Qwen90M step 950000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 945000,
+  "step": 950000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "d6e242a319b84a2ddbfbdee1f013cedca15ac92daa2465e2dc7e8ee27334bfa5",
+    "model.safetensors": "2e6800a411ea52928020ac842716553e33b1e7f874b99a56f4a6d2a6bbbab114",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "1b526c93db35739f32941c9bc4333b3a51815b006101025f8a5000c026886d7a",
-    "training_config.json": "aafacc05aaf9ea874dcfa3d02457799efaf5525cf7b211a4208541f6e989efee",
+    "trainer.pt": "8ff2ea6bb34408b69ebbd7193690d1ba5ccb5272bc038889553875b472839158",
+    "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4",
-    "shard_scheduler.json": "50acacccffc7b89ec0e3a19cfe617edd4d68fb6d7b4e7825701fea014ffa4daf",
-    "coverage.json": "bdb36980d69eb7e333e36c43d928aead76bc2f1818c6f4d3d8019f2cc03b9b09",
+    "shard_scheduler.json": "b552fa7d9f27edf138aa08049c5810dc5f2939827c4512222fbf3c1ba1750c42",
+    "coverage.json": "b37ea7e378a52a99096460a590042a103f46dd59f031c3a7ade67958795c5436",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,101 +1,107 @@
 {
-  "step": 945000,
+  "step": 950000,
   "stats": {
-    "climbmix/blocks": 34020000,
-    "climbmix/targets": 56870861475,
-    "climbmix/input_tokens": 56874465591,
-    "ultra_qa/blocks": 12757500,
-    "ultra_qa/targets": 20812763599,
-    "ultra_qa/input_tokens": 20812803585,
-    "cortex/blocks": 12757500,
-    "cortex/targets": 485107565,
-    "cortex/input_tokens": 20279107177,
-    "ultra_style/blocks": 12757500,
-    "ultra_style/targets": 20177960264,
-    "ultra_style/input_tokens": 20179017624,
-    "openmath/blocks": 8505000,
-    "openmath/targets": 13141616824,
-    "openmath/input_tokens": 13145995633,
-    "rewrite6/blocks": 4252500,
-    "rewrite6/targets": 6971421582,
-    "rewrite6/input_tokens": 6979064617
+    "climbmix/blocks": 34200000,
+    "climbmix/targets": 57171948898,
+    "climbmix/input_tokens": 57175571843,
+    "ultra_qa/blocks": 12825000,
+    "ultra_qa/targets": 20922958426,
+    "ultra_qa/input_tokens": 20922998603,
+    "cortex/blocks": 12825000,
+    "cortex/targets": 487674525,
+    "cortex/input_tokens": 20386543238,
+    "ultra_style/blocks": 12825000,
+    "ultra_style/targets": 20285002197,
+    "ultra_style/input_tokens": 20286064763,
+    "openmath/blocks": 8550000,
+    "openmath/targets": 13211092129,
+    "openmath/input_tokens": 13215494259,
+    "rewrite6/blocks": 4275000,
+    "rewrite6/targets": 7008276799,
+    "rewrite6/input_tokens": 7015960317
   },
   "sources": {
     "climbmix": {
-      "page": 4659,
-      "row": 1681,
+      "page": 4683,
+      "row": 3857,
       "start": {
-        "file": 225,
+        "file": 226,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 43,
+          "row_group": 57,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 19084945
+      "offset": 8192,
+      "documents": 19185425
     },
     "rewrite6": {
-      "page": 0,
-      "row": 2278,
-      "start": null,
+      "page": 14,
+      "row": 2553,
+      "start": {
+        "file": 0,
+        "reader": {
+          "format": "jsonl-row-v1",
+          "row": 57344
+        }
+      },
       "epoch": 53,
-      "offset": 2048,
-      "documents": 10888690
+      "offset": 0,
+      "documents": 10946309
     },
     "openmath": {
-      "page": 2911,
-      "row": 4058,
+      "page": 2927,
+      "row": 1756,
       "start": {
         "file": 27,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 133,
-          "row": 910
+          "row_group": 199,
+          "row": 446
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 11927514
+      "documents": 11990748
     },
     "ultra_qa": {
-      "page": 1258,
-      "row": 4095,
+      "page": 1265,
+      "row": 2680,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 24209
+          "row": 52881
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 5156863
+      "offset": 4096,
+      "documents": 5184120
     },
     "ultra_style": {
-      "page": 1921,
-      "row": 1309,
+      "page": 1931,
+      "row": 1294,
       "start": {
         "file": 11,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 78628
+          "row_group": 4,
+          "row": 35497
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 7869725
+      "offset": 0,
+      "documents": 7910670
     },
     "cortex": {
-      "page": 247415,
-      "row": 48,
+      "page": 248725,
+      "row": 90,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 31669168
+      "documents": 31836890
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 945000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "d6e242a319b84a2ddbfbdee1f013cedca15ac92daa2465e2dc7e8ee27334bfa5", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "1b526c93db35739f32941c9bc4333b3a51815b006101025f8a5000c026886d7a", "training_config.json": "aafacc05aaf9ea874dcfa3d02457799efaf5525cf7b211a4208541f6e989efee", "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4", "shard_scheduler.json": "50acacccffc7b89ec0e3a19cfe617edd4d68fb6d7b4e7825701fea014ffa4daf", "coverage.json": "bdb36980d69eb7e333e36c43d928aead76bc2f1818c6f4d3d8019f2cc03b9b09", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000945000"}
+{"step": 950000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "2e6800a411ea52928020ac842716553e33b1e7f874b99a56f4a6d2a6bbbab114", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "8ff2ea6bb34408b69ebbd7193690d1ba5ccb5272bc038889553875b472839158", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4", "shard_scheduler.json": "b552fa7d9f27edf138aa08049c5810dc5f2939827c4512222fbf3c1ba1750c42", "coverage.json": "b37ea7e378a52a99096460a590042a103f46dd59f031c3a7ade67958795c5436", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000950000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d6e242a319b84a2ddbfbdee1f013cedca15ac92daa2465e2dc7e8ee27334bfa5
+oid sha256:2e6800a411ea52928020ac842716553e33b1e7f874b99a56f4a6d2a6bbbab114
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 598315.063954628,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 77415.48812361607,
-      "required_pages_per_second_upper_estimate": 0.005355783097072209
+      "required_token_positions_per_second": 76668.71420286849,
+      "required_pages_per_second_upper_estimate": 0.005304119544480668
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 8533.957213384612,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9676.936015452009,
-      "required_pages_per_second_upper_estimate": 0.004051531032266766
+      "required_token_positions_per_second": 9583.589275358561,
+      "required_pages_per_second_upper_estimate": 0.004012448701491235
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 380144.35761751566,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 19353.872030904018,
-      "required_pages_per_second_upper_estimate": 0.004030901148109212
+      "required_token_positions_per_second": 19167.178550717123,
+      "required_pages_per_second_upper_estimate": 0.003992017819624442
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 43742.16556775435,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 29030.808046356025,
-      "required_pages_per_second_upper_estimate": 0.001740352614901801
+      "required_token_positions_per_second": 28750.767826075684,
+      "required_pages_per_second_upper_estimate": 0.0017235646312926524
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 22018.86276523721,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 29030.808046356025,
-      "required_pages_per_second_upper_estimate": 0.0022570176325746004
+      "required_token_positions_per_second": 28750.767826075684,
+      "required_pages_per_second_upper_estimate": 0.0022352457372145557
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 910315.184299418,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 23,
-      "required_token_positions_per_second": 29030.808046356025,
-      "required_pages_per_second_upper_estimate": 0.37237603476553693
+      "required_token_positions_per_second": 28750.767826075684,
+      "required_pages_per_second_upper_estimate": 0.3687839795035426
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,412 +155,412 @@
   "shard_benchmark_scope": "Local packing/read measurements; upload capacity measured only on real data commits. Targets larger than available pages are sample-limited.",
   "estimated_fresh_generation_batch_seconds": 2.558590814491007,
   "selected": {
-    "prefetch_batches": 64,
+    "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.4394334593517044,
-    "producer_p95_seconds": 0.21852520599713898,
-    "gpu_update_seconds": 0.9523675660647138,
-    "producer_batches_per_second": 2.275657391850176,
-    "gpu_batches_per_second": 1.0500147586210948,
-    "producer_headroom": 2.167262291473527,
+    "producer_batch_seconds": 0.10526186014067207,
+    "producer_p95_seconds": 0.27886092599874246,
+    "gpu_update_seconds": 0.9616438826000493,
+    "producer_batches_per_second": 9.500117123748325,
+    "gpu_batches_per_second": 1.039885989079705,
+    "producer_headroom": 9.135729516036552,
     "sustainable": true,
-    "buffer_seconds": 60.951524228141686,
+    "buffer_seconds": 60.583564603803104,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 81,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 296988.82874613756,
-    "upload_bytes_per_second": 4767584.704755423,
+    "generation_bytes_per_second": 357854.62686666014,
+    "upload_bytes_per_second": 4807959.195163555,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 63.726889102822255,
+    "estimated_full_shards_per_hour": 76.78727249622204,
     "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
   },
   "upload_measurements": [
     {
-      "seconds": 1.5571684259994072,
-      "bytes": 5848121,
-      "pages": 28,
+      "seconds": 1.4908248419997108,
+      "bytes": 6812358,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.1272584719990846,
-      "bytes": 2509117,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.521788887999719,
+      "bytes": 7493690,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.3743308060002164,
-      "bytes": 6082381,
-      "pages": 31,
+      "seconds": 1.3138526910006476,
+      "bytes": 6032843,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.292268752000382,
-      "bytes": 11000537,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4258081890002359,
-      "bytes": 8234938,
-      "pages": 31,
+      "seconds": 1.2020050520004588,
+      "bytes": 6391888,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.4250404260001233,
-      "bytes": 6072497,
-      "pages": 36,
+      "seconds": 1.6235010039999906,
+      "bytes": 6476306,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.7697906350003905,
-      "bytes": 6241903,
-      "pages": 36,
+      "seconds": 3.9122833619985613,
+      "bytes": 6289054,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.7708508949999668,
-      "bytes": 6504956,
+      "seconds": 1.8021375259995693,
+      "bytes": 8272426,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1802944610008126,
-      "bytes": 6153792,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.257975387001352,
+      "bytes": 10890364,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.060665737999443,
-      "bytes": 2571654,
-      "pages": 33,
+      "seconds": 1.6343959760015423,
+      "bytes": 2597952,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.9573821910016704,
-      "bytes": 16491610,
+      "seconds": 1.3388132420004695,
+      "bytes": 5864353,
       "pages": 35,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 1.2731761789982556,
-      "bytes": 6636284,
-      "pages": 37,
+      "seconds": 1.5614973880001344,
+      "bytes": 5966551,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.2044296969997959,
-      "bytes": 2640995,
-      "pages": 37,
-      "shards": 2
+      "seconds": 3.168707209999411,
+      "bytes": 6305368,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.348795689000326,
-      "bytes": 5804474,
+      "seconds": 1.9134112159990764,
+      "bytes": 11699377,
       "pages": 28,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.1589008739993005,
-      "bytes": 2402568,
-      "pages": 31,
-      "shards": 2
+      "seconds": 0.9410961429966846,
+      "bytes": 1236103,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.2708631089990376,
-      "bytes": 9724430,
+      "seconds": 1.262457610002457,
+      "bytes": 7619870,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.795558247000372,
-      "bytes": 12019830,
-      "pages": 36,
+      "seconds": 1.496031877999485,
+      "bytes": 10106377,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9979082219997508,
-      "bytes": 2716683,
+      "seconds": 1.0667447009982425,
+      "bytes": 2533809,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3286046470002475,
-      "bytes": 11093643,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2714691889996175,
-      "bytes": 7462349,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0150521480009047,
-      "bytes": 842077,
-      "pages": 26,
+      "seconds": 1.2533954940008698,
+      "bytes": 6219875,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1759212130000378,
-      "bytes": 6164782,
-      "pages": 33,
+      "seconds": 1.2242291910006315,
+      "bytes": 6745128,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.4110855310009356,
-      "bytes": 8184458,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.3833632300011232,
+      "bytes": 11874470,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.2917386739991343,
-      "bytes": 10061132,
-      "pages": 42,
-      "shards": 3
+      "seconds": 1.439283288000297,
+      "bytes": 6018491,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.2036675580002338,
-      "bytes": 326353,
-      "pages": 9,
+      "seconds": 0.925738228001137,
+      "bytes": 863516,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.3246287520014448,
-      "bytes": 7379080,
-      "pages": 33,
+      "seconds": 1.338383825001074,
+      "bytes": 7489532,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.0026750559991342,
-      "bytes": 938764,
-      "pages": 31,
-      "shards": 1
+      "seconds": 1.0948180369996408,
+      "bytes": 10021223,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.5529996709992702,
-      "bytes": 10526115,
-      "pages": 38,
+      "seconds": 1.3469298159980099,
+      "bytes": 8271893,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.8902240070001426,
-      "bytes": 13319811,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.2550481799989939,
+      "bytes": 6009424,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.1297152579991234,
-      "bytes": 1247239,
-      "pages": 35,
+      "seconds": 0.9456898729986278,
+      "bytes": 975770,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.127117382000506,
-      "bytes": 7535662,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.7259603879974748,
+      "bytes": 12302933,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 1.3609303779994661,
-      "bytes": 10666543,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.298020391001046,
+      "bytes": 6312632,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.843333258999337,
-      "bytes": 1195587,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.6241299210014404,
+      "bytes": 2390001,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.4843279429987888,
-      "bytes": 7415076,
-      "pages": 39,
+      "seconds": 1.5602491789977648,
+      "bytes": 11841247,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.515185850999842,
-      "bytes": 11700777,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.3097548330006248,
+      "bytes": 6099339,
+      "pages": 40,
+      "shards": 2
     },
     {
-      "seconds": 1.3114747930012527,
-      "bytes": 6594730,
-      "pages": 38,
+      "seconds": 1.2393856140006392,
+      "bytes": 7638444,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1013290299997607,
-      "bytes": 5841716,
-      "pages": 29,
+      "seconds": 1.58254548900004,
+      "bytes": 5975386,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9914871209984994,
-      "bytes": 1095861,
-      "pages": 35,
+      "seconds": 0.9414393419974658,
+      "bytes": 1125185,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.156548383001791,
-      "bytes": 7676323,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.356874795001204,
+      "bytes": 11516865,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.4210737489993335,
-      "bytes": 10660170,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.9571627120021731,
+      "bytes": 762513,
+      "pages": 23,
+      "shards": 1
     },
     {
-      "seconds": 1.4908248419997108,
-      "bytes": 6812358,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.4426862469990738,
+      "bytes": 13691203,
+      "pages": 49,
+      "shards": 4
     },
     {
-      "seconds": 1.521788887999719,
-      "bytes": 7493690,
-      "pages": 39,
-      "shards": 3
+      "seconds": 2.0410720630025025,
+      "bytes": 6120614,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.3138526910006476,
-      "bytes": 6032843,
-      "pages": 36,
+      "seconds": 1.370815718000813,
+      "bytes": 5076360,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2020050520004588,
-      "bytes": 6391888,
-      "pages": 29,
+      "seconds": 1.1959222149998823,
+      "bytes": 7412774,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.6235010039999906,
-      "bytes": 6476306,
+      "seconds": 1.488441301000421,
+      "bytes": 6011041,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 3.9122833619985613,
-      "bytes": 6289054,
-      "pages": 37,
+      "seconds": 1.0648157450013969,
+      "bytes": 668578,
+      "pages": 20,
       "shards": 2
     },
     {
-      "seconds": 1.8021375259995693,
-      "bytes": 8272426,
+      "seconds": 1.9570103289988765,
+      "bytes": 18490744,
+      "pages": 47,
+      "shards": 5
+    },
+    {
+      "seconds": 0.8552882639996824,
+      "bytes": 1175820,
       "pages": 35,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 1.257975387001352,
-      "bytes": 10890364,
+      "seconds": 1.1064946840015182,
+      "bytes": 7293995,
       "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.6343959760015423,
-      "bytes": 2597952,
-      "pages": 37,
+      "seconds": 1.736566771996877,
+      "bytes": 6191401,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3388132420004695,
-      "bytes": 5864353,
-      "pages": 35,
+      "seconds": 1.1331273229989165,
+      "bytes": 4688984,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.5614973880001344,
-      "bytes": 5966551,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.3685046969985706,
+      "bytes": 7267431,
+      "pages": 27,
+      "shards": 3
     },
     {
-      "seconds": 3.168707209999411,
-      "bytes": 6305368,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.4039977110005566,
+      "bytes": 12182760,
+      "pages": 48,
+      "shards": 3
     },
     {
-      "seconds": 1.9134112159990764,
-      "bytes": 11699377,
-      "pages": 28,
+      "seconds": 1.015954297999997,
+      "bytes": 2619113,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3503473790005955,
+      "bytes": 10477535,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.9410961429966846,
-      "bytes": 1236103,
-      "pages": 35,
+      "seconds": 0.9818654609989608,
+      "bytes": 963058,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.262457610002457,
-      "bytes": 7619870,
-      "pages": 38,
+      "seconds": 1.1789819430014177,
+      "bytes": 7759537,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.496031877999485,
-      "bytes": 10106377,
-      "pages": 30,
+      "seconds": 1.2661478619993431,
+      "bytes": 11756535,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.0667447009982425,
-      "bytes": 2533809,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.2254978609998943,
+      "bytes": 7186634,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.2533954940008698,
-      "bytes": 6219875,
-      "pages": 36,
+      "seconds": 1.2788624959976005,
+      "bytes": 6210071,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2242291910006315,
-      "bytes": 6745128,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.412427707000461,
+      "bytes": 887856,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.3833632300011232,
-      "bytes": 11874470,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.2455697520017566,
+      "bytes": 7910132,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.439283288000297,
-      "bytes": 6018491,
+      "seconds": 1.3195631069975207,
+      "bytes": 10098822,
       "pages": 36,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.925738228001137,
-      "bytes": 863516,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.6743710149967228,
+      "bytes": 8331470,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.338383825001074,
-      "bytes": 7489532,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.060381152001355,
+      "bytes": 6181629,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.0948180369996408,
-      "bytes": 10021223,
-      "pages": 34,
+      "seconds": 1.3323454749988741,
+      "bytes": 6231984,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3298568760001217,
+      "bytes": 7493698,
+      "pages": 36,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b526c93db35739f32941c9bc4333b3a51815b006101025f8a5000c026886d7a
-size 721788287
+oid sha256:8ff2ea6bb34408b69ebbd7193690d1ba5ccb5272bc038889553875b472839158
+size 721758527

training_config.json

@@ -53,8 +53,8 @@
   "cpu_threads": 4,
   "page_rows": 4096,
   "cortex_page_rows": 128,
-  "prefetch_batches": 64,
-  "prefetch_low_water": 32,
+  "prefetch_batches": 63,
+  "prefetch_low_water": 31,
   "memory_pages": 12,
   "cache_gb": 20,
   "force_rebenchmark": false,