Qwen90M step 25000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 20000,
+  "step": 25000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "8ac28a22873400a2ad81f120bca87505f54dd1c44997b9d1fb50fb7b5e06c38a",
+    "model.safetensors": "df7303e39f8a38cb32caa015de73f0c83cb91ddb60382d6c2876d432a817c840",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "f18e92b7be914c509e752f35a75c1d1da97d9597baf2c06a4ed1fd4a3d17d365",
+    "trainer.pt": "c0a2189fdee90c71c35753bc51ffdb3afecbeb4a346bd3889656ce4fe54fb477",
     "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2",
     "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
-    "shard_scheduler.json": "ba70b7966dd2918d6c10d0e338ae611c7dc6c0c15e14439a5011a3b293e76f49",
-    "coverage.json": "4e370713d47399509a8d42fcc8692de21ea4cb81a32e0196eac9278823ef761f",
+    "shard_scheduler.json": "61528e6c5607c5d05d8a1210d80299914e613d3ffcdb126da4557c3c47fdf788",
+    "coverage.json": "de500062e0edf2f10dd1cb2862bfcdb202710953fdc9e56a645d42f5467026c1",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 20000,
+  "step": 25000,
   "stats": {
-    "climbmix/blocks": 720000,
-    "climbmix/targets": 1202690209,
-    "climbmix/input_tokens": 1202766834,
-    "ultra_qa/blocks": 270000,
-    "ultra_qa/targets": 440598577,
-    "ultra_qa/input_tokens": 440599431,
-    "cortex/blocks": 270000,
-    "cortex/targets": 10270926,
-    "cortex/input_tokens": 429238220,
-    "ultra_style/blocks": 270000,
-    "ultra_style/targets": 426974586,
-    "ultra_style/input_tokens": 426997226,
-    "openmath/blocks": 180000,
-    "openmath/targets": 278145153,
-    "openmath/input_tokens": 278238017,
-    "rewrite6/blocks": 90000,
-    "rewrite6/targets": 147537674,
-    "rewrite6/input_tokens": 147699508
+    "climbmix/blocks": 900000,
+    "climbmix/targets": 1503316012,
+    "climbmix/input_tokens": 1503411635,
+    "ultra_qa/blocks": 337500,
+    "ultra_qa/targets": 550725921,
+    "ultra_qa/input_tokens": 550727011,
+    "cortex/blocks": 337500,
+    "cortex/targets": 12839020,
+    "cortex/input_tokens": 536531199,
+    "ultra_style/blocks": 337500,
+    "ultra_style/targets": 533604810,
+    "ultra_style/input_tokens": 533633023,
+    "openmath/blocks": 225000,
+    "openmath/targets": 347653521,
+    "openmath/input_tokens": 347769506,
+    "rewrite6/blocks": 112500,
+    "rewrite6/targets": 184371575,
+    "rewrite6/input_tokens": 184573987
   },
   "sources": {
     "climbmix": {
-      "page": 98,
-      "row": 3561,
+      "page": 123,
+      "row": 2056,
       "start": {
-        "file": 4,
+        "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 58,
+          "row_group": 76,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 404969
+      "offset": 0,
+      "documents": 505864
     },
     "rewrite6": {
-      "page": 6,
-      "row": 493,
+      "page": 20,
+      "row": 960,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 24576
+          "row": 81920
         }
       },
       "epoch": 1,
       "offset": 2048,
-      "documents": 230473
+      "documents": 288284
     },
     "openmath": {
-      "page": 61,
-      "row": 2771,
+      "page": 77,
+      "row": 320,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 249,
-          "row": 856
+          "row_group": 315,
+          "row": 392
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 252627
+      "documents": 315712
     },
     "ultra_qa": {
-      "page": 26,
-      "row": 2708,
+      "page": 33,
+      "row": 1362,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 41926
+          "row_group": 2,
+          "row": 5892
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 109204
+      "documents": 136530
     },
     "ultra_style": {
-      "page": 40,
-      "row": 3229,
+      "page": 50,
+      "row": 3964,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 78473
+          "row_group": 2,
+          "row": 33851
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 167069
+      "documents": 208764
     },
     "cortex": {
-      "page": 5235,
-      "row": 127,
+      "page": 6544,
+      "row": 9,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 670207
+      "documents": 837641
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 20000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "8ac28a22873400a2ad81f120bca87505f54dd1c44997b9d1fb50fb7b5e06c38a", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "f18e92b7be914c509e752f35a75c1d1da97d9597baf2c06a4ed1fd4a3d17d365", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "ba70b7966dd2918d6c10d0e338ae611c7dc6c0c15e14439a5011a3b293e76f49", "coverage.json": "4e370713d47399509a8d42fcc8692de21ea4cb81a32e0196eac9278823ef761f", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000020000"}
+{"step": 25000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "df7303e39f8a38cb32caa015de73f0c83cb91ddb60382d6c2876d432a817c840", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "c0a2189fdee90c71c35753bc51ffdb3afecbeb4a346bd3889656ce4fe54fb477", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "61528e6c5607c5d05d8a1210d80299914e613d3ffcdb126da4557c3c47fdf788", "coverage.json": "de500062e0edf2f10dd1cb2862bfcdb202710953fdc9e56a645d42f5467026c1", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000025000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ac28a22873400a2ad81f120bca87505f54dd1c44997b9d1fb50fb7b5e06c38a
+oid sha256:df7303e39f8a38cb32caa015de73f0c83cb91ddb60382d6c2876d432a817c840
 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": 75780.15754960763,
-      "required_pages_per_second_upper_estimate": 0.006568431993346605
+      "required_token_positions_per_second": 75999.12812604562,
+      "required_pages_per_second_upper_estimate": 0.006587411807936398
     },
     {
       "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": 9472.519693700953,
-      "required_pages_per_second_upper_estimate": 0.0037787297326076883
+      "required_token_positions_per_second": 9499.891015755702,
+      "required_pages_per_second_upper_estimate": 0.0037896485622130613
     },
     {
       "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": 18945.039387401906,
-      "required_pages_per_second_upper_estimate": 0.004382601440969964
+      "required_token_positions_per_second": 18999.782031511404,
+      "required_pages_per_second_upper_estimate": 0.004395265188247066
     },
     {
       "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": 28417.55908110286,
-      "required_pages_per_second_upper_estimate": 0.001730480252629366
+      "required_token_positions_per_second": 28499.673047267104,
+      "required_pages_per_second_upper_estimate": 0.0017354805623500836
     },
     {
       "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": 28417.55908110286,
-      "required_pages_per_second_upper_estimate": 0.0021876623629016084
+      "required_token_positions_per_second": 28499.673047267104,
+      "required_pages_per_second_upper_estimate": 0.0021939837233229385
     },
     {
       "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": 28417.55908110286,
-      "required_pages_per_second_upper_estimate": 0.3528509763351361
+      "required_token_positions_per_second": 28499.673047267104,
+      "required_pages_per_second_upper_estimate": 0.3538705569771298
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.09428515504697543,
-    "producer_p95_seconds": 0.2934562519985775,
-    "gpu_update_seconds": 0.9729195924637102,
-    "producer_batches_per_second": 10.60612351437268,
-    "gpu_batches_per_second": 1.0278341681533152,
-    "producer_headroom": 10.318905367223241,
+    "producer_batch_seconds": 0.12395837693742351,
+    "producer_p95_seconds": 0.32236783699772786,
+    "gpu_update_seconds": 0.970116392358095,
+    "producer_batches_per_second": 8.067224052996584,
+    "gpu_batches_per_second": 1.0308041466748807,
+    "producer_headroom": 7.826146294637495,
     "sustainable": true,
-    "buffer_seconds": 60.32101473275003,
+    "buffer_seconds": 60.14721632620189,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 83,
+    "upload_pages_per_commit": 80,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 321461.5944819022,
-    "upload_bytes_per_second": 5587316.407845459,
+    "generation_bytes_per_second": 466257.7871084776,
+    "upload_bytes_per_second": 5728821.4302538335,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 68.97817493288802,
+    "estimated_full_shards_per_hour": 100.04806718769785,
     "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": 0.7792264220006473,
-      "bytes": 714539,
-      "pages": 19,
-      "shards": 1
+      "seconds": 1.0224713220013655,
+      "bytes": 6909652,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.1150978799996665,
-      "bytes": 7568611,
-      "pages": 45,
-      "shards": 3
+      "seconds": 1.018912650000857,
+      "bytes": 5819747,
+      "pages": 22,
+      "shards": 2
     },
     {
-      "seconds": 1.1983624430013151,
-      "bytes": 7531256,
-      "pages": 35,
+      "seconds": 0.8826371689974621,
+      "bytes": 4172226,
+      "pages": 46,
       "shards": 3
     },
     {
-      "seconds": 1.255419142999017,
-      "bytes": 11119971,
+      "seconds": 2.508591282999987,
+      "bytes": 5665007,
       "pages": 21,
-      "shards": 3
-    },
-    {
-      "seconds": 1.229843424000137,
-      "bytes": 7832261,
-      "pages": 45,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.9293085339995741,
-      "bytes": 2351431,
+      "seconds": 1.1267430049992981,
+      "bytes": 4735418,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.204088584998317,
-      "bytes": 7306502,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0363285590010491,
-      "bytes": 11683428,
-      "pages": 36,
+      "seconds": 1.0588101130015275,
+      "bytes": 8209294,
+      "pages": 47,
       "shards": 4
     },
     {
-      "seconds": 1.0476808559997153,
-      "bytes": 995690,
+      "seconds": 1.0802435129990045,
+      "bytes": 11583568,
       "pages": 29,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5069925719999446,
-      "bytes": 1775692,
-      "pages": 21,
-      "shards": 2
-    },
-    {
-      "seconds": 1.199250067000321,
-      "bytes": 13436756,
-      "pages": 47,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.2015854260007472,
-      "bytes": 10309364,
-      "pages": 34,
+      "seconds": 1.2433872929977952,
+      "bytes": 7696542,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9169029310014594,
-      "bytes": 7362656,
-      "pages": 34,
-      "shards": 3
+      "seconds": 0.9942282960000739,
+      "bytes": 5730133,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.9597948790014925,
-      "bytes": 2508350,
+      "seconds": 0.9958830260002287,
+      "bytes": 6097937,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1389318259989523,
-      "bytes": 6045358,
-      "pages": 35,
+      "seconds": 1.0204584569983126,
+      "bytes": 2643615,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1304748169986851,
-      "bytes": 12071070,
-      "pages": 27,
+      "seconds": 1.1489640459985822,
+      "bytes": 6010015,
+      "pages": 32,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3807983080005215,
+      "bytes": 15814135,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 3.228665867998643,
-      "bytes": 7676896,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1238995200001227,
+      "bytes": 2339467,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.9868028299988509,
-      "bytes": 6101148,
-      "pages": 35,
+      "seconds": 1.365642591001233,
+      "bytes": 6320146,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8392134009991423,
-      "bytes": 3718175,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.0016697349965398,
+      "bytes": 761758,
+      "pages": 22,
+      "shards": 2
     },
     {
-      "seconds": 1.0338640699992538,
-      "bytes": 9654277,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.1483951949994662,
+      "bytes": 11735955,
+      "pages": 47,
+      "shards": 4
     },
     {
-      "seconds": 0.8135462400005053,
-      "bytes": 1110572,
+      "seconds": 1.0657177410030272,
+      "bytes": 1129396,
       "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.3182481060011924,
-      "bytes": 8487979,
-      "pages": 27,
-      "shards": 4
+      "seconds": 1.0460155590008071,
+      "bytes": 6970566,
+      "pages": 24,
+      "shards": 3
     },
     {
-      "seconds": 1.0571894279983098,
-      "bytes": 6049663,
-      "pages": 10,
-      "shards": 2
+      "seconds": 2.029252282998641,
+      "bytes": 16530913,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.061286989999644,
-      "bytes": 6189988,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.8704905230006261,
+      "bytes": 1186552,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.1359487030003947,
-      "bytes": 6670820,
-      "pages": 29,
+      "seconds": 3.2236978209984954,
+      "bytes": 7889017,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9071375989988155,
-      "bytes": 7422614,
-      "pages": 37,
-      "shards": 4
+      "seconds": 0.9988030870008515,
+      "bytes": 1104863,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 0.8035606240009656,
-      "bytes": 1171760,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0168222699976468,
+      "bytes": 4937209,
+      "pages": 2,
+      "shards": 2
     },
     {
-      "seconds": 1.2951343000004272,
-      "bytes": 7367275,
-      "pages": 27,
-      "shards": 3
+      "seconds": 1.0983402380006737,
+      "bytes": 4472235,
+      "pages": 20,
+      "shards": 2
     },
     {
-      "seconds": 1.1229757969995262,
-      "bytes": 5828242,
-      "pages": 34,
+      "seconds": 1.3725059809985396,
+      "bytes": 7833557,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.539654905001953,
-      "bytes": 12116214,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.04416848999972,
+      "bytes": 7007035,
+      "pages": 42,
+      "shards": 2
     },
     {
-      "seconds": 1.2743941750013619,
-      "bytes": 7473487,
+      "seconds": 1.018632732997503,
+      "bytes": 7180638,
       "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.8017673559988907,
-      "bytes": 2201439,
-      "pages": 35,
+      "seconds": 1.1994012059985835,
+      "bytes": 10151059,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9821173639975314,
+      "bytes": 2386369,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2722073730001284,
-      "bytes": 12361137,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.2013670170017576,
+      "bytes": 6446691,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 0.8463285300022108,
-      "bytes": 752773,
-      "pages": 23,
+      "seconds": 1.0164053049993527,
+      "bytes": 1100982,
+      "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.0778829359987867,
-      "bytes": 7678507,
-      "pages": 43,
-      "shards": 3
+      "seconds": 1.2673739959973318,
+      "bytes": 17927773,
+      "pages": 41,
+      "shards": 5
     },
     {
-      "seconds": 1.0977264680004737,
-      "bytes": 13270237,
-      "pages": 39,
-      "shards": 4
+      "seconds": 0.8297740179987159,
+      "bytes": 928094,
+      "pages": 28,
+      "shards": 1
     },
     {
-      "seconds": 0.8527823730000819,
-      "bytes": 4778684,
-      "pages": 20,
+      "seconds": 0.8634128480007348,
+      "bytes": 6113242,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.9591597530015861,
-      "bytes": 7365649,
-      "pages": 37,
+      "seconds": 0.979562092001288,
+      "bytes": 7577775,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.119754101997387,
-      "bytes": 2574245,
-      "pages": 36,
+      "seconds": 0.9862535339998431,
+      "bytes": 4746157,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0843618420003622,
-      "bytes": 5928833,
-      "pages": 24,
-      "shards": 2
+      "seconds": 1.0129832909988181,
+      "bytes": 7553178,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.330314847000409,
-      "bytes": 13379246,
-      "pages": 41,
-      "shards": 5
+      "seconds": 1.2109974509985477,
+      "bytes": 11754963,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.0224713220013655,
-      "bytes": 6909652,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.878410489000089,
+      "bytes": 971318,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 1.018912650000857,
-      "bytes": 5819747,
-      "pages": 22,
-      "shards": 2
+      "seconds": 1.079362858999957,
+      "bytes": 12108906,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 0.8826371689974621,
-      "bytes": 4172226,
-      "pages": 46,
-      "shards": 3
+      "seconds": 0.808165985999949,
+      "bytes": 1135299,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 2.508591282999987,
-      "bytes": 5665007,
-      "pages": 21,
+      "seconds": 1.9975283729982038,
+      "bytes": 6018894,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.1267430049992981,
-      "bytes": 4735418,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.1784233340003993,
+      "bytes": 7683890,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.0588101130015275,
-      "bytes": 8209294,
-      "pages": 47,
-      "shards": 4
+      "seconds": 1.1247381559987844,
+      "bytes": 11305256,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.0802435129990045,
-      "bytes": 11583568,
+      "seconds": 0.900414362000447,
+      "bytes": 7293854,
       "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.2433872929977952,
-      "bytes": 7696542,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0943440850023762,
+      "bytes": 1102979,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 0.9942282960000739,
-      "bytes": 5730133,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.0738140280009247,
+      "bytes": 6022869,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 0.9958830260002287,
-      "bytes": 6097937,
+      "seconds": 1.1826581230016018,
+      "bytes": 10993511,
       "pages": 35,
-      "shards": 2
+      "shards": 4
     },
     {
-      "seconds": 1.0204584569983126,
-      "bytes": 2643615,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.020132988000114,
+      "bytes": 882161,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.1489640459985822,
-      "bytes": 6010015,
+      "seconds": 1.030573132000427,
+      "bytes": 11789498,
       "pages": 32,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8314693719985371,
+      "bytes": 2720719,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3807983080005215,
-      "bytes": 15814135,
-      "pages": 39,
-      "shards": 4
+      "seconds": 0.9525506049976684,
+      "bytes": 10860039,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1238995200001227,
-      "bytes": 2339467,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.2853242279998085,
+      "bytes": 7445426,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.365642591001233,
-      "bytes": 6320146,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.7597367620001023,
+      "bytes": 831447,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.0016697349965398,
-      "bytes": 761758,
-      "pages": 22,
+      "seconds": 1.0118765539991728,
+      "bytes": 6408285,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1483951949994662,
-      "bytes": 11735955,
-      "pages": 47,
+      "seconds": 1.0685828210007458,
+      "bytes": 12919611,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.0657177410030272,
-      "bytes": 1129396,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.0399747580013354,
+      "bytes": 5973768,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.0460155590008071,
-      "bytes": 6970566,
-      "pages": 24,
-      "shards": 3
+      "seconds": 1.0144561509987398,
+      "bytes": 6361193,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 2.029252282998641,
-      "bytes": 16530913,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.0571298460017715,
+      "bytes": 2527314,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8704905230006261,
-      "bytes": 1186552,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1346876719981083,
+      "bytes": 10084333,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 3.2236978209984954,
-      "bytes": 7889017,
-      "pages": 38,
+      "seconds": 0.9850222569984908,
+      "bytes": 2356425,
+      "pages": 28,
+      "shards": 2
+    },
+    {
+      "seconds": 1.0524133049984812,
+      "bytes": 12178249,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.9988030870008515,
-      "bytes": 1104863,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.109467931000836,
+      "bytes": 5783909,
+      "pages": 28,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f18e92b7be914c509e752f35a75c1d1da97d9597baf2c06a4ed1fd4a3d17d365
-size 721771775
+oid sha256:c0a2189fdee90c71c35753bc51ffdb3afecbeb4a346bd3889656ce4fe54fb477
+size 721767167