349,
"grad_norm": 6.722035884857178,
"learning_rate": 9.179114286565026e-05,
"loss": 2.3064193725585938,
"step": 11225
},
{
"epoch": 5.416466056812711,
"grad_norm": 8.097888946533203,
"learning_rate": 9.140266881701192e-05,
"loss": 2.280697937011719,
"step": 11250
},
{
"epoch": 5.428502648050072,
"grad_norm": 7.873524188995361,
"learning_rate": 9.101432543516e-05,
"loss": 2.3224221801757814,
"step": 11275
},
{
"epoch": 5.440539239287434,
"grad_norm": 7.560741424560547,
"learning_rate": 9.062611862234254e-05,
"loss": 2.3603790283203123,
"step": 11300
},
{
"epoch": 5.452575830524795,
"grad_norm": 7.508003234863281,
"learning_rate": 9.023805427873206e-05,
"loss": 2.2686558532714844,
"step": 11325
},
{
"epoch": 5.464612421762157,
"grad_norm": 6.625362873077393,
"learning_rate": 8.985013830233573e-05,
"loss": 2.2339849853515625,
"step": 11350
},
{
"epoch": 5.476649012999519,
"grad_norm": 7.706381320953369,
"learning_rate": 8.946237658890564e-05,
"loss": 2.3134950256347655,
"step": 11375
},
{
"epoch": 5.48868560423688,
"grad_norm": 7.328007698059082,
"learning_rate": 8.90747750318494e-05,
"loss": 2.3151211547851562,
"step": 11400
},
{
"epoch": 5.500722195474242,
"grad_norm": 7.3665385246276855,
"learning_rate": 8.868733952214055e-05,
"loss": 2.2758139038085936,
"step": 11425
},
{
"epoch": 5.512758786711603,
"grad_norm": 7.762585163116455,
"learning_rate": 8.830007594822882e-05,
"loss": 2.269935302734375,
"step": 11450
},
{
"epoch": 5.524795377948965,
"grad_norm": 6.537332057952881,
"learning_rate": 8.791299019595082e-05,
"loss": 2.325925750732422,
"step": 11475
},
{
"epoch": 5.536831969186326,
"grad_norm": 6.946711540222168,
"learning_rate": 8.752608814844059e-05,
"loss": 2.3549649047851564,
"step": 11500
},
{
"epoch": 5.5488685604236885,
"grad_norm": 7.574845314025879,
"learning_rate": 8.713937568604005e-05,
"loss": 2.273140563964844,
"step": 11525
},
{
"epoch": 5.56090515166105,
"grad_norm": 7.267978668212891,
"learning_rate": 8.675285868620971e-05,
"loss": 2.3256234741210937,
"step": 11550
},
{
"epoch": 5.572941742898411,
"grad_norm": 6.712131500244141,
"learning_rate": 8.636654302343941e-05,
"loss": 2.31070556640625,
"step": 11575
},
{
"epoch": 5.584978334135773,
"grad_norm": 7.248861789703369,
"learning_rate": 8.598043456915887e-05,
"loss": 2.327481231689453,
"step": 11600
},
{
"epoch": 5.597014925373134,
"grad_norm": 7.805160045623779,
"learning_rate": 8.559453919164867e-05,
"loss": 2.3892562866210936,
"step": 11625
},
{
"epoch": 5.609051516610496,
"grad_norm": 8.140661239624023,
"learning_rate": 8.520886275595072e-05,
"loss": 2.2867330932617187,
"step": 11650
},
{
"epoch": 5.621088107847857,
"grad_norm": 7.571468353271484,
"learning_rate": 8.482341112377956e-05,
"loss": 2.328978729248047,
"step": 11675
},
{
"epoch": 5.633124699085219,
"grad_norm": 6.735067367553711,
"learning_rate": 8.443819015343295e-05,
"loss": 2.302386932373047,
"step": 11700
},
{
"epoch": 5.645161290322581,
"grad_norm": 8.074029922485352,
"learning_rate": 8.40532056997029e-05,
"loss": 2.277113952636719,
"step": 11725
},
{
"epoch": 5.657197881559942,
"grad_norm": 8.525930404663086,
"learning_rate": 8.366846361378682e-05,
"loss": 2.329440002441406,
"step": 11750
},
{
"epoch": 5.669234472797304,
"grad_norm": 7.668955326080322,
"learning_rate": 8.328396974319835e-05,
"loss": 2.279188232421875,
"step": 11775
},
{
"epoch": 5.681271064034665,
"grad_norm": 7.235455513000488,
"learning_rate": 8.289972993167873e-05,
"loss": 2.3442225646972656,
"step": 11800
},
{
"epoch": 5.693307655272027,
"grad_norm": 6.819657325744629,
"learning_rate": 8.251575001910785e-05,
"loss": 2.2605599975585937,
"step": 11825
},
{
"epoch": 5.705344246509388,
"grad_norm": 7.259073734283447,
"learning_rate": 8.213203584141545e-05,
"loss": 2.3369517517089844,
"step": 11850
},
{
"epoch": 5.7173808377467505,
"grad_norm": 8.00145149230957,
"learning_rate": 8.174859323049262e-05,
"loss": 2.298871612548828,
"step": 11875
},
{
"epoch": 5.729417428984112,
"grad_norm": 7.8096723556518555,
"learning_rate": 8.136542801410292e-05,
"loss": 2.2825933837890626,
"step": 11900
},
{
"epoch": 5.741454020221473,
"grad_norm": 7.057461261749268,
"learning_rate": 8.098254601579398e-05,
"loss": 2.265797119140625,
"step": 11925
},
{
"epoch": 5.753490611458835,
"grad_norm": 7.485247611999512,
"learning_rate": 8.059995305480885e-05,
"loss": 2.2326194763183596,
"step": 11950
},
{
"epoch": 5.765527202696196,
"grad_norm": 6.613664627075195,
"learning_rate": 8.02176549459978e-05,
"loss": 2.288837585449219,
"step": 11975
},
{
"epoch": 5.777563793933558,
"grad_norm": 7.495915412902832,
"learning_rate": 7.983565749972961e-05,
"loss": 2.341431884765625,
"step": 12000
},
{
"epoch": 5.789600385170919,
"grad_norm": 7.257405757904053,
"learning_rate": 7.945396652180352e-05,
"loss": 2.2813145446777345,
"step": 12025
},
{
"epoch": 5.801636976408281,
"grad_norm": 7.222981929779053,
"learning_rate": 7.907258781336091e-05,
"loss": 2.299049987792969,
"step": 12050
},
{
"epoch": 5.813673567645643,
"grad_norm": 7.421830177307129,
"learning_rate": 7.86915271707971e-05,
"loss": 2.2596804809570314,
"step": 12075
},
{
"epoch": 5.8257101588830045,
"grad_norm": 7.540179252624512,
"learning_rate": 7.831079038567323e-05,
"loss": 2.2943069458007814,
"step": 12100
},
{
"epoch": 5.837746750120366,
"grad_norm": 7.248844623565674,
"learning_rate": 7.793038324462838e-05,
"loss": 2.279644775390625,
"step": 12125
},
{
"epoch": 5.849783341357727,
"grad_norm": 7.413556098937988,
"learning_rate": 7.755031152929149e-05,
"loss": 2.2773260498046874,
"step": 12150
},
{
"epoch": 5.861819932595089,
"grad_norm": 7.529948711395264,
"learning_rate": 7.717058101619343e-05,
"loss": 2.2864813232421874,
"step": 12175
},
{
"epoch": 5.87385652383245,
"grad_norm": 6.944470405578613,
"learning_rate": 7.679119747667942e-05,
"loss": 2.292400207519531,
"step": 12200
},
{
"epoch": 5.885893115069813,
"grad_norm": 6.74458122253418,
"learning_rate": 7.641216667682115e-05,
"loss": 2.2836102294921874,
"step": 12225
},
{
"epoch": 5.897929706307174,
"grad_norm": 7.522007942199707,
"learning_rate": 7.603349437732916e-05,
"loss": 2.2751849365234373,
"step": 12250
},
{
"epoch": 5.9099662975445355,
"grad_norm": 7.773344039916992,
"learning_rate": 7.565518633346537e-05,
"loss": 2.3335125732421873,
"step": 12275
},
{
"epoch": 5.922002888781897,
"grad_norm": 7.2289042472839355,
"learning_rate": 7.527724829495551e-05,
"loss": 2.2528985595703124,
"step": 12300
},
{
"epoch": 5.934039480019258,
"grad_norm": 6.855514049530029,
"learning_rate": 7.489968600590174e-05,
"loss": 2.337420654296875,
"step": 12325
},
{
"epoch": 5.94607607125662,
"grad_norm": 7.413761615753174,
"learning_rate": 7.452250520469548e-05,
"loss": 2.3095712280273437,
"step": 12350
},
{
"epoch": 5.958112662493981,
"grad_norm": 7.350351333618164,
"learning_rate": 7.414571162392998e-05,
"loss": 2.2485447692871094,
"step": 12375
},
{
"epoch": 5.970149253731344,
"grad_norm": 7.017465114593506,
"learning_rate": 7.37693109903134e-05,
"loss": 2.3014979553222656,
"step": 12400
},
{
"epoch": 5.982185844968705,
"grad_norm": 7.562068939208984,
"learning_rate": 7.339330902458163e-05,
"loss": 2.297460479736328,
"step": 12425
},
{
"epoch": 5.9942224362060665,
"grad_norm": 7.372097015380859,
"learning_rate": 7.301771144141141e-05,
"loss": 2.3374281311035157,
"step": 12450
},
{
"epoch": 6.0,
"eval_loss": 2.547119379043579,
"eval_runtime": 4.1959,
"eval_samples_per_second": 357.491,
"eval_steps_per_second": 11.201,
"step": 12462
}
],
"logging_steps": 25,
"max_steps": 20770,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}