Qwen90M step 540000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 535000,
+  "step": 540000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "746a63165b4986046476ae84c2713e36ed3fa23737687659143ada4b831f1654",
+    "model.safetensors": "b38e10d38ed7ada06841ff8982da7d676daf791bf507a6cf07a68d71c4a4078d",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "2b45969c25feb44bef2bbc28e1a234c7ba36a717f6375ea0dba3053855528f94",
+    "trainer.pt": "043647c8c26e8fed077bb197406f27a2662b7f2ed8742ee4065fa206a57818d0",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "3070ebbdf9b2b0fd6a0a237819043aa2c2a35c74a97edfa223e5c324471af4af",
-    "shard_scheduler.json": "dc3b4c9c9c5634b43786ba061a7ead1b9ed759c5b1eacff95382e91a9ef4f364",
-    "coverage.json": "7998ce90c7aacd5ffa1b3b45381b7a23cce87145599cd5217ab5d192996cc5db",
+    "shard_scheduler.json": "de12fe13ce1f77cd03863bd480ed4c92361aad4e582dff7ec3d7a242555cb0c8",
+    "coverage.json": "c6a7d6cc92b6319e9b013aeaaaf6b0e9a72597bf84b77a74dadf487e898d80d9",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,101 +1,107 @@
 {
-  "step": 535000,
+  "step": 540000,
   "stats": {
-    "climbmix/blocks": 19260000,
-    "climbmix/targets": 32195111082,
-    "climbmix/input_tokens": 32197151068,
-    "ultra_qa/blocks": 7222500,
-    "ultra_qa/targets": 11782590681,
-    "ultra_qa/input_tokens": 11782613431,
-    "cortex/blocks": 7222500,
-    "cortex/targets": 274657059,
-    "cortex/input_tokens": 11480802169,
-    "ultra_style/blocks": 7222500,
-    "ultra_style/targets": 11423290047,
-    "ultra_style/input_tokens": 11423888845,
-    "openmath/blocks": 4815000,
-    "openmath/targets": 7439991699,
-    "openmath/input_tokens": 7442470149,
-    "rewrite6/blocks": 2407500,
-    "rewrite6/targets": 3946780306,
-    "rewrite6/input_tokens": 3951107292
+    "climbmix/blocks": 19440000,
+    "climbmix/targets": 32495587833,
+    "climbmix/input_tokens": 32497646947,
+    "ultra_qa/blocks": 7290000,
+    "ultra_qa/targets": 11892698357,
+    "ultra_qa/input_tokens": 11892721312,
+    "cortex/blocks": 7290000,
+    "cortex/targets": 277235924,
+    "cortex/input_tokens": 11588137812,
+    "ultra_style/blocks": 7290000,
+    "ultra_style/targets": 11529724861,
+    "ultra_style/input_tokens": 11530329511,
+    "openmath/blocks": 4860000,
+    "openmath/targets": 7509510059,
+    "openmath/input_tokens": 7512011367,
+    "rewrite6/blocks": 2430000,
+    "rewrite6/targets": 3983637088,
+    "rewrite6/input_tokens": 3988004561
   },
   "sources": {
     "climbmix": {
-      "page": 2637,
-      "row": 4073,
+      "page": 2662,
+      "row": 3055,
       "start": {
-        "file": 127,
+        "file": 128,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 51,
+          "row_group": 68,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 10805225
+      "documents": 10906607
     },
     "rewrite6": {
-      "page": 0,
-      "row": 2348,
-      "start": null,
+      "page": 14,
+      "row": 2627,
+      "start": {
+        "file": 0,
+        "reader": {
+          "format": "jsonl-row-v1",
+          "row": 57344
+        }
+      },
       "epoch": 30,
       "offset": 2048,
-      "documents": 6164468
+      "documents": 6222091
     },
     "openmath": {
-      "page": 1648,
-      "row": 1424,
+      "page": 1663,
+      "row": 2767,
       "start": {
         "file": 15,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 200,
-          "row": 458
+          "row_group": 261,
+          "row": 898
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 6751632
+      "documents": 6814415
     },
     "ultra_qa": {
-      "page": 712,
-      "row": 3263,
+      "page": 719,
+      "row": 1872,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 60132
+          "row_group": 5,
+          "row": 24371
         }
       },
       "epoch": 0,
-      "offset": 6144,
-      "documents": 2919615
+      "offset": 2048,
+      "documents": 2946896
     },
     "ultra_style": {
-      "page": 1087,
-      "row": 3342,
+      "page": 1098,
+      "row": 616,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 2030
+          "row": 47086
         }
       },
       "epoch": 0,
       "offset": 4096,
-      "documents": 4455694
+      "documents": 4498024
     },
     "cortex": {
-      "page": 140074,
-      "row": 15,
+      "page": 141384,
+      "row": 37,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 17929487
+      "documents": 18097189
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 535000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "746a63165b4986046476ae84c2713e36ed3fa23737687659143ada4b831f1654", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "2b45969c25feb44bef2bbc28e1a234c7ba36a717f6375ea0dba3053855528f94", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "3070ebbdf9b2b0fd6a0a237819043aa2c2a35c74a97edfa223e5c324471af4af", "shard_scheduler.json": "dc3b4c9c9c5634b43786ba061a7ead1b9ed759c5b1eacff95382e91a9ef4f364", "coverage.json": "7998ce90c7aacd5ffa1b3b45381b7a23cce87145599cd5217ab5d192996cc5db", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000535000"}
+{"step": 540000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b38e10d38ed7ada06841ff8982da7d676daf791bf507a6cf07a68d71c4a4078d", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "043647c8c26e8fed077bb197406f27a2662b7f2ed8742ee4065fa206a57818d0", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "3070ebbdf9b2b0fd6a0a237819043aa2c2a35c74a97edfa223e5c324471af4af", "shard_scheduler.json": "de12fe13ce1f77cd03863bd480ed4c92361aad4e582dff7ec3d7a242555cb0c8", "coverage.json": "c6a7d6cc92b6319e9b013aeaaaf6b0e9a72597bf84b77a74dadf487e898d80d9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000540000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:746a63165b4986046476ae84c2713e36ed3fa23737687659143ada4b831f1654
+oid sha256:b38e10d38ed7ada06841ff8982da7d676daf791bf507a6cf07a68d71c4a4078d
 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": 76499.76095720792,
-      "required_pages_per_second_upper_estimate": 0.00529243096692033
+      "required_token_positions_per_second": 76448.66075247723,
+      "required_pages_per_second_upper_estimate": 0.005288895736188249
     },
     {
       "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": 9562.47011965099,
-      "required_pages_per_second_upper_estimate": 0.004003606552014596
+      "required_token_positions_per_second": 9556.082594059653,
+      "required_pages_per_second_upper_estimate": 0.004000932228436205
     },
     {
       "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": 19124.94023930198,
-      "required_pages_per_second_upper_estimate": 0.003983220693255846
+      "required_token_positions_per_second": 19112.165188119307,
+      "required_pages_per_second_upper_estimate": 0.00398055998699525
     },
     {
       "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": 28687.41035895297,
-      "required_pages_per_second_upper_estimate": 0.0017197664478798965
+      "required_token_positions_per_second": 28668.24778217896,
+      "required_pages_per_second_upper_estimate": 0.0017186176806618552
     },
     {
       "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": 28687.41035895297,
-      "required_pages_per_second_upper_estimate": 0.002230319972825811
+      "required_token_positions_per_second": 28668.24778217896,
+      "required_pages_per_second_upper_estimate": 0.0022288301667688993
     },
     {
       "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": 28687.41035895297,
-      "required_pages_per_second_upper_estimate": 0.36797129794324046
+      "required_token_positions_per_second": 28668.24778217896,
+      "required_pages_per_second_upper_estimate": 0.3677255009835554
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 704566,
-    "producer_batch_seconds": 0.08245735881254745,
-    "producer_p95_seconds": 0.19908109500102,
-    "gpu_update_seconds": 0.9637677174081841,
-    "producer_batches_per_second": 12.127480365619363,
-    "gpu_batches_per_second": 1.03759441402463,
-    "producer_headroom": 11.688074069885543,
+    "producer_batch_seconds": 0.3263542902969334,
+    "producer_p95_seconds": 0.2078722389996983,
+    "gpu_update_seconds": 0.9644119239539579,
+    "producer_batches_per_second": 3.064154600480815,
+    "gpu_batches_per_second": 1.0369013231401534,
+    "producer_headroom": 2.9551072335420745,
     "sustainable": true,
-    "buffer_seconds": 60.7173661967156,
+    "buffer_seconds": 60.757951209099346,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 87,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 322888.0851025507,
-    "upload_bytes_per_second": 4628498.063736584,
+    "generation_bytes_per_second": 393197.65477415506,
+    "upload_bytes_per_second": 5782840.946028825,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 69.28426661307708,
+    "estimated_full_shards_per_hour": 84.37106354158868,
     "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": 1.011093388000063,
-      "bytes": 1199201,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0240878049999083,
-      "bytes": 7205273,
-      "pages": 28,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1896957620001558,
-      "bytes": 10758587,
-      "pages": 42,
+      "seconds": 1.4530193070004316,
+      "bytes": 7690650,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9081728179999118,
-      "bytes": 2358450,
-      "pages": 28,
+      "seconds": 1.0416680760008603,
+      "bytes": 5628205,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.152849502999743,
-      "bytes": 6119990,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0915457699993567,
+      "bytes": 13214903,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.0355856849996599,
-      "bytes": 6832540,
-      "pages": 32,
+      "seconds": 1.4333472569996957,
+      "bytes": 6210705,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 20.55054765299974,
-      "bytes": 11612692,
+      "seconds": 1.232096203000765,
+      "bytes": 1201273,
       "pages": 35,
-      "shards": 4
+      "shards": 1
     },
     {
-      "seconds": 1.2162191410006926,
-      "bytes": 6401722,
-      "pages": 45,
-      "shards": 2
+      "seconds": 1.2199994860002334,
+      "bytes": 7355179,
+      "pages": 26,
+      "shards": 3
     },
     {
-      "seconds": 1.3737986060004914,
-      "bytes": 7819084,
-      "pages": 35,
+      "seconds": 1.1187948159986263,
+      "bytes": 10435269,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.079906239000593,
-      "bytes": 5671187,
-      "pages": 35,
+      "seconds": 0.8605066399995849,
+      "bytes": 757336,
+      "pages": 24,
+      "shards": 1
+    },
+    {
+      "seconds": 1.4440002779992938,
+      "bytes": 13477323,
+      "pages": 45,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0930595740010176,
+      "bytes": 6005049,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.1921040119996178,
-      "bytes": 5850261,
+      "seconds": 1.1409170160004578,
+      "bytes": 7751245,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.385997141000189,
+      "bytes": 5854774,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9688384539995241,
-      "bytes": 8315017,
+      "seconds": 0.8962630369987892,
+      "bytes": 1124581,
       "pages": 35,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 1.0787890209994657,
-      "bytes": 6174644,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.1323460120001982,
+      "bytes": 7094094,
+      "pages": 25,
+      "shards": 3
     },
     {
-      "seconds": 1.431994874999873,
-      "bytes": 7632897,
+      "seconds": 1.3368760009998368,
+      "bytes": 10726853,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.237401497999599,
-      "bytes": 5052734,
-      "pages": 33,
+      "seconds": 0.9546346269999049,
+      "bytes": 5999756,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0236900689997128,
-      "bytes": 6421100,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1835868019988993,
+      "bytes": 7840439,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.9477791999997862,
-      "bytes": 2414897,
-      "pages": 27,
+      "seconds": 1.00542208799925,
+      "bytes": 5360697,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.0987587250001525,
-      "bytes": 11899150,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4687352659993849,
-      "bytes": 10700369,
-      "pages": 42,
+      "seconds": 1.2967056479992607,
+      "bytes": 7612762,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1858354729993152,
-      "bytes": 2742758,
-      "pages": 37,
-      "shards": 2
+      "seconds": 0.9550784620005288,
+      "bytes": 883731,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.046531567000784,
-      "bytes": 5904437,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1562814480002999,
+      "bytes": 11922123,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.4689589510007863,
-      "bytes": 7230449,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.2267546230013977,
+      "bytes": 11778233,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.0110029770003166,
-      "bytes": 6117097,
-      "pages": 37,
+      "seconds": 0.9305929850015673,
+      "bytes": 1085853,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1878640209997684,
-      "bytes": 11776096,
-      "pages": 38,
+      "seconds": 1.4772485220000817,
+      "bytes": 7565247,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.224845865000134,
-      "bytes": 2065038,
-      "pages": 19,
+      "seconds": 1.035932449000029,
+      "bytes": 5952411,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 0.9610936170001878,
-      "bytes": 5798019,
-      "pages": 27,
-      "shards": 2
+      "seconds": 0.7514245589991333,
+      "bytes": 257606,
+      "pages": 8,
+      "shards": 1
     },
     {
-      "seconds": 1.1097549460000664,
-      "bytes": 5080154,
-      "pages": 36,
+      "seconds": 0.9524619530002383,
+      "bytes": 4638620,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.1122178149998945,
-      "bytes": 7607293,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.3739831450002384,
+      "bytes": 13357834,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.0907007900004828,
-      "bytes": 6092133,
+      "seconds": 1.1055001610002364,
+      "bytes": 6014129,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8937167260000933,
-      "bytes": 2214439,
-      "pages": 24,
+      "seconds": 0.9146980179993989,
+      "bytes": 2516554,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.3153861170003438,
-      "bytes": 17177113,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.1637235170001077,
+      "bytes": 10485207,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.9797320220004622,
-      "bytes": 1103154,
-      "pages": 35,
+      "seconds": 1.0128666689997772,
+      "bytes": 6455974,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.195222697000645,
-      "bytes": 7323521,
-      "pages": 29,
+      "seconds": 1.0892136950005806,
+      "bytes": 8273419,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.158046825999918,
-      "bytes": 6457008,
+      "seconds": 1.3893709189997026,
+      "bytes": 6184789,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1688710530006574,
-      "bytes": 6589194,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0017207049995704,
-      "bytes": 5856566,
-      "pages": 27,
+      "seconds": 1.226723518999279,
+      "bytes": 5211595,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.1244407199992565,
-      "bytes": 12346633,
-      "pages": 40,
+      "seconds": 1.4460434569991776,
+      "bytes": 7613200,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9011206420000235,
-      "bytes": 2301339,
-      "pages": 25,
+      "seconds": 1.303638638000848,
+      "bytes": 6027470,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4037659219993657,
-      "bytes": 10492690,
-      "pages": 42,
-      "shards": 3
+      "seconds": 1.9445103569996718,
+      "bytes": 2479005,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 0.9956328799999028,
-      "bytes": 1276481,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.5074067269997613,
+      "bytes": 15537281,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.4530193070004316,
-      "bytes": 7690650,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9092938880003203,
+      "bytes": 1113077,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.0416680760008603,
-      "bytes": 5628205,
-      "pages": 24,
-      "shards": 2
+      "seconds": 1.0412252780006384,
+      "bytes": 7665353,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.0915457699993567,
-      "bytes": 13214903,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.1747847249989718,
+      "bytes": 10716254,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.4333472569996957,
-      "bytes": 6210705,
-      "pages": 36,
+      "seconds": 1.0711626960001013,
+      "bytes": 2631436,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.232096203000765,
-      "bytes": 1201273,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.137612402000741,
+      "bytes": 5722821,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.2199994860002334,
-      "bytes": 7355179,
-      "pages": 26,
+      "seconds": 1.4303736129986646,
+      "bytes": 11919616,
+      "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 1.1187948159986263,
-      "bytes": 10435269,
-      "pages": 45,
+      "seconds": 1.0499967649993778,
+      "bytes": 7310677,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.8605066399995849,
-      "bytes": 757336,
-      "pages": 24,
-      "shards": 1
+      "seconds": 1.0670534790006059,
+      "bytes": 5802765,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.4440002779992938,
-      "bytes": 13477323,
-      "pages": 45,
-      "shards": 4
+      "seconds": 0.9111343770000531,
+      "bytes": 2649643,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0930595740010176,
-      "bytes": 6005049,
-      "pages": 30,
+      "seconds": 1.2196053769985156,
+      "bytes": 6057143,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1409170160004578,
-      "bytes": 7751245,
-      "pages": 38,
+      "seconds": 1.2347321990000637,
+      "bytes": 9643472,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.385997141000189,
-      "bytes": 5854774,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1648727040010272,
+      "bytes": 8469052,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.8962630369987892,
-      "bytes": 1124581,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.1223673130007228,
+      "bytes": 6068911,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.1323460120001982,
-      "bytes": 7094094,
-      "pages": 25,
-      "shards": 3
+      "seconds": 1.1270621679996111,
+      "bytes": 5842224,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.3368760009998368,
-      "bytes": 10726853,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.3927226420000807,
+      "bytes": 7163586,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 0.9546346269999049,
-      "bytes": 5999756,
-      "pages": 29,
+      "seconds": 1.230830958000297,
+      "bytes": 6020552,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1835868019988993,
-      "bytes": 7840439,
-      "pages": 38,
+      "seconds": 1.2729082169989852,
+      "bytes": 7700586,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.00542208799925,
-      "bytes": 5360697,
+      "seconds": 0.9891643420014589,
+      "bytes": 6883017,
       "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.2967056479992607,
-      "bytes": 7612762,
-      "pages": 38,
+      "seconds": 1.2779713299987634,
+      "bytes": 9455856,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9550784620005288,
-      "bytes": 883731,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.159703276000073,
+      "bytes": 2570411,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.1562814480002999,
-      "bytes": 11922123,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.2064628600001015,
+      "bytes": 6189523,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.2267546230013977,
-      "bytes": 11778233,
-      "pages": 33,
+      "seconds": 1.4404870660000597,
+      "bytes": 11907122,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 0.9305929850015673,
-      "bytes": 1085853,
+      "seconds": 0.842472636999446,
+      "bytes": 857193,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0425618379995285,
+      "bytes": 11991809,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.4772485220000817,
-      "bytes": 7565247,
-      "pages": 36,
+      "seconds": 1.543527111000003,
+      "bytes": 7770797,
+      "pages": 39,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b45969c25feb44bef2bbc28e1a234c7ba36a717f6375ea0dba3053855528f94
-size 721759935
+oid sha256:043647c8c26e8fed077bb197406f27a2662b7f2ed8742ee4065fa206a57818d0
+size 721757375