Qwen90M step 20000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 15000,
+  "step": 20000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "cd4b4a336ca854cc63b56daafd448bbdabc975be153afb5b40dbcfeed84c0a5c",
+    "model.safetensors": "8ac28a22873400a2ad81f120bca87505f54dd1c44997b9d1fb50fb7b5e06c38a",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "9ee3cdbc8544ee947af4a7d860b9cd7d6a5386d7eb87fa5e677065b54efe108a",
+    "trainer.pt": "f18e92b7be914c509e752f35a75c1d1da97d9597baf2c06a4ed1fd4a3d17d365",
     "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2",
     "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
-    "shard_scheduler.json": "4d92f7ac5793b7b10fc6dcd78ea5d5fc6a289ca95e0d0b954b4effa3df49ceff",
-    "coverage.json": "ae6b95743a2e517acaa8daca9ca6ec0408a66feeeb897332ac3aa2b7585e4311",
+    "shard_scheduler.json": "ba70b7966dd2918d6c10d0e338ae611c7dc6c0c15e14439a5011a3b293e76f49",
+    "coverage.json": "4e370713d47399509a8d42fcc8692de21ea4cb81a32e0196eac9278823ef761f",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 15000,
+  "step": 20000,
   "stats": {
-    "climbmix/blocks": 540000,
-    "climbmix/targets": 901514897,
-    "climbmix/input_tokens": 901572469,
-    "ultra_qa/blocks": 202500,
-    "ultra_qa/targets": 330561975,
-    "ultra_qa/input_tokens": 330562626,
-    "cortex/blocks": 202500,
-    "cortex/targets": 7710129,
-    "cortex/input_tokens": 321940622,
-    "ultra_style/blocks": 202500,
-    "ultra_style/targets": 320384397,
-    "ultra_style/input_tokens": 320401468,
-    "openmath/blocks": 135000,
-    "openmath/targets": 208599081,
-    "openmath/input_tokens": 208668782,
-    "rewrite6/blocks": 67500,
-    "rewrite6/targets": 110633666,
-    "rewrite6/input_tokens": 110755126
+    "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
   },
   "sources": {
     "climbmix": {
-      "page": 74,
-      "row": 1396,
+      "page": 98,
+      "row": 3561,
       "start": {
-        "file": 3,
+        "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 44,
+          "row_group": 58,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 304500
+      "documents": 404969
     },
     "rewrite6": {
-      "page": 42,
-      "row": 1009,
+      "page": 6,
+      "row": 493,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 172032
+          "row": 24576
         }
       },
-      "epoch": 0,
-      "offset": 0,
-      "documents": 173041
+      "epoch": 1,
+      "offset": 2048,
+      "documents": 230473
     },
     "openmath": {
-      "page": 46,
-      "row": 1133,
+      "page": 61,
+      "row": 2771,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 188,
-          "row": 416
+          "row_group": 249,
+          "row": 856
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 189549
+      "documents": 252627
     },
     "ultra_qa": {
-      "page": 20,
-      "row": 1,
+      "page": 26,
+      "row": 2708,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 17350
+          "row": 41926
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 81921
+      "offset": 2048,
+      "documents": 109204
     },
     "ultra_style": {
-      "page": 30,
-      "row": 2561,
+      "page": 40,
+      "row": 3229,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 37513
+          "row": 78473
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 125441
+      "documents": 167069
     },
     "cortex": {
-      "page": 3926,
-      "row": 10,
+      "page": 5235,
+      "row": 127,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 502538
+      "documents": 670207
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 15000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "cd4b4a336ca854cc63b56daafd448bbdabc975be153afb5b40dbcfeed84c0a5c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "9ee3cdbc8544ee947af4a7d860b9cd7d6a5386d7eb87fa5e677065b54efe108a", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "4d92f7ac5793b7b10fc6dcd78ea5d5fc6a289ca95e0d0b954b4effa3df49ceff", "coverage.json": "ae6b95743a2e517acaa8daca9ca6ec0408a66feeeb897332ac3aa2b7585e4311", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000015000"}
+{"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"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cd4b4a336ca854cc63b56daafd448bbdabc975be153afb5b40dbcfeed84c0a5c
+oid sha256:8ac28a22873400a2ad81f120bca87505f54dd1c44997b9d1fb50fb7b5e06c38a
 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": 75908.27937948794,
-      "required_pages_per_second_upper_estimate": 0.006579537268838822
+      "required_token_positions_per_second": 75780.15754960763,
+      "required_pages_per_second_upper_estimate": 0.006568431993346605
     },
     {
       "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": 9488.534922435992,
-      "required_pages_per_second_upper_estimate": 0.0037851184467991034
+      "required_token_positions_per_second": 9472.519693700953,
+      "required_pages_per_second_upper_estimate": 0.0037787297326076883
     },
     {
       "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": 18977.069844871985,
-      "required_pages_per_second_upper_estimate": 0.004390011123588869
+      "required_token_positions_per_second": 18945.039387401906,
+      "required_pages_per_second_upper_estimate": 0.004382601440969964
     },
     {
       "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": 28465.604767307974,
-      "required_pages_per_second_upper_estimate": 0.0017334059828430235
+      "required_token_positions_per_second": 28417.55908110286,
+      "required_pages_per_second_upper_estimate": 0.001730480252629366
     },
     {
       "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": 28465.604767307974,
-      "required_pages_per_second_upper_estimate": 0.0021913610528246498
+      "required_token_positions_per_second": 28417.55908110286,
+      "required_pages_per_second_upper_estimate": 0.0021876623629016084
     },
     {
       "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": 28465.604767307974,
-      "required_pages_per_second_upper_estimate": 0.3534475429592358
+      "required_token_positions_per_second": 28417.55908110286,
+      "required_pages_per_second_upper_estimate": 0.3528509763351361
     }
   ],
   "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": 0.6275874297735697,
-    "producer_p95_seconds": 0.29346390000137035,
-    "gpu_update_seconds": 0.9712774496101002,
-    "producer_batches_per_second": 1.5934034885956763,
-    "gpu_batches_per_second": 1.029571931687933,
-    "producer_headroom": 1.5476368766030448,
+    "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,
     "sustainable": true,
-    "buffer_seconds": 60.219201875826215,
+    "buffer_seconds": 60.32101473275003,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 234605.4285099929,
-    "upload_bytes_per_second": 6302700.702389904,
+    "generation_bytes_per_second": 321461.5944819022,
+    "upload_bytes_per_second": 5587316.407845459,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 50.340863623379136,
+    "estimated_full_shards_per_hour": 68.97817493288802,
     "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": 1.0970937430001868,
-      "bytes": 8464356,
-      "pages": 36,
-      "shards": 4
+      "seconds": 0.7792264220006473,
+      "bytes": 714539,
+      "pages": 19,
+      "shards": 1
     },
     {
-      "seconds": 1.18106462400101,
-      "bytes": 6740941,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.1150978799996665,
+      "bytes": 7568611,
+      "pages": 45,
+      "shards": 3
     },
     {
-      "seconds": 1.0259071060008864,
-      "bytes": 6033717,
+      "seconds": 1.1983624430013151,
+      "bytes": 7531256,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.3882136779993743,
-      "bytes": 13040894,
-      "pages": 40,
-      "shards": 5
-    },
-    {
-      "seconds": 0.7154928489999293,
-      "bytes": 835967,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.255419142999017,
+      "bytes": 11119971,
+      "pages": 21,
+      "shards": 3
     },
     {
-      "seconds": 1.0692436869994708,
-      "bytes": 7577178,
-      "pages": 36,
+      "seconds": 1.229843424000137,
+      "bytes": 7832261,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.0859860309992655,
-      "bytes": 2128705,
-      "pages": 34,
+      "seconds": 0.9293085339995741,
+      "bytes": 2351431,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.1122738659996685,
-      "bytes": 11663489,
-      "pages": 30,
+      "seconds": 1.204088584998317,
+      "bytes": 7306502,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1172626269999455,
-      "bytes": 11245903,
+      "seconds": 1.0363285590010491,
+      "bytes": 11683428,
       "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 0.9253301299995655,
-      "bytes": 2328336,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.028129626000009,
-      "bytes": 6087526,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.0476808559997153,
+      "bytes": 995690,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 0.7597279170004185,
-      "bytes": 2156570,
+      "seconds": 1.5069925719999446,
+      "bytes": 1775692,
       "pages": 21,
       "shards": 2
     },
     {
-      "seconds": 0.9871070610006427,
-      "bytes": 12008396,
-      "pages": 45,
+      "seconds": 1.199250067000321,
+      "bytes": 13436756,
+      "pages": 47,
       "shards": 4
     },
     {
-      "seconds": 1.228597503999481,
-      "bytes": 13077636,
+      "seconds": 1.2015854260007472,
+      "bytes": 10309364,
       "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1013024290004978,
-      "bytes": 1132593,
-      "pages": 33,
-      "shards": 1
-    },
-    {
-      "seconds": 0.9999866519992793,
-      "bytes": 6918002,
-      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.2065209279990086,
-      "bytes": 11423370,
-      "pages": 39,
-      "shards": 4
+      "seconds": 0.9169029310014594,
+      "bytes": 7362656,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.0073676210013218,
-      "bytes": 932900,
-      "pages": 27,
-      "shards": 1
+      "seconds": 0.9597948790014925,
+      "bytes": 2508350,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.0279631710000103,
-      "bytes": 7068347,
+      "seconds": 1.1389318259989523,
+      "bytes": 6045358,
       "pages": 35,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.8251235360003193,
-      "bytes": 2046300,
-      "pages": 19,
-      "shards": 2
+      "seconds": 1.1304748169986851,
+      "bytes": 12071070,
+      "pages": 27,
+      "shards": 4
     },
     {
-      "seconds": 1.6571312910000415,
-      "bytes": 16761254,
+      "seconds": 3.228665867998643,
+      "bytes": 7676896,
       "pages": 38,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.7977973630004271,
-      "bytes": 2044807,
-      "pages": 28,
+      "seconds": 0.9868028299988509,
+      "bytes": 6101148,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0466349409998656,
-      "bytes": 7179273,
-      "pages": 20,
+      "seconds": 0.8392134009991423,
+      "bytes": 3718175,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.7437304710001627,
-      "bytes": 927772,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.0338640699992538,
+      "bytes": 9654277,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.1199327470003482,
-      "bytes": 6072182,
-      "pages": 30,
-      "shards": 2
+      "seconds": 0.8135462400005053,
+      "bytes": 1110572,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 0.9855451060011546,
-      "bytes": 8735177,
-      "pages": 38,
+      "seconds": 1.3182481060011924,
+      "bytes": 8487979,
+      "pages": 27,
       "shards": 4
     },
     {
-      "seconds": 0.964613582000311,
-      "bytes": 5800981,
-      "pages": 27,
+      "seconds": 1.0571894279983098,
+      "bytes": 6049663,
+      "pages": 10,
       "shards": 2
     },
     {
-      "seconds": 0.8288159710009495,
-      "bytes": 7059901,
-      "pages": 37,
+      "seconds": 1.061286989999644,
+      "bytes": 6189988,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9880495930010511,
-      "bytes": 8851756,
-      "pages": 43,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0367009409983439,
-      "bytes": 10231090,
-      "pages": 31,
+      "seconds": 1.1359487030003947,
+      "bytes": 6670820,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.8111973979994218,
-      "bytes": 875148,
-      "pages": 25,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2982671780009696,
-      "bytes": 8839899,
-      "pages": 39,
+      "seconds": 0.9071375989988155,
+      "bytes": 7422614,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 0.756701359001454,
-      "bytes": 1071685,
-      "pages": 31,
+      "seconds": 0.8035606240009656,
+      "bytes": 1171760,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.2869488490014191,
-      "bytes": 16651954,
-      "pages": 29,
-      "shards": 5
+      "seconds": 1.2951343000004272,
+      "bytes": 7367275,
+      "pages": 27,
+      "shards": 3
     },
     {
-      "seconds": 0.9651243050011544,
-      "bytes": 7745172,
-      "pages": 44,
-      "shards": 4
+      "seconds": 1.1229757969995262,
+      "bytes": 5828242,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.7505128630000399,
-      "bytes": 1172556,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.539654905001953,
+      "bytes": 12116214,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.1141684859994712,
-      "bytes": 8597889,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.2743941750013619,
+      "bytes": 7473487,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.1277135729997099,
-      "bytes": 5536865,
-      "pages": 31,
+      "seconds": 0.8017673559988907,
+      "bytes": 2201439,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9057988900003693,
-      "bytes": 5888539,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.2722073730001284,
+      "bytes": 12361137,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.030597670998759,
-      "bytes": 7980071,
-      "pages": 26,
-      "shards": 3
+      "seconds": 0.8463285300022108,
+      "bytes": 752773,
+      "pages": 23,
+      "shards": 1
     },
     {
-      "seconds": 0.9780282619994978,
-      "bytes": 7468851,
-      "pages": 44,
+      "seconds": 1.0778829359987867,
+      "bytes": 7678507,
+      "pages": 43,
       "shards": 3
     },
     {
-      "seconds": 1.20604977199946,
-      "bytes": 11385510,
-      "pages": 35,
+      "seconds": 1.0977264680004737,
+      "bytes": 13270237,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 0.7792264220006473,
-      "bytes": 714539,
-      "pages": 19,
-      "shards": 1
+      "seconds": 0.8527823730000819,
+      "bytes": 4778684,
+      "pages": 20,
+      "shards": 2
     },
     {
-      "seconds": 1.1150978799996665,
-      "bytes": 7568611,
-      "pages": 45,
+      "seconds": 0.9591597530015861,
+      "bytes": 7365649,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1983624430013151,
-      "bytes": 7531256,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.119754101997387,
+      "bytes": 2574245,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.255419142999017,
-      "bytes": 11119971,
-      "pages": 21,
-      "shards": 3
+      "seconds": 1.0843618420003622,
+      "bytes": 5928833,
+      "pages": 24,
+      "shards": 2
     },
     {
-      "seconds": 1.229843424000137,
-      "bytes": 7832261,
-      "pages": 45,
-      "shards": 3
+      "seconds": 1.330314847000409,
+      "bytes": 13379246,
+      "pages": 41,
+      "shards": 5
     },
     {
-      "seconds": 0.9293085339995741,
-      "bytes": 2351431,
-      "pages": 32,
+      "seconds": 1.0224713220013655,
+      "bytes": 6909652,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.204088584998317,
-      "bytes": 7306502,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.018912650000857,
+      "bytes": 5819747,
+      "pages": 22,
+      "shards": 2
     },
     {
-      "seconds": 1.0363285590010491,
-      "bytes": 11683428,
-      "pages": 36,
-      "shards": 4
+      "seconds": 0.8826371689974621,
+      "bytes": 4172226,
+      "pages": 46,
+      "shards": 3
     },
     {
-      "seconds": 1.0476808559997153,
-      "bytes": 995690,
-      "pages": 29,
-      "shards": 1
+      "seconds": 2.508591282999987,
+      "bytes": 5665007,
+      "pages": 21,
+      "shards": 2
     },
     {
-      "seconds": 1.5069925719999446,
-      "bytes": 1775692,
-      "pages": 21,
+      "seconds": 1.1267430049992981,
+      "bytes": 4735418,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.199250067000321,
-      "bytes": 13436756,
+      "seconds": 1.0588101130015275,
+      "bytes": 8209294,
       "pages": 47,
       "shards": 4
     },
     {
-      "seconds": 1.2015854260007472,
-      "bytes": 10309364,
-      "pages": 34,
+      "seconds": 1.0802435129990045,
+      "bytes": 11583568,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.9169029310014594,
-      "bytes": 7362656,
-      "pages": 34,
+      "seconds": 1.2433872929977952,
+      "bytes": 7696542,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9597948790014925,
-      "bytes": 2508350,
-      "pages": 35,
+      "seconds": 0.9942282960000739,
+      "bytes": 5730133,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1389318259989523,
-      "bytes": 6045358,
+      "seconds": 0.9958830260002287,
+      "bytes": 6097937,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1304748169986851,
-      "bytes": 12071070,
-      "pages": 27,
+      "seconds": 1.0204584569983126,
+      "bytes": 2643615,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "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,
+      "seconds": 1.0460155590008071,
+      "bytes": 6970566,
+      "pages": 24,
+      "shards": 3
+    },
+    {
+      "seconds": 2.029252282998641,
+      "bytes": 16530913,
+      "pages": 39,
       "shards": 4
+    },
+    {
+      "seconds": 0.8704905230006261,
+      "bytes": 1186552,
+      "pages": 33,
+      "shards": 1
+    },
+    {
+      "seconds": 3.2236978209984954,
+      "bytes": 7889017,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9988030870008515,
+      "bytes": 1104863,
+      "pages": 33,
+      "shards": 1
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ee3cdbc8544ee947af4a7d860b9cd7d6a5386d7eb87fa5e677065b54efe108a
-size 721787967
+oid sha256:f18e92b7be914c509e752f35a75c1d1da97d9597baf2c06a4ed1fd4a3d17d365
+size 721771775