Qwen90M step 150000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 145000,
+  "step": 150000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "f82ef0b41a9f5a444a6d4c213cf42228dda7b7856086639bd309c8f1264486c9",
+    "model.safetensors": "cf66ab2f55cbc1fb6e95bfaef3a16d5c7c8084d9d77f534a346ea64534d3636b",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "81681141f6dfe127e198e16da88d8c38af13ada9c9f0a4963e0cd992eb0b365c",
+    "trainer.pt": "f838656d4de1ef990d79c3418bb1cd36ab78cb5ecc62a6d1f5a9fbe1d6a5d857",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "10fa55d76d1a4ec4f9132ef494d950eba8aa9e6cecd5fa1ceb70ba6de35537e4",
-    "shard_scheduler.json": "33c9db18de388458070d6df886e5bda2cedf9dabd3e716f543d2e756f18de9cc",
-    "coverage.json": "088c76466a16e8303fb88ede838eb2896c816dbf26f02d114c54169dacd144e5",
+    "shard_scheduler.json": "34c9075a0f66a3d08c183399d506bb22fcf18c184ad147564967e4708a38ba3f",
+    "coverage.json": "a2f785366449384f55526192ff2e3551a8f132be1f93c9467f8d4fa84286825c",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 145000,
+  "step": 150000,
   "stats": {
-    "climbmix/blocks": 5220000,
-    "climbmix/targets": 8723475320,
-    "climbmix/input_tokens": 8724028649,
-    "ultra_qa/blocks": 1957500,
-    "ultra_qa/targets": 3193055973,
-    "ultra_qa/input_tokens": 3193062132,
-    "cortex/blocks": 1957500,
-    "cortex/targets": 74442843,
-    "cortex/input_tokens": 3111300467,
-    "ultra_style/blocks": 1957500,
-    "ultra_style/targets": 3096311658,
-    "ultra_style/input_tokens": 3096473872,
-    "openmath/blocks": 1305000,
-    "openmath/targets": 2016316779,
-    "openmath/input_tokens": 2016988997,
-    "rewrite6/blocks": 652500,
-    "rewrite6/targets": 1069684284,
-    "rewrite6/input_tokens": 1070857144
+    "climbmix/blocks": 5400000,
+    "climbmix/targets": 9024151403,
+    "climbmix/input_tokens": 9024723849,
+    "ultra_qa/blocks": 2025000,
+    "ultra_qa/targets": 3303216837,
+    "ultra_qa/input_tokens": 3303223204,
+    "cortex/blocks": 2025000,
+    "cortex/targets": 77013811,
+    "cortex/input_tokens": 3218659538,
+    "ultra_style/blocks": 2025000,
+    "ultra_style/targets": 3203109017,
+    "ultra_style/input_tokens": 3203277056,
+    "openmath/blocks": 1350000,
+    "openmath/targets": 2085884301,
+    "openmath/input_tokens": 2086579643,
+    "rewrite6/blocks": 675000,
+    "rewrite6/targets": 1106511308,
+    "rewrite6/input_tokens": 1107724623
   },
   "sources": {
     "climbmix": {
-      "page": 715,
-      "row": 2681,
+      "page": 740,
+      "row": 1687,
       "start": {
-        "file": 34,
+        "file": 35,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 44,
+          "row_group": 60,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 2931321
+      "offset": 2048,
+      "documents": 3032727
     },
     "rewrite6": {
-      "page": 6,
-      "row": 3005,
+      "page": 20,
+      "row": 3357,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 24576
+          "row": 81920
         }
       },
       "epoch": 8,
-      "offset": 2048,
-      "documents": 1670813
+      "offset": 0,
+      "documents": 1728509
     },
     "openmath": {
-      "page": 446,
-      "row": 3868,
+      "page": 462,
+      "row": 1399,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 80,
-          "row": 216
+          "row_group": 145,
+          "row": 752
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 1830684
+      "offset": 2048,
+      "documents": 1893751
     },
     "ultra_qa": {
-      "page": 193,
-      "row": 848,
+      "page": 199,
+      "row": 3551,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 12512
+          "row": 37088
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 791376
+      "offset": 2048,
+      "documents": 818655
     },
     "ultra_style": {
-      "page": 294,
-      "row": 3383,
+      "page": 305,
+      "row": 256,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 9157
+          "row": 54213
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1207607
+      "documents": 1249536
     },
     "cortex": {
-      "page": 37962,
-      "row": 8,
+      "page": 39271,
+      "row": 122,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 4859144
+      "documents": 5026810
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 145000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "f82ef0b41a9f5a444a6d4c213cf42228dda7b7856086639bd309c8f1264486c9", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "81681141f6dfe127e198e16da88d8c38af13ada9c9f0a4963e0cd992eb0b365c", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "10fa55d76d1a4ec4f9132ef494d950eba8aa9e6cecd5fa1ceb70ba6de35537e4", "shard_scheduler.json": "33c9db18de388458070d6df886e5bda2cedf9dabd3e716f543d2e756f18de9cc", "coverage.json": "088c76466a16e8303fb88ede838eb2896c816dbf26f02d114c54169dacd144e5", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000145000"}
+{"step": 150000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "cf66ab2f55cbc1fb6e95bfaef3a16d5c7c8084d9d77f534a346ea64534d3636b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "f838656d4de1ef990d79c3418bb1cd36ab78cb5ecc62a6d1f5a9fbe1d6a5d857", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "10fa55d76d1a4ec4f9132ef494d950eba8aa9e6cecd5fa1ceb70ba6de35537e4", "shard_scheduler.json": "34c9075a0f66a3d08c183399d506bb22fcf18c184ad147564967e4708a38ba3f", "coverage.json": "a2f785366449384f55526192ff2e3551a8f132be1f93c9467f8d4fa84286825c", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000150000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f82ef0b41a9f5a444a6d4c213cf42228dda7b7856086639bd309c8f1264486c9
+oid sha256:cf66ab2f55cbc1fb6e95bfaef3a16d5c7c8084d9d77f534a346ea64534d3636b
 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": 75819.42597709605,
-      "required_pages_per_second_upper_estimate": 0.006571835681116382
+      "required_token_positions_per_second": 75368.18777639217,
+      "required_pages_per_second_upper_estimate": 0.006532723497532133
     },
     {
       "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": 9477.428247137006,
-      "required_pages_per_second_upper_estimate": 0.003780687827962744
+      "required_token_positions_per_second": 9421.02347204902,
+      "required_pages_per_second_upper_estimate": 0.0037581871198536063
     },
     {
       "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": 18954.856494274012,
-      "required_pages_per_second_upper_estimate": 0.0043848724558696464
+      "required_token_positions_per_second": 18842.04694409804,
+      "required_pages_per_second_upper_estimate": 0.004358775951816709
     },
     {
       "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": 28432.28474141102,
-      "required_pages_per_second_upper_estimate": 0.0017313769680825642
+      "required_token_positions_per_second": 28263.070416147064,
+      "required_pages_per_second_upper_estimate": 0.0017210727034729413
     },
     {
       "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": 28432.28474141102,
-      "required_pages_per_second_upper_estimate": 0.002188795984995367
+      "required_token_positions_per_second": 28263.070416147064,
+      "required_pages_per_second_upper_estimate": 0.0021757693978213155
     },
     {
       "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": 28432.28474141102,
-      "required_pages_per_second_upper_estimate": 0.35303381975254877
+      "required_token_positions_per_second": 28263.070416147064,
+      "required_pages_per_second_upper_estimate": 0.3509327441566865
     }
   ],
   "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": 688946,
-    "producer_batch_seconds": 0.22319741781251423,
-    "producer_p95_seconds": 0.37801911600035965,
-    "gpu_update_seconds": 0.9724156975584616,
-    "producer_batches_per_second": 4.4803385711209245,
-    "gpu_batches_per_second": 1.0283667802883036,
-    "producer_headroom": 4.3567515569346345,
+    "producer_batch_seconds": 0.1606457544066302,
+    "producer_p95_seconds": 0.3911977770039812,
+    "gpu_update_seconds": 0.9782376646595458,
+    "producer_batches_per_second": 6.2248766155922,
+    "gpu_batches_per_second": 1.0222464704914302,
+    "producer_headroom": 6.089408763230731,
     "sustainable": true,
-    "buffer_seconds": 60.28977324862462,
+    "buffer_seconds": 60.65073520889184,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 80,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 549008.0145004862,
-    "upload_bytes_per_second": 5070479.625359649,
+    "generation_bytes_per_second": 362750.0021895322,
+    "upload_bytes_per_second": 5470118.08714139,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 117.80433965931834,
+    "estimated_full_shards_per_hour": 77.83770608200525,
     "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": 0.8942368999996688,
-      "bytes": 850966,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.7747774790004769,
+      "bytes": 7374031,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.4387539480012492,
-      "bytes": 11588272,
-      "pages": 38,
-      "shards": 4
+      "seconds": 0.9477112800013856,
+      "bytes": 5792070,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.6636693889995513,
-      "bytes": 11864781,
-      "pages": 37,
+      "seconds": 1.0662493220006581,
+      "bytes": 5942385,
+      "pages": 30,
+      "shards": 2
+    },
+    {
+      "seconds": 1.245676413000183,
+      "bytes": 7693974,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1722596879990306,
-      "bytes": 1208158,
-      "pages": 35,
+      "seconds": 1.0202759780004271,
+      "bytes": 6615812,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2069239850025042,
-      "bytes": 7251816,
-      "pages": 27,
+      "seconds": 1.4830753700007335,
+      "bytes": 9974834,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.169119137000962,
-      "bytes": 6109087,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.272736505001376,
+      "bytes": 7611348,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.4141185940025025,
-      "bytes": 7704490,
-      "pages": 38,
+      "seconds": 0.8344998959983059,
+      "bytes": 1093697,
+      "pages": 33,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3662152600008994,
+      "bytes": 7153184,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.2197817560008843,
-      "bytes": 5784149,
-      "pages": 28,
+      "seconds": 1.6717624239972793,
+      "bytes": 5422811,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3641967180010397,
-      "bytes": 7009240,
+      "seconds": 1.1521110989997396,
+      "bytes": 12146138,
       "pages": 38,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.4704428500008362,
-      "bytes": 11028707,
+      "seconds": 1.1019259579989011,
+      "bytes": 7338055,
       "pages": 30,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.9246157949964982,
-      "bytes": 1204930,
-      "pages": 34,
+      "seconds": 0.7986074580003333,
+      "bytes": 1164850,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.0427682160006952,
-      "bytes": 5985007,
+      "seconds": 1.6704007100015588,
+      "bytes": 10030549,
+      "pages": 31,
+      "shards": 3
+    },
+    {
+      "seconds": 2.951283398000669,
+      "bytes": 2617533,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1606040469996515,
+      "bytes": 6236824,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1093607039983908,
-      "bytes": 7338185,
+      "seconds": 4.655135680000967,
+      "bytes": 7374926,
       "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9098767819996283,
-      "bytes": 5901703,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.311550709000585,
+      "bytes": 11722471,
+      "pages": 42,
+      "shards": 3
     },
     {
-      "seconds": 1.1044195929971465,
-      "bytes": 13283238,
-      "pages": 34,
-      "shards": 4
+      "seconds": 0.7739134580006066,
+      "bytes": 5816616,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.8937192140001571,
-      "bytes": 6207827,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.117633730002126,
+      "bytes": 7565346,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.9678468759993848,
-      "bytes": 1151038,
+      "seconds": 0.8082868439996673,
+      "bytes": 1195987,
       "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.3402203920013562,
-      "bytes": 12178587,
-      "pages": 33,
+      "seconds": 1.157742789000622,
+      "bytes": 11040923,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 0.8894950190006057,
-      "bytes": 1180133,
-      "pages": 34,
+      "seconds": 0.8454459890017461,
+      "bytes": 981466,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.1408973270008573,
-      "bytes": 5952533,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4523514239990618,
-      "bytes": 13500548,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.0536112129993853,
+      "bytes": 11943546,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.8281688100032625,
-      "bytes": 661491,
+      "seconds": 1.0575292020002962,
+      "bytes": 2082389,
       "pages": 19,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.42470605500057,
-      "bytes": 11031230,
-      "pages": 41,
+      "seconds": 1.061086557998351,
+      "bytes": 10157591,
+      "pages": 32,
       "shards": 4
     },
     {
-      "seconds": 0.8167821940005524,
-      "bytes": 1161839,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.351508649000607,
+      "bytes": 6405900,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.9240367780002998,
-      "bytes": 5199805,
-      "pages": 10,
+      "seconds": 0.9840362190006999,
+      "bytes": 2390157,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4070672229972843,
-      "bytes": 1010316,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.0104034229989338,
+      "bytes": 6117004,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.181387053002254,
-      "bytes": 11521167,
-      "pages": 38,
+      "seconds": 1.275026989998878,
+      "bytes": 16016482,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.134901017001539,
-      "bytes": 6770130,
-      "pages": 32,
+      "seconds": 0.8882673849984712,
+      "bytes": 2492218,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.1226096480022534,
-      "bytes": 7935280,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.982757900001161,
-      "bytes": 5911177,
-      "pages": 25,
+      "seconds": 1.2180680759993265,
+      "bytes": 6022567,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.134857662000286,
-      "bytes": 5917855,
-      "pages": 41,
+      "seconds": 1.038481719999254,
+      "bytes": 2380066,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5188524260011036,
-      "bytes": 7655478,
-      "pages": 39,
+      "seconds": 1.0937937710004917,
+      "bytes": 10154198,
+      "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 0.7887198950011225,
-      "bytes": 905384,
-      "pages": 25,
-      "shards": 1
-    },
-    {
-      "seconds": 4.007056702997943,
-      "bytes": 13411793,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.047995336000895,
+      "bytes": 5924124,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.3467935920016316,
-      "bytes": 10736953,
-      "pages": 38,
+      "seconds": 1.1032993179978803,
+      "bytes": 8345826,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.8973097269990831,
-      "bytes": 792959,
-      "pages": 26,
+      "seconds": 0.9737631519965362,
+      "bytes": 6188431,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2533608519988775,
-      "bytes": 7837535,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.3008259939997515,
+      "bytes": 12099101,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.1466966719999618,
-      "bytes": 6019789,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8470429570006672,
+      "bytes": 932725,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.0688578599983884,
-      "bytes": 4660632,
-      "pages": 31,
+      "seconds": 1.1173031580001407,
+      "bytes": 6247577,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4090757320009288,
-      "bytes": 13354980,
+      "seconds": 0.892657821001194,
+      "bytes": 2615407,
       "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2370568289989023,
+      "bytes": 16078869,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.0085502450019703,
-      "bytes": 1017261,
-      "pages": 26,
+      "seconds": 0.8470632479984488,
+      "bytes": 947755,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.7747774790004769,
-      "bytes": 7374031,
-      "pages": 33,
+      "seconds": 1.2166951500003051,
+      "bytes": 7600114,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9477112800013856,
-      "bytes": 5792070,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0662493220006581,
-      "bytes": 5942385,
+      "seconds": 1.056514903997595,
+      "bytes": 5902748,
       "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.245676413000183,
-      "bytes": 7693974,
-      "pages": 38,
+      "seconds": 1.0406464490006329,
+      "bytes": 6219293,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0202759780004271,
-      "bytes": 6615812,
-      "pages": 28,
+      "seconds": 1.1162217770033749,
+      "bytes": 6495200,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4830753700007335,
-      "bytes": 9974834,
-      "pages": 35,
+      "seconds": 1.1027901979978196,
+      "bytes": 11991791,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.272736505001376,
-      "bytes": 7611348,
-      "pages": 38,
+      "seconds": 1.5872819810028886,
+      "bytes": 6214149,
+      "pages": 24,
       "shards": 3
     },
     {
-      "seconds": 0.8344998959983059,
-      "bytes": 1093697,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.0351565999953891,
+      "bytes": 6216009,
+      "pages": 41,
+      "shards": 2
     },
     {
-      "seconds": 1.3662152600008994,
-      "bytes": 7153184,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.4164216759963892,
+      "bytes": 2450413,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.6717624239972793,
-      "bytes": 5422811,
-      "pages": 35,
+      "seconds": 1.0167613629964762,
+      "bytes": 6214181,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.1521110989997396,
-      "bytes": 12146138,
-      "pages": 38,
+      "seconds": 1.3737143940015812,
+      "bytes": 10463664,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1019259579989011,
-      "bytes": 7338055,
-      "pages": 30,
+      "seconds": 0.8019685460021719,
+      "bytes": 2186825,
+      "pages": 23,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2952469239971833,
+      "bytes": 12034549,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 0.7986074580003333,
-      "bytes": 1164850,
-      "pages": 35,
-      "shards": 1
+      "seconds": 0.8924442770003225,
+      "bytes": 6155396,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.6704007100015588,
-      "bytes": 10030549,
+      "seconds": 1.1965257189949625,
+      "bytes": 6621394,
       "pages": 31,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 2.951283398000669,
-      "bytes": 2617533,
-      "pages": 35,
+      "seconds": 1.1455287000062526,
+      "bytes": 6484150,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1606040469996515,
-      "bytes": 6236824,
-      "pages": 36,
+      "seconds": 0.8978742559993407,
+      "bytes": 2461928,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 4.655135680000967,
-      "bytes": 7374926,
-      "pages": 30,
-      "shards": 3
+      "seconds": 0.9092525360028958,
+      "bytes": 6015320,
+      "pages": 26,
+      "shards": 2
     },
     {
-      "seconds": 1.311550709000585,
-      "bytes": 11722471,
+      "seconds": 1.3533267960010562,
+      "bytes": 11703451,
       "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 0.7739134580006066,
-      "bytes": 5816616,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.117633730002126,
-      "bytes": 7565346,
-      "pages": 38,
+      "seconds": 1.2227107450016774,
+      "bytes": 7523214,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.8082868439996673,
-      "bytes": 1195987,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.157742789000622,
-      "bytes": 11040923,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.0604035779979313,
+      "bytes": 6119324,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8454459890017461,
-      "bytes": 981466,
-      "pages": 29,
+      "seconds": 1.0826924209977733,
+      "bytes": 902486,
+      "pages": 27,
       "shards": 1
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:81681141f6dfe127e198e16da88d8c38af13ada9c9f0a4963e0cd992eb0b365c
-size 721762943
+oid sha256:f838656d4de1ef990d79c3418bb1cd36ab78cb5ecc62a6d1f5a9fbe1d6a5d857
+size 721765055