Qwen90M step 210000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 205000,
+  "step": 210000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "81479c438b659ae76d14a43cced464395f2a4a922f5a7d702440d60c97a8d05d",
+    "model.safetensors": "921781da103eac594d0d849dee7135488c4fb169435d08bae6486e287aaf16c7",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "7e9a9266ee128ed17a9d65fb5caa62965e42f185628b8d60fc5b3a810de04245",
+    "trainer.pt": "1a7597fc28402f093934d8c6ac10a76d19bc865c277e196a6a417b18ff76bfba",
     "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf",
     "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522",
-    "shard_scheduler.json": "6b78eed1b3798fed0667fc8eba1b21df9d8e0ade64a2f028654262590a5a3502",
-    "coverage.json": "cca4a03276b249b32504e7328f2319c84bec1a8153f03dd9d2c1e39fa1f9e3a6",
+    "shard_scheduler.json": "12ca185f760b35fc0306bf401e7f24f0c1ef6c13f220f2c3f65d1588dbf74548",
+    "coverage.json": "0784ef6420a747808e25c92ea80f2676691796080a1e53c62ef8213031463920",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 205000,
+  "step": 210000,
   "stats": {
-    "climbmix/blocks": 7380000,
-    "climbmix/targets": 12335030804,
-    "climbmix/input_tokens": 12335812509,
-    "ultra_qa/blocks": 2767500,
-    "ultra_qa/targets": 4515170109,
-    "ultra_qa/input_tokens": 4515178820,
-    "cortex/blocks": 2767500,
-    "cortex/targets": 105260709,
-    "cortex/input_tokens": 4399023648,
-    "ultra_style/blocks": 2767500,
-    "ultra_style/targets": 4377672891,
-    "ultra_style/input_tokens": 4377902456,
-    "openmath/blocks": 1845000,
-    "openmath/targets": 2850633588,
-    "openmath/input_tokens": 2851583832,
-    "rewrite6/blocks": 922500,
-    "rewrite6/targets": 1512304899,
-    "rewrite6/input_tokens": 1513963391
+    "climbmix/blocks": 7560000,
+    "climbmix/targets": 12635674730,
+    "climbmix/input_tokens": 12636475801,
+    "ultra_qa/blocks": 2835000,
+    "ultra_qa/targets": 4625316379,
+    "ultra_qa/input_tokens": 4625325318,
+    "cortex/blocks": 2835000,
+    "cortex/targets": 107843015,
+    "cortex/input_tokens": 4506362531,
+    "ultra_style/blocks": 2835000,
+    "ultra_style/targets": 4484394327,
+    "ultra_style/input_tokens": 4484629586,
+    "openmath/blocks": 1890000,
+    "openmath/targets": 2920105808,
+    "openmath/input_tokens": 2921079037,
+    "rewrite6/blocks": 945000,
+    "rewrite6/targets": 1549174599,
+    "rewrite6/input_tokens": 1550873200
   },
   "sources": {
     "climbmix": {
-      "page": 1011,
-      "row": 415,
+      "page": 1035,
+      "row": 3928,
       "start": {
-        "file": 48,
+        "file": 50,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 74,
+          "row_group": 4,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4141471
+      "documents": 4243288
     },
     "rewrite6": {
-      "page": 25,
-      "row": 802,
+      "page": 39,
+      "row": 727,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 102400
+          "row": 159744
         }
       },
       "epoch": 11,
       "offset": 0,
-      "documents": 2362646
+      "documents": 2419915
     },
     "openmath": {
-      "page": 631,
-      "row": 3535,
+      "page": 647,
+      "row": 820,
       "start": {
-        "file": 5,
+        "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 401,
-          "row": 326
+          "row_group": 30,
+          "row": 212
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 2588111
+      "documents": 2650932
     },
     "ultra_qa": {
-      "page": 273,
-      "row": 561,
+      "page": 279,
+      "row": 3264,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 14394
+          "row": 38970
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1118769
+      "documents": 1146048
     },
     "ultra_style": {
-      "page": 416,
-      "row": 3531,
+      "page": 427,
+      "row": 231,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 78532
+          "row_group": 4,
+          "row": 38971
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 1707467
+      "offset": 2048,
+      "documents": 1749223
     },
     "cortex": {
-      "page": 53673,
-      "row": 65,
+      "page": 54982,
+      "row": 96,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 6870209
+      "documents": 7037792
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 205000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "81479c438b659ae76d14a43cced464395f2a4a922f5a7d702440d60c97a8d05d", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "7e9a9266ee128ed17a9d65fb5caa62965e42f185628b8d60fc5b3a810de04245", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522", "shard_scheduler.json": "6b78eed1b3798fed0667fc8eba1b21df9d8e0ade64a2f028654262590a5a3502", "coverage.json": "cca4a03276b249b32504e7328f2319c84bec1a8153f03dd9d2c1e39fa1f9e3a6", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000205000"}
+{"step": 210000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "921781da103eac594d0d849dee7135488c4fb169435d08bae6486e287aaf16c7", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "1a7597fc28402f093934d8c6ac10a76d19bc865c277e196a6a417b18ff76bfba", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522", "shard_scheduler.json": "12ca185f760b35fc0306bf401e7f24f0c1ef6c13f220f2c3f65d1588dbf74548", "coverage.json": "0784ef6420a747808e25c92ea80f2676691796080a1e53c62ef8213031463920", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000210000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:81479c438b659ae76d14a43cced464395f2a4a922f5a7d702440d60c97a8d05d
+oid sha256:921781da103eac594d0d849dee7135488c4fb169435d08bae6486e287aaf16c7
 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": 74801.05554614915,
-      "required_pages_per_second_upper_estimate": 0.006483565913198165
+      "required_token_positions_per_second": 74616.38181288075,
+      "required_pages_per_second_upper_estimate": 0.006467558862049758
     },
     {
       "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": 9350.131943268643,
-      "required_pages_per_second_upper_estimate": 0.0037299074291003045
+      "required_token_positions_per_second": 9327.047726610093,
+      "required_pages_per_second_upper_estimate": 0.0037206987899354132
     },
     {
       "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": 18700.263886537286,
-      "required_pages_per_second_upper_estimate": 0.0043259769367466165
+      "required_token_positions_per_second": 18654.095453220187,
+      "required_pages_per_second_upper_estimate": 0.004315296682235381
     },
     {
       "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": 28050.39582980593,
-      "required_pages_per_second_upper_estimate": 0.0017081219369820858
+      "required_token_positions_per_second": 27981.14317983028,
+      "required_pages_per_second_upper_estimate": 0.0017039048139391429
     },
     {
       "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": 28050.39582980593,
-      "required_pages_per_second_upper_estimate": 0.00215939711944384
+      "required_token_positions_per_second": 27981.14317983028,
+      "required_pages_per_second_upper_estimate": 0.00215406585874511
     },
     {
       "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": 21,
-      "required_token_positions_per_second": 28050.39582980593,
-      "required_pages_per_second_upper_estimate": 0.3482920375703829
+      "required_token_positions_per_second": 27981.14317983028,
+      "required_pages_per_second_upper_estimate": 0.3474321514313953
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,373 +158,409 @@
     "prefetch_batches": 61,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.18545458216408406,
-    "producer_p95_seconds": 0.21071651100101008,
-    "gpu_update_seconds": 0.9856545400554259,
-    "producer_batches_per_second": 5.392155795402419,
-    "gpu_batches_per_second": 1.0145542473164761,
-    "producer_headroom": 5.31480284042457,
+    "producer_batch_seconds": 0.11913276657806193,
+    "producer_p95_seconds": 0.32301659399854543,
+    "gpu_update_seconds": 0.9880940111099384,
+    "producer_batches_per_second": 8.393996284345066,
+    "gpu_batches_per_second": 1.0120494495019632,
+    "producer_headroom": 8.294057457840436,
     "sustainable": true,
-    "buffer_seconds": 60.12492694338098,
+    "buffer_seconds": 60.273734677706244,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 86,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 359053.68233167636,
-    "upload_bytes_per_second": 5324605.411232466,
+    "generation_bytes_per_second": 355577.373587176,
+    "upload_bytes_per_second": 5217106.874362177,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 77.04456188643186,
+    "estimated_full_shards_per_hour": 76.29862695418795,
     "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": 1.3317575169999145,
-      "bytes": 24816,
-      "pages": 1,
-      "shards": 1
+      "seconds": 0.9532060609999462,
+      "bytes": 7422959,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.540608198999962,
-      "bytes": 9815303,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.2087593340002059,
+      "bytes": 6148456,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.441364390000217,
-      "bytes": 13613764,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.2602108099999896,
+      "bytes": 2561145,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.8236200929995903,
-      "bytes": 1218836,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.5218778089993066,
+      "bytes": 10001472,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.1421018979999644,
-      "bytes": 6132811,
-      "pages": 34,
+      "seconds": 1.655338631999257,
+      "bytes": 6858339,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2374535429999014,
-      "bytes": 7166128,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.3612198569990142,
+      "bytes": 7511604,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.0863462409997737,
-      "bytes": 6279325,
+      "seconds": 1.111224463000326,
+      "bytes": 5878363,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9579987879997134,
-      "bytes": 7302822,
-      "pages": 28,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9735736169996017,
-      "bytes": 6894198,
-      "pages": 36,
+      "seconds": 1.1176849410003342,
+      "bytes": 5784032,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.282220471999608,
-      "bytes": 9425496,
+      "seconds": 1.2254727769995952,
+      "bytes": 7561832,
       "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.236358336000194,
-      "bytes": 7782594,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1802483909996226,
+      "bytes": 1055971,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.0319873409998763,
-      "bytes": 964782,
-      "pages": 28,
-      "shards": 1
+      "seconds": 1.59867495799881,
+      "bytes": 13042114,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.3289059700000507,
-      "bytes": 6144064,
-      "pages": 29,
+      "seconds": 1.1432860509994498,
+      "bytes": 5906480,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2417195279995212,
+      "bytes": 5623378,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.1848053769999751,
-      "bytes": 6754320,
+      "seconds": 0.9634915199985699,
+      "bytes": 7597611,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.6214569329995356,
-      "bytes": 11854578,
+      "seconds": 0.761399758999687,
+      "bytes": 1061264,
       "pages": 32,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 0.9589299640001627,
-      "bytes": 2653813,
-      "pages": 36,
+      "seconds": 1.0512442660001398,
+      "bytes": 5862459,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.7666170649999913,
-      "bytes": 10799529,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.2080343199995696,
+      "bytes": 17359027,
+      "pages": 39,
+      "shards": 5
     },
     {
-      "seconds": 0.9170148619996326,
-      "bytes": 257577,
-      "pages": 8,
+      "seconds": 1.0087294929999189,
+      "bytes": 1073260,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.4457078639998144,
-      "bytes": 7654224,
-      "pages": 37,
+      "seconds": 1.4433996169991588,
+      "bytes": 7239065,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.9410484550007823,
-      "bytes": 1134403,
-      "pages": 34,
+      "seconds": 0.894821537000098,
+      "bytes": 1079304,
+      "pages": 31,
       "shards": 1
     },
     {
-      "seconds": 1.2444776140000613,
-      "bytes": 6243354,
-      "pages": 26,
+      "seconds": 1.1567702649990679,
+      "bytes": 10579644,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 1.6480453410003975,
+      "bytes": 7697533,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2853153289997863,
+      "bytes": 6668992,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.5528372400003718,
-      "bytes": 12775308,
+      "seconds": 1.2017098229989642,
+      "bytes": 7521074,
       "pages": 38,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.527598386000136,
-      "bytes": 5658593,
+      "seconds": 1.335906720998537,
+      "bytes": 5362985,
       "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.7142034060007063,
-      "bytes": 6133193,
-      "pages": 36,
+      "seconds": 1.582250990999455,
+      "bytes": 5900156,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.0039940289998412,
-      "bytes": 2541534,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2782940739998594,
+      "bytes": 7640268,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 0.723643687000731,
+      "bytes": 1209843,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.7366310759998669,
-      "bytes": 10198556,
+      "seconds": 1.586552576000031,
+      "bytes": 9572867,
       "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.7939092509996044,
-      "bytes": 1225371,
-      "pages": 32,
-      "shards": 1
+      "seconds": 1.2363076699984958,
+      "bytes": 8177813,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.2473491540004034,
-      "bytes": 13161042,
+      "seconds": 1.3350522609998734,
+      "bytes": 6097996,
       "pages": 35,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 1.0770104249995711,
-      "bytes": 6141587,
-      "pages": 36,
+      "seconds": 0.9499315409993869,
+      "bytes": 5779343,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1525904230002197,
-      "bytes": 7408784,
-      "pages": 38,
+      "seconds": 1.612262433000069,
+      "bytes": 7066450,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.0340140519992929,
-      "bytes": 5739884,
-      "pages": 26,
+      "seconds": 1.120285896000496,
+      "bytes": 6092850,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9941408290005711,
-      "bytes": 1118169,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.532518773999982,
+      "bytes": 2518854,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.4089016780008023,
-      "bytes": 7539453,
-      "pages": 38,
+      "seconds": 1.2675222210000356,
+      "bytes": 11869039,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.256968259000132,
-      "bytes": 11567319,
-      "pages": 33,
+      "seconds": 1.6028917950006871,
+      "bytes": 10461677,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1466406099998494,
-      "bytes": 5899615,
-      "pages": 35,
+      "seconds": 1.3869329340013792,
+      "bytes": 2462257,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9532060609999462,
-      "bytes": 7422959,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.3128174869998475,
+      "bytes": 5972623,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.2087593340002059,
-      "bytes": 6148456,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.8817397439997876,
+      "bytes": 1063209,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.2602108099999896,
-      "bytes": 2561145,
+      "seconds": 1.6997340180005267,
+      "bytes": 11573294,
+      "pages": 33,
+      "shards": 4
+    },
+    {
+      "seconds": 1.4719399390014587,
+      "bytes": 11935199,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.5218778089993066,
-      "bytes": 10001472,
-      "pages": 31,
-      "shards": 3
+      "seconds": 0.9283750220001821,
+      "bytes": 2657482,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.655338631999257,
-      "bytes": 6858339,
-      "pages": 37,
+      "seconds": 0.9979908650002471,
+      "bytes": 5962542,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.3612198569990142,
-      "bytes": 7511604,
-      "pages": 29,
+      "seconds": 1.4787909560000116,
+      "bytes": 11055954,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.111224463000326,
-      "bytes": 5878363,
+      "seconds": 1.0256725109993567,
+      "bytes": 2318063,
+      "pages": 28,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9941544879984576,
+      "bytes": 6059545,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1176849410003342,
-      "bytes": 5784032,
-      "pages": 33,
+      "seconds": 1.1885242320004181,
+      "bytes": 7012163,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2254727769995952,
-      "bytes": 7561832,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.3779209780004749,
+      "bytes": 11061746,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.1802483909996226,
-      "bytes": 1055971,
-      "pages": 33,
+      "seconds": 0.9865021240002534,
+      "bytes": 862370,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.59867495799881,
-      "bytes": 13042114,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.1108503920004296,
+      "bytes": 7694144,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.1432860509994498,
-      "bytes": 5906480,
-      "pages": 31,
+      "seconds": 2.1076598479994573,
+      "bytes": 6179293,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2417195279995212,
-      "bytes": 5623378,
-      "pages": 33,
+      "seconds": 1.0649322700010089,
+      "bytes": 5823579,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 0.9634915199985699,
-      "bytes": 7597611,
-      "pages": 38,
+      "seconds": 1.1941432809999242,
+      "bytes": 7137236,
+      "pages": 22,
       "shards": 3
     },
     {
-      "seconds": 0.761399758999687,
-      "bytes": 1061264,
-      "pages": 32,
-      "shards": 1
+      "seconds": 1.292083440001079,
+      "bytes": 7124472,
+      "pages": 44,
+      "shards": 2
     },
     {
-      "seconds": 1.0512442660001398,
-      "bytes": 5862459,
-      "pages": 28,
+      "seconds": 1.2287268149993906,
+      "bytes": 6045685,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2080343199995696,
-      "bytes": 17359027,
-      "pages": 39,
-      "shards": 5
+      "seconds": 1.297586668999429,
+      "bytes": 7207958,
+      "pages": 28,
+      "shards": 3
     },
     {
-      "seconds": 1.0087294929999189,
-      "bytes": 1073260,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.0006021399985912,
+      "bytes": 4975022,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.4433996169991588,
-      "bytes": 7239065,
+      "seconds": 1.1054393590002292,
+      "bytes": 5686965,
       "pages": 29,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.894821537000098,
-      "bytes": 1079304,
-      "pages": 31,
-      "shards": 1
+      "seconds": 0.952104601999963,
+      "bytes": 2531508,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.1567702649990679,
-      "bytes": 10579644,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.6972403340005258,
+      "bytes": 16066476,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.6480453410003975,
-      "bytes": 7697533,
-      "pages": 38,
+      "seconds": 1.3755702170001314,
+      "bytes": 7451679,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.2853153289997863,
-      "bytes": 6668992,
-      "pages": 31,
+      "seconds": 1.598954668999795,
+      "bytes": 1262837,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0224317270003667,
+      "bytes": 6094784,
+      "pages": 35,
       "shards": 2
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7e9a9266ee128ed17a9d65fb5caa62965e42f185628b8d60fc5b3a810de04245
-size 721767807
+oid sha256:1a7597fc28402f093934d8c6ac10a76d19bc865c277e196a6a417b18ff76bfba
+size 721768575