Qwen90M step 780000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 775000,
+  "step": 780000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "62d59c2938f0295d3b0f8368ee314c8fc85d4acb69d47b1b11a15d84b82da2a7",
+    "model.safetensors": "861f90278f68313b09eb4b302450aa4b5f8e1fe53d075de34cb4897b13546b54",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "3b0cdf7569ff3f756bba96adbbd77bced2a88ef4f965a14cbe15078a1fd95894",
+    "trainer.pt": "bb7ceaeda93bfb3b23d8a7a9fc6b379ef73c40abcbb8d58fb638329c569f90e7",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d",
-    "shard_scheduler.json": "9a616e2bb05ee9b5cb075c6eccfe164bbcf57cf736739b2ec6ff6e8eead85217",
-    "coverage.json": "df852170d83f78171ae60b5d208710eaa74621666ecccf6bcd6b78262cb0b924",
+    "shard_scheduler.json": "060f57d01a3356f0f8395d272cc24eb06656aeba401c5dcbc708fe202e44878a",
+    "coverage.json": "f641f693c6260b4b353bf8683e17a82b7b292ad28deab6832a52269599d71b92",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 775000,
+  "step": 780000,
   "stats": {
-    "climbmix/blocks": 27900000,
-    "climbmix/targets": 46638920719,
-    "climbmix/input_tokens": 46641876738,
-    "ultra_qa/blocks": 10462500,
-    "ultra_qa/targets": 17068384083,
-    "ultra_qa/input_tokens": 17068416928,
-    "cortex/blocks": 10462500,
-    "cortex/targets": 397833446,
-    "cortex/input_tokens": 16630731446,
-    "ultra_style/blocks": 10462500,
-    "ultra_style/targets": 16549240635,
-    "ultra_style/input_tokens": 16550108681,
-    "openmath/blocks": 6975000,
-    "openmath/targets": 10777571114,
-    "openmath/input_tokens": 10781162121,
-    "rewrite6/blocks": 3487500,
-    "rewrite6/targets": 5717285134,
-    "rewrite6/input_tokens": 5723553756
+    "climbmix/blocks": 28080000,
+    "climbmix/targets": 46940530276,
+    "climbmix/input_tokens": 46943505263,
+    "ultra_qa/blocks": 10530000,
+    "ultra_qa/targets": 17178444425,
+    "ultra_qa/input_tokens": 17178477489,
+    "cortex/blocks": 10530000,
+    "cortex/targets": 400402885,
+    "cortex/input_tokens": 16738077716,
+    "ultra_style/blocks": 10530000,
+    "ultra_style/targets": 16655737210,
+    "ultra_style/input_tokens": 16656610869,
+    "openmath/blocks": 7020000,
+    "openmath/targets": 10847110567,
+    "openmath/input_tokens": 10850724850,
+    "rewrite6/blocks": 3510000,
+    "rewrite6/targets": 5754125788,
+    "rewrite6/input_tokens": 5760434426
   },
   "sources": {
     "climbmix": {
-      "page": 3821,
-      "row": 1836,
+      "page": 3845,
+      "row": 3683,
       "start": {
-        "file": 184,
+        "file": 186,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 78,
+          "row_group": 10,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 15652652
+      "offset": 4096,
+      "documents": 15752803
     },
     "rewrite6": {
-      "page": 23,
-      "row": 3880,
+      "page": 37,
+      "row": 3740,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 94208
+          "row": 151552
         }
       },
       "epoch": 43,
-      "offset": 0,
-      "documents": 8930460
+      "offset": 2048,
+      "documents": 8987664
     },
     "openmath": {
-      "page": 2388,
-      "row": 276,
+      "page": 2403,
+      "row": 2100,
       "start": {
         "file": 22,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 174,
-          "row": 948
+          "row_group": 236,
+          "row": 388
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 9781524
+      "documents": 9844788
     },
     "ultra_qa": {
-      "page": 1032,
-      "row": 2254,
+      "page": 1039,
+      "row": 799,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 5540
+          "row": 34212
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 4229326
+      "offset": 2048,
+      "documents": 4256543
     },
     "ultra_style": {
-      "page": 1575,
-      "row": 3932,
+      "page": 1586,
+      "row": 561,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 31988
+          "row": 77044
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6455132
+      "documents": 6496817
     },
     "cortex": {
-      "page": 202908,
-      "row": 41,
+      "page": 204216,
+      "row": 101,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 25972265
+      "documents": 26139749
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 775000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "62d59c2938f0295d3b0f8368ee314c8fc85d4acb69d47b1b11a15d84b82da2a7", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "3b0cdf7569ff3f756bba96adbbd77bced2a88ef4f965a14cbe15078a1fd95894", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "9a616e2bb05ee9b5cb075c6eccfe164bbcf57cf736739b2ec6ff6e8eead85217", "coverage.json": "df852170d83f78171ae60b5d208710eaa74621666ecccf6bcd6b78262cb0b924", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000775000"}
+{"step": 780000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "861f90278f68313b09eb4b302450aa4b5f8e1fe53d075de34cb4897b13546b54", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "bb7ceaeda93bfb3b23d8a7a9fc6b379ef73c40abcbb8d58fb638329c569f90e7", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "060f57d01a3356f0f8395d272cc24eb06656aeba401c5dcbc708fe202e44878a", "coverage.json": "f641f693c6260b4b353bf8683e17a82b7b292ad28deab6832a52269599d71b92", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000780000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:62d59c2938f0295d3b0f8368ee314c8fc85d4acb69d47b1b11a15d84b82da2a7
+oid sha256:861f90278f68313b09eb4b302450aa4b5f8e1fe53d075de34cb4897b13546b54
 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": 77130.99093735527,
-      "required_pages_per_second_upper_estimate": 0.0053361009216022675
+      "required_token_positions_per_second": 77198.30405065721,
+      "required_pages_per_second_upper_estimate": 0.005340757798968437
     },
     {
       "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": 9641.373867169408,
-      "required_pages_per_second_upper_estimate": 0.004036641903402943
+      "required_token_positions_per_second": 9649.788006332152,
+      "required_pages_per_second_upper_estimate": 0.0040401647277631784
     },
     {
       "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": 19282.747734338816,
-      "required_pages_per_second_upper_estimate": 0.004016087832808515
+      "required_token_positions_per_second": 19299.576012664304,
+      "required_pages_per_second_upper_estimate": 0.004019592719392171
     },
     {
       "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": 28924.121601508225,
-      "required_pages_per_second_upper_estimate": 0.0017339569254339446
+      "required_token_positions_per_second": 28949.364018996454,
+      "required_pages_per_second_upper_estimate": 0.001735470169819424
     },
     {
       "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": 28924.121601508225,
-      "required_pages_per_second_upper_estimate": 0.0022487232307517617
+      "required_token_positions_per_second": 28949.364018996454,
+      "required_pages_per_second_upper_estimate": 0.0022506857176818108
     },
     {
       "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": 28924.121601508225,
-      "required_pages_per_second_upper_estimate": 0.37100757560200903
+      "required_token_positions_per_second": 28949.364018996454,
+      "required_pages_per_second_upper_estimate": 0.37133135823035174
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.07639025929684351,
-    "producer_p95_seconds": 0.21046823000142467,
-    "gpu_update_seconds": 0.9558803679817995,
-    "producer_batches_per_second": 13.090674245706095,
-    "gpu_batches_per_second": 1.0461560185730696,
-    "producer_headroom": 12.513118515115409,
+    "producer_batch_seconds": 0.39113858623451847,
+    "producer_p95_seconds": 0.258143648999976,
+    "gpu_update_seconds": 0.9550468874500143,
+    "producer_batches_per_second": 2.5566385807827743,
+    "gpu_batches_per_second": 1.047069011103749,
+    "producer_headroom": 2.4417097189112105,
     "sustainable": true,
-    "buffer_seconds": 60.22046318285337,
+    "buffer_seconds": 60.1679539093509,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 80,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 462776.96354896243,
-    "upload_bytes_per_second": 5621658.650253022,
+    "generation_bytes_per_second": 310654.2999911742,
+    "upload_bytes_per_second": 4935605.042243135,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 99.30116348125128,
+    "estimated_full_shards_per_hour": 66.65918111611767,
     "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": 1.0387209649998113,
-      "bytes": 1064900,
-      "pages": 32,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1239516590003404,
-      "bytes": 11388095,
-      "pages": 24,
-      "shards": 3
-    },
-    {
-      "seconds": 1.270762184999512,
-      "bytes": 12213784,
-      "pages": 48,
+      "seconds": 2.3685029429998394,
+      "bytes": 11410659,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.305295600000136,
-      "bytes": 1053111,
+      "seconds": 0.8315555410008528,
+      "bytes": 1017212,
       "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.13043969499995,
-      "bytes": 7715241,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1387970109999515,
-      "bytes": 10453428,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3777619700003925,
-      "bytes": 819264,
-      "pages": 26,
+      "seconds": 1.2100102580006933,
+      "bytes": 5596380,
+      "pages": 21,
       "shards": 2
     },
     {
-      "seconds": 1.2043475289992784,
-      "bytes": 13122126,
-      "pages": 40,
+      "seconds": 1.2646026780003012,
+      "bytes": 13750693,
+      "pages": 48,
       "shards": 4
     },
     {
-      "seconds": 1.1971378379994349,
-      "bytes": 5971358,
-      "pages": 36,
+      "seconds": 1.3428173240008618,
+      "bytes": 5558514,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 0.9979249650004931,
-      "bytes": 6639528,
-      "pages": 32,
+      "seconds": 1.273476110000047,
+      "bytes": 7479055,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0738694510000641,
-      "bytes": 6116466,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.084104961999401,
+      "bytes": 947822,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 0.993601779000528,
-      "bytes": 1093689,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1121349749992078,
+      "bytes": 6275461,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0732885580000584,
-      "bytes": 7199351,
-      "pages": 29,
+      "seconds": 1.2975029250010266,
+      "bytes": 6264701,
+      "pages": 26,
       "shards": 3
     },
     {
-      "seconds": 1.3326483809996716,
-      "bytes": 16444186,
+      "seconds": 1.4136529820007127,
+      "bytes": 11858220,
       "pages": 37,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.9647154859994771,
-      "bytes": 2573191,
+      "seconds": 1.0470604990005086,
+      "bytes": 6246532,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2671719289992325,
-      "bytes": 6070484,
-      "pages": 36,
+      "seconds": 1.0089250809996884,
+      "bytes": 2580280,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9431054610013234,
-      "bytes": 842654,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.3556453979999787,
-      "bytes": 11128099,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1785594149987446,
+      "bytes": 10228291,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1991763389996777,
-      "bytes": 5989028,
-      "pages": 39,
+      "seconds": 0.9172461540001677,
+      "bytes": 2354642,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.483354228999815,
-      "bytes": 13516314,
+      "seconds": 1.1485085579988663,
+      "bytes": 12167168,
       "pages": 40,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.855230520999612,
-      "bytes": 838138,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.0151017729986052,
+      "bytes": 6037243,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.1489447340009065,
-      "bytes": 10403004,
-      "pages": 36,
+      "seconds": 1.2390309599995817,
+      "bytes": 6886865,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.9511843500004034,
-      "bytes": 2374318,
-      "pages": 29,
+      "seconds": 1.3525706070013257,
+      "bytes": 5675921,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.678345239999544,
-      "bytes": 6141170,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2551854249995813,
+      "bytes": 7559192,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.0750775129999965,
-      "bytes": 1235188,
-      "pages": 35,
+      "seconds": 0.825367109000581,
+      "bytes": 1141341,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.2133236099998612,
-      "bytes": 11087663,
-      "pages": 9,
-      "shards": 3
+      "seconds": 1.5828043500005151,
+      "bytes": 15814359,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.060693833998812,
-      "bytes": 7602919,
-      "pages": 45,
+      "seconds": 1.3585308149995399,
+      "bytes": 7486305,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.202748789000907,
-      "bytes": 6204281,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8041142309994029,
+      "bytes": 1148050,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.4798138580008526,
-      "bytes": 7393031,
+      "seconds": 1.0195367809992604,
+      "bytes": 6121053,
       "pages": 30,
+      "shards": 2
+    },
+    {
+      "seconds": 2.6576393979994464,
+      "bytes": 1839915,
+      "pages": 12,
+      "shards": 2
+    },
+    {
+      "seconds": 1.339241851999759,
+      "bytes": 10779600,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.8912479129994608,
-      "bytes": 1116344,
+      "seconds": 4.318171066001014,
+      "bytes": 1094935,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.331704049000109,
-      "bytes": 10313766,
+      "seconds": 1.3115034270012984,
+      "bytes": 13301880,
+      "pages": 34,
+      "shards": 5
+    },
+    {
+      "seconds": 1.1869826189995365,
+      "bytes": 6445004,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.5044031500001438,
-      "bytes": 8199304,
-      "pages": 34,
+      "seconds": 1.1019184790002328,
+      "bytes": 6210962,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.4808365860008053,
-      "bytes": 6083362,
-      "pages": 37,
+      "seconds": 1.991184726000938,
+      "bytes": 5824172,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2456822759995703,
-      "bytes": 12102575,
-      "pages": 32,
+      "seconds": 1.041803035999692,
+      "bytes": 1241755,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3904252129996166,
+      "bytes": 12223668,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 0.872640274999867,
-      "bytes": 1185922,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.4052468339996267,
+      "bytes": 11659745,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.1524197270009608,
-      "bytes": 6081827,
-      "pages": 36,
+      "seconds": 1.587534140999196,
+      "bytes": 2763901,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1447629150006833,
-      "bytes": 7321171,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.0879035409998323,
+      "bytes": 6026132,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.0708932610014017,
-      "bytes": 6812845,
-      "pages": 32,
-      "shards": 3
+      "seconds": 0.9889250640007958,
+      "bytes": 952891,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 1.235942547000377,
-      "bytes": 11771801,
-      "pages": 40,
+      "seconds": 1.1961971269993228,
+      "bytes": 12036916,
+      "pages": 46,
       "shards": 4
     },
     {
-      "seconds": 1.195207675000347,
-      "bytes": 1134238,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1770325620018411,
+      "bytes": 6029556,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.0769621990002634,
-      "bytes": 6468284,
-      "pages": 36,
+      "seconds": 1.1039014870002575,
+      "bytes": 6886174,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 2.3685029429998394,
-      "bytes": 11410659,
-      "pages": 34,
+      "seconds": 1.8041386129989405,
+      "bytes": 11898323,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 0.8315555410008528,
-      "bytes": 1017212,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.2189035779992992,
+      "bytes": 6954573,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.2100102580006933,
-      "bytes": 5596380,
-      "pages": 21,
+      "seconds": 1.5599068500014255,
+      "bytes": 2415324,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2646026780003012,
-      "bytes": 13750693,
-      "pages": 48,
-      "shards": 4
+      "seconds": 0.996803005997208,
+      "bytes": 6353091,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.3428173240008618,
-      "bytes": 5558514,
-      "pages": 31,
+      "seconds": 1.149405411000771,
+      "bytes": 4840926,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.273476110000047,
-      "bytes": 7479055,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.5049040789999708,
+      "bytes": 13497203,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.084104961999401,
-      "bytes": 947822,
-      "pages": 26,
-      "shards": 1
+      "seconds": 2.2471259019985155,
+      "bytes": 6049971,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1121349749992078,
-      "bytes": 6275461,
+      "seconds": 1.1292916710008285,
+      "bytes": 2686677,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2975029250010266,
-      "bytes": 6264701,
-      "pages": 26,
+      "seconds": 1.3082805390004069,
+      "bytes": 10490550,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.4136529820007127,
-      "bytes": 11858220,
+      "seconds": 1.171707494999282,
+      "bytes": 6154958,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.0470604990005086,
-      "bytes": 6246532,
-      "pages": 37,
+      "seconds": 1.3786574580008164,
+      "bytes": 2401890,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0089250809996884,
-      "bytes": 2580280,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.469864372000302,
+      "bytes": 11763636,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.1785594149987446,
-      "bytes": 10228291,
-      "pages": 34,
+      "seconds": 1.3772630089988525,
+      "bytes": 6008251,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9172461540001677,
-      "bytes": 2354642,
-      "pages": 28,
+      "seconds": 1.3044562169998244,
+      "bytes": 6256652,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1485085579988663,
-      "bytes": 12167168,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.3504138789976423,
+      "bytes": 6609036,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0151017729986052,
-      "bytes": 6037243,
-      "pages": 35,
+      "seconds": 0.9880936410008871,
+      "bytes": 2162409,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.2390309599995817,
-      "bytes": 6886865,
-      "pages": 35,
+      "seconds": 1.2899976480002806,
+      "bytes": 11585751,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.3525706070013257,
-      "bytes": 5675921,
+      "seconds": 1.4168642419972457,
+      "bytes": 6615057,
       "pages": 28,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.2551854249995813,
-      "bytes": 7559192,
+      "seconds": 1.1043323689991666,
+      "bytes": 7549149,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.825367109000581,
-      "bytes": 1141341,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1122719240011065,
+      "bytes": 6287535,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.5828043500005151,
-      "bytes": 15814359,
+      "seconds": 1.1806445659967721,
+      "bytes": 6303271,
       "pages": 33,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.3585308149995399,
-      "bytes": 7486305,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.2656237250012055,
+      "bytes": 5785840,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.8041142309994029,
-      "bytes": 1148050,
-      "pages": 34,
+      "seconds": 1.1340223679981136,
+      "bytes": 1097457,
+      "pages": 35,
       "shards": 1
+    },
+    {
+      "seconds": 3.1717748339979153,
+      "bytes": 13032899,
+      "pages": 32,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3b0cdf7569ff3f756bba96adbbd77bced2a88ef4f965a14cbe15078a1fd95894
-size 721780543
+oid sha256:bb7ceaeda93bfb3b23d8a7a9fc6b379ef73c40abcbb8d58fb638329c569f90e7
+size 721774783