Qwen90M step 250000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 245000,
+  "step": 250000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "80d6299138410b2e8858ba12acf696cdf3176dcd9f400bb585856b76acf7d993",
+    "model.safetensors": "b39daf23cbfc7a4db45dc715dda0257d7a0aceacebd6722e70e52b392a3d914f",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "727525a389ec5f45c0e2113ae9043f1f725bf96b0fc45fa3f07a6899b9d03b8b",
+    "trainer.pt": "e91afb32a91ba214e7473f46d6f99a0a931e405920d457559aea60d50123c7e4",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7",
-    "shard_scheduler.json": "53335450f9b4b65852a99c22ef387c6c7d63ed363bbf58d23d279cc75e51158e",
-    "coverage.json": "74e3063ad472df43fbe8188bf2dafe200f98ec7afd81e2c0b072fc086e471069",
+    "shard_scheduler.json": "51e06602336cf2ff9c7fbf2702cb9edfe3a4879775d6e82cb900f0895494a34a",
+    "coverage.json": "fa6f7ebd74b32c3a670cb83966d87b06345ed8c181b0803237a454541a8aa878",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 245000,
+  "step": 250000,
   "stats": {
-    "climbmix/blocks": 8820000,
-    "climbmix/targets": 14743955462,
-    "climbmix/input_tokens": 14744889647,
-    "ultra_qa/blocks": 3307500,
-    "ultra_qa/targets": 5395722257,
-    "ultra_qa/input_tokens": 5395732630,
-    "cortex/blocks": 3307500,
-    "cortex/targets": 125836183,
-    "cortex/input_tokens": 5257347187,
-    "ultra_style/blocks": 3307500,
-    "ultra_style/targets": 5231835740,
-    "ultra_style/input_tokens": 5232110004,
-    "openmath/blocks": 2205000,
-    "openmath/targets": 3406996250,
-    "openmath/input_tokens": 3408131868,
-    "rewrite6/blocks": 1102500,
-    "rewrite6/targets": 1807346506,
-    "rewrite6/input_tokens": 1809328100
+    "climbmix/blocks": 9000000,
+    "climbmix/targets": 15044821319,
+    "climbmix/input_tokens": 15045774685,
+    "ultra_qa/blocks": 3375000,
+    "ultra_qa/targets": 5505900381,
+    "ultra_qa/input_tokens": 5505910963,
+    "cortex/blocks": 3375000,
+    "cortex/targets": 128400391,
+    "cortex/input_tokens": 5364742698,
+    "ultra_style/blocks": 3375000,
+    "ultra_style/targets": 5338964418,
+    "ultra_style/input_tokens": 5339244172,
+    "openmath/blocks": 2250000,
+    "openmath/targets": 3476637803,
+    "openmath/input_tokens": 3477796696,
+    "rewrite6/blocks": 1125000,
+    "rewrite6/targets": 1844270111,
+    "rewrite6/input_tokens": 1846292063
   },
   "sources": {
     "climbmix": {
-      "page": 1207,
-      "row": 3232,
+      "page": 1232,
+      "row": 2049,
       "start": {
-        "file": 58,
+        "file": 59,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 35,
+          "row_group": 53,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 12288,
-      "documents": 4947104
+      "offset": 2048,
+      "documents": 5048321
     },
     "rewrite6": {
-      "page": 37,
-      "row": 1322,
+      "page": 1,
+      "row": 847,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 151552
+          "row": 4096
         }
       },
-      "epoch": 13,
+      "epoch": 14,
       "offset": 0,
-      "documents": 2823126
+      "documents": 2880599
     },
     "openmath": {
-      "page": 755,
-      "row": 365,
+      "page": 770,
+      "row": 2180,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 35,
-          "row": 930
+          "row_group": 97,
+          "row": 370
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 3092845
+      "documents": 3156100
     },
     "ultra_qa": {
-      "page": 326,
-      "row": 1777,
+      "page": 333,
+      "row": 364,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 37728
+          "row_group": 5,
+          "row": 1742
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1337073
+      "documents": 1364332
     },
     "ultra_style": {
-      "page": 498,
-      "row": 978,
+      "page": 508,
+      "row": 1215,
       "start": {
-        "file": 2,
+        "file": 3,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 7,
-          "row": 74335
+          "row_group": 0,
+          "row": 25940
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 2040786
+      "offset": 2048,
+      "documents": 2081983
     },
     "cortex": {
-      "page": 64142,
-      "row": 98,
+      "page": 65453,
+      "row": 34,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 8210274
+      "documents": 8378018
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 245000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "80d6299138410b2e8858ba12acf696cdf3176dcd9f400bb585856b76acf7d993", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "727525a389ec5f45c0e2113ae9043f1f725bf96b0fc45fa3f07a6899b9d03b8b", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7", "shard_scheduler.json": "53335450f9b4b65852a99c22ef387c6c7d63ed363bbf58d23d279cc75e51158e", "coverage.json": "74e3063ad472df43fbe8188bf2dafe200f98ec7afd81e2c0b072fc086e471069", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000245000"}
+{"step": 250000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b39daf23cbfc7a4db45dc715dda0257d7a0aceacebd6722e70e52b392a3d914f", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "e91afb32a91ba214e7473f46d6f99a0a931e405920d457559aea60d50123c7e4", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7", "shard_scheduler.json": "51e06602336cf2ff9c7fbf2702cb9edfe3a4879775d6e82cb900f0895494a34a", "coverage.json": "fa6f7ebd74b32c3a670cb83966d87b06345ed8c181b0803237a454541a8aa878", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000250000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:80d6299138410b2e8858ba12acf696cdf3176dcd9f400bb585856b76acf7d993
+oid sha256:b39daf23cbfc7a4db45dc715dda0257d7a0aceacebd6722e70e52b392a3d914f
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 1062967.6168812835,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 75420.9647685022,
-      "required_pages_per_second_upper_estimate": 0.006537298073446168
+      "required_token_positions_per_second": 75570.71051583829,
+      "required_pages_per_second_upper_estimate": 0.006550277655298134
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 1206672.263377472,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9427.620596062776,
-      "required_pages_per_second_upper_estimate": 0.0037608188112584874
+      "required_token_positions_per_second": 9446.338814479786,
+      "required_pages_per_second_upper_estimate": 0.003768285788447338
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 410042.7581134458,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 18855.24119212555,
-      "required_pages_per_second_upper_estimate": 0.0043618282088870395
+      "required_token_positions_per_second": 18892.677628959573,
+      "required_pages_per_second_upper_estimate": 0.004370488469689805
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 2163486.5624068812,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28282.86178818833,
-      "required_pages_per_second_upper_estimate": 0.0017222778941929503
+      "required_token_positions_per_second": 28339.01644343936,
+      "required_pages_per_second_upper_estimate": 0.0017256974180770313
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 1546694.2737633225,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28282.86178818833,
-      "required_pages_per_second_upper_estimate": 0.0021772929924270767
+      "required_token_positions_per_second": 28339.01644343936,
+      "required_pages_per_second_upper_estimate": 0.0021816159332343353
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 1015177.8395898147,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28282.86178818833,
-      "required_pages_per_second_upper_estimate": 0.35117848675997776
+      "required_token_positions_per_second": 28339.01644343936,
+      "required_pages_per_second_upper_estimate": 0.3518757396406541
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 690670,
-    "producer_batch_seconds": 0.5026795613593151,
-    "producer_p95_seconds": 0.20843738300027326,
-    "gpu_update_seconds": 0.9775531276522568,
-    "producer_batches_per_second": 1.9893388887661585,
-    "gpu_batches_per_second": 1.0229623042602838,
-    "producer_headroom": 1.9446844526736233,
+    "producer_batch_seconds": 0.129124772499992,
+    "producer_p95_seconds": 0.24966627399953722,
+    "gpu_update_seconds": 0.9756160752855156,
+    "producer_batches_per_second": 7.744447332908656,
+    "gpu_batches_per_second": 1.0249933609461575,
+    "producer_headroom": 7.555607312187723,
     "sustainable": true,
-    "buffer_seconds": 60.60829391443992,
+    "buffer_seconds": 60.48819666770197,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
     "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 284788.4705175513,
-    "upload_bytes_per_second": 3726836.968530263,
+    "generation_bytes_per_second": 382211.29352878686,
+    "upload_bytes_per_second": 4854486.8576622605,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 61.108976236771625,
+    "estimated_full_shards_per_hour": 82.01364616773323,
     "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": 1.4821450189999723,
-      "bytes": 5748205,
-      "pages": 24,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3858465779999278,
-      "bytes": 775414,
-      "pages": 24,
+      "seconds": 0.7338976749997528,
+      "bytes": 1218142,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.4328443340000376,
-      "bytes": 11483713,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.1807161469996572,
+      "bytes": 7525394,
+      "pages": 28,
+      "shards": 3
     },
     {
-      "seconds": 0.909485917999973,
-      "bytes": 1076059,
-      "pages": 32,
-      "shards": 1
+      "seconds": 1.1362227980007447,
+      "bytes": 11027017,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.961313237000013,
-      "bytes": 6243214,
+      "seconds": 0.8288988310005152,
+      "bytes": 5801101,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.4556228910000755,
-      "bytes": 7655201,
-      "pages": 35,
+      "seconds": 1.2185628970000835,
+      "bytes": 7809668,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1608069779999823,
-      "bytes": 11166576,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.2807725910006411,
+      "bytes": 1082933,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.0247551580000618,
-      "bytes": 7489968,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.2944532310002614,
+      "bytes": 11460253,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 0.7696208980000847,
-      "bytes": 1062863,
+      "seconds": 1.090632098000242,
+      "bytes": 1122627,
       "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.0704197729999123,
-      "bytes": 6206676,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.176693084999897,
+      "bytes": 12183576,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 2.8963105739999264,
-      "bytes": 7237253,
-      "pages": 32,
+      "seconds": 1.0236794259999442,
+      "bytes": 7487764,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.3083305300001484,
-      "bytes": 5580567,
-      "pages": 35,
+      "seconds": 1.1006291130006502,
+      "bytes": 6137509,
+      "pages": 39,
       "shards": 2
     },
     {
-      "seconds": 1.1845111140000881,
-      "bytes": 11699180,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.0797216399996614,
+      "bytes": 5987959,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.859392975999981,
-      "bytes": 2445063,
+      "seconds": 1.322158891000072,
+      "bytes": 2479484,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0558128229999966,
-      "bytes": 6261983,
-      "pages": 34,
+      "seconds": 1.3229940370001714,
+      "bytes": 6465970,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3157395409998571,
-      "bytes": 12201352,
-      "pages": 34,
+      "seconds": 1.3629516780001723,
+      "bytes": 11728657,
+      "pages": 30,
       "shards": 4
     },
     {
-      "seconds": 2.8152504160002536,
-      "bytes": 1180754,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.096408940999936,
+      "bytes": 6091544,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.0497820410000713,
-      "bytes": 5934890,
-      "pages": 35,
+      "seconds": 1.042788374000338,
+      "bytes": 6163555,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.250543007000033,
-      "bytes": 8195692,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.04696464799963,
+      "bytes": 2343472,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.6225744519997534,
-      "bytes": 9437094,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.3791698659997564,
+      "bytes": 10729274,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.1101808009998422,
-      "bytes": 6059135,
-      "pages": 35,
+      "seconds": 1.2665546050002376,
+      "bytes": 6293373,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0066997960002482,
-      "bytes": 2574525,
+      "seconds": 1.4158054800000173,
+      "bytes": 8082619,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.100743914999839,
+      "bytes": 6375983,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 3.0401188460000412,
-      "bytes": 10860861,
+      "seconds": 1.1518813020011294,
+      "bytes": 6877765,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.9114932880002016,
-      "bytes": 506194,
-      "pages": 16,
+      "seconds": 1.007327716999498,
+      "bytes": 296016,
+      "pages": 9,
       "shards": 1
     },
     {
-      "seconds": 1.1319729980000375,
-      "bytes": 7544705,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.1130705590003345,
+      "bytes": 6042311,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.008947303000241,
-      "bytes": 7015503,
-      "pages": 38,
-      "shards": 2
+      "seconds": 0.9439528090006206,
+      "bytes": 7900957,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.0751419640000677,
-      "bytes": 5874473,
-      "pages": 27,
-      "shards": 2
+      "seconds": 0.7852815879996342,
+      "bytes": 838334,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.191119417000209,
-      "bytes": 6703440,
-      "pages": 36,
-      "shards": 3
+      "seconds": 4.030324741001095,
+      "bytes": 15386927,
+      "pages": 28,
+      "shards": 4
     },
     {
-      "seconds": 1.0874789810000038,
-      "bytes": 6434937,
-      "pages": 34,
-      "shards": 3
+      "seconds": 0.845626510999864,
+      "bytes": 2613623,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.6576894219997484,
-      "bytes": 7512513,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.1139741530005267,
+      "bytes": 5909960,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 0.866141241999685,
-      "bytes": 1214946,
-      "pages": 34,
+      "seconds": 1.2427591689993278,
+      "bytes": 910923,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.2982788010003787,
-      "bytes": 16654234,
-      "pages": 34,
+      "seconds": 1.2945012599993788,
+      "bytes": 12063808,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 0.9026169319995461,
-      "bytes": 2322454,
-      "pages": 27,
+      "seconds": 0.8872203939990868,
+      "bytes": 6084404,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0145137629997407,
-      "bytes": 6320844,
-      "pages": 35,
+      "seconds": 1.511142709001433,
+      "bytes": 6655300,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.26741677300015,
-      "bytes": 7410356,
+      "seconds": 1.0988184660000115,
+      "bytes": 7486365,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.306584923000628,
-      "bytes": 936913,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.2685245129996474,
+      "bytes": 10318034,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 33.908926936999706,
-      "bytes": 9910795,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.292125334000957,
+      "bytes": 1081552,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 2.0074904029997924,
-      "bytes": 13527574,
-      "pages": 44,
-      "shards": 4
+      "seconds": 1.197562700999697,
+      "bytes": 7759288,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.034237903999383,
-      "bytes": 890095,
-      "pages": 25,
+      "seconds": 0.8769548000000214,
+      "bytes": 1135557,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.0843116149999332,
-      "bytes": 11777166,
-      "pages": 39,
+      "seconds": 3.838755348999257,
+      "bytes": 11603941,
+      "pages": 24,
       "shards": 4
     },
     {
-      "seconds": 1.0721420089994353,
-      "bytes": 6211390,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.57969898599913,
+      "bytes": 12094782,
+      "pages": 45,
+      "shards": 3
     },
     {
-      "seconds": 0.7338976749997528,
-      "bytes": 1218142,
+      "seconds": 0.8754016399998363,
+      "bytes": 1113722,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1807161469996572,
-      "bytes": 7525394,
-      "pages": 28,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1362227980007447,
-      "bytes": 11027017,
-      "pages": 35,
+      "seconds": 4.401627224999174,
+      "bytes": 7622267,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 0.8288988310005152,
-      "bytes": 5801101,
-      "pages": 35,
+      "seconds": 0.9423913660011749,
+      "bytes": 5379592,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2185628970000835,
-      "bytes": 7809668,
+      "seconds": 1.2426582910011348,
+      "bytes": 7714336,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.2807725910006411,
-      "bytes": 1082933,
-      "pages": 33,
+      "seconds": 1.0709264350007288,
+      "bytes": 727110,
+      "pages": 23,
       "shards": 1
     },
     {
-      "seconds": 1.2944532310002614,
-      "bytes": 11460253,
+      "seconds": 1.4975259919992823,
+      "bytes": 12239566,
+      "pages": 44,
+      "shards": 3
+    },
+    {
+      "seconds": 1.4755159030009963,
+      "bytes": 11647558,
       "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 1.090632098000242,
-      "bytes": 1122627,
+      "seconds": 0.8895111490001,
+      "bytes": 1197830,
       "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.176693084999897,
-      "bytes": 12183576,
-      "pages": 36,
+      "seconds": 1.169628391000515,
+      "bytes": 7496810,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0236794259999442,
-      "bytes": 7487764,
-      "pages": 29,
+      "seconds": 1.5263510319982743,
+      "bytes": 5728692,
+      "pages": 27,
       "shards": 3
     },
     {
-      "seconds": 1.1006291130006502,
-      "bytes": 6137509,
-      "pages": 39,
+      "seconds": 1.2046007990011276,
+      "bytes": 5154137,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.0797216399996614,
-      "bytes": 5987959,
-      "pages": 27,
-      "shards": 2
+      "seconds": 2.8337960279986874,
+      "bytes": 11685666,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.322158891000072,
-      "bytes": 2479484,
+      "seconds": 1.282873722999284,
+      "bytes": 2666361,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3229940370001714,
-      "bytes": 6465970,
-      "pages": 36,
+      "seconds": 1.1408444070002588,
+      "bytes": 5962047,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3629516780001723,
-      "bytes": 11728657,
-      "pages": 30,
+      "seconds": 1.150664095999673,
+      "bytes": 11982825,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.096408940999936,
-      "bytes": 6091544,
+      "seconds": 0.9085387359991728,
+      "bytes": 1084848,
+      "pages": 33,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2088771979997546,
+      "bytes": 6082441,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.042788374000338,
-      "bytes": 6163555,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0392706679995172,
+      "bytes": 8238622,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.04696464799963,
-      "bytes": 2343472,
-      "pages": 28,
+      "seconds": 1.0565556260007725,
+      "bytes": 5902213,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.3791698659997564,
-      "bytes": 10729274,
-      "pages": 36,
+      "seconds": 1.6428124459998799,
+      "bytes": 11197872,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 1.2665546050002376,
-      "bytes": 6293373,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4158054800000173,
-      "bytes": 8082619,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.1795239030016091,
+      "bytes": 829951,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.100743914999839,
-      "bytes": 6375983,
+      "seconds": 1.183389565001562,
+      "bytes": 6316614,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1518813020011294,
-      "bytes": 6877765,
-      "pages": 35,
+      "seconds": 1.805858101000922,
+      "bytes": 7507792,
+      "pages": 34,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:727525a389ec5f45c0e2113ae9043f1f725bf96b0fc45fa3f07a6899b9d03b8b
-size 721770815
+oid sha256:e91afb32a91ba214e7473f46d6f99a0a931e405920d457559aea60d50123c7e4
+size 721772479