Qwen90M step 775000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 770000,
+  "step": 775000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "e0a55fb5bc214d02022f2eff8a8e4d4a8989e853edc73e1a2ebb5d0b7571c743",
+    "model.safetensors": "62d59c2938f0295d3b0f8368ee314c8fc85d4acb69d47b1b11a15d84b82da2a7",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "a498688babdf2cae771028b6c89d36575e2ed244953a97468586e9e5595fe80e",
+    "trainer.pt": "3b0cdf7569ff3f756bba96adbbd77bced2a88ef4f965a14cbe15078a1fd95894",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d",
-    "shard_scheduler.json": "d4f4b8859106eed29ba6de07b880ae2a733ce84e3ecc5289a12c47202ed328b5",
-    "coverage.json": "0b5c9661fa1c47243f8665233277a741ea19c000bbb01a96a0cb91fec8b18598",
+    "shard_scheduler.json": "9a616e2bb05ee9b5cb075c6eccfe164bbcf57cf736739b2ec6ff6e8eead85217",
+    "coverage.json": "df852170d83f78171ae60b5d208710eaa74621666ecccf6bcd6b78262cb0b924",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 770000,
+  "step": 775000,
   "stats": {
-    "climbmix/blocks": 27720000,
-    "climbmix/targets": 46338195596,
-    "climbmix/input_tokens": 46341132368,
-    "ultra_qa/blocks": 10395000,
-    "ultra_qa/targets": 16958238643,
-    "ultra_qa/input_tokens": 16958271297,
-    "cortex/blocks": 10395000,
-    "cortex/targets": 395254793,
-    "cortex/input_tokens": 16523352599,
-    "ultra_style/blocks": 10395000,
-    "ultra_style/targets": 16442103528,
-    "ultra_style/input_tokens": 16442965847,
-    "openmath/blocks": 6930000,
-    "openmath/targets": 10708073683,
-    "openmath/input_tokens": 10711641641,
-    "rewrite6/blocks": 3465000,
-    "rewrite6/targets": 5680408647,
-    "rewrite6/input_tokens": 5686636248
+    "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
   },
   "sources": {
     "climbmix": {
-      "page": 3796,
-      "row": 2700,
+      "page": 3821,
+      "row": 1836,
       "start": {
-        "file": 183,
+        "file": 184,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 61,
+          "row_group": 78,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 15551116
+      "offset": 0,
+      "documents": 15652652
     },
     "rewrite6": {
-      "page": 9,
-      "row": 2929,
+      "page": 23,
+      "row": 3880,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 36864
+          "row": 94208
         }
       },
       "epoch": 43,
       "offset": 0,
-      "documents": 8872165
+      "documents": 8930460
     },
     "openmath": {
-      "page": 2372,
-      "row": 2873,
+      "page": 2388,
+      "row": 276,
       "start": {
         "file": 22,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 109,
-          "row": 412
+          "row_group": 174,
+          "row": 948
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 9718585
+      "documents": 9781524
     },
     "ultra_qa": {
-      "page": 1025,
-      "row": 3695,
+      "page": 1032,
+      "row": 2254,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 41389
+          "row_group": 1,
+          "row": 5540
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 4202095
+      "documents": 4229326
     },
     "ultra_style": {
-      "page": 1565,
-      "row": 3159,
+      "page": 1575,
+      "row": 3932,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
-          "row": 75510
+          "row_group": 3,
+          "row": 31988
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6413399
+      "documents": 6455132
     },
     "cortex": {
-      "page": 201598,
-      "row": 95,
+      "page": 202908,
+      "row": 41,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 25804639
+      "documents": 25972265
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 770000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e0a55fb5bc214d02022f2eff8a8e4d4a8989e853edc73e1a2ebb5d0b7571c743", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a498688babdf2cae771028b6c89d36575e2ed244953a97468586e9e5595fe80e", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "d4f4b8859106eed29ba6de07b880ae2a733ce84e3ecc5289a12c47202ed328b5", "coverage.json": "0b5c9661fa1c47243f8665233277a741ea19c000bbb01a96a0cb91fec8b18598", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000770000"}
+{"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"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e0a55fb5bc214d02022f2eff8a8e4d4a8989e853edc73e1a2ebb5d0b7571c743
+oid sha256:62d59c2938f0295d3b0f8368ee314c8fc85d4acb69d47b1b11a15d84b82da2a7
 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": 76400.702468659,
-      "required_pages_per_second_upper_estimate": 0.005285577870835155
+      "required_token_positions_per_second": 77130.99093735527,
+      "required_pages_per_second_upper_estimate": 0.0053361009216022675
     },
     {
       "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": 9550.087808582375,
-      "required_pages_per_second_upper_estimate": 0.003998422336942226
+      "required_token_positions_per_second": 9641.373867169408,
+      "required_pages_per_second_upper_estimate": 0.004036641903402943
     },
     {
       "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": 19100.17561716475,
-      "required_pages_per_second_upper_estimate": 0.003978062875551665
+      "required_token_positions_per_second": 19282.747734338816,
+      "required_pages_per_second_upper_estimate": 0.004016087832808515
     },
     {
       "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": 28650.263425747125,
-      "required_pages_per_second_upper_estimate": 0.0017175395459542889
+      "required_token_positions_per_second": 28924.121601508225,
+      "required_pages_per_second_upper_estimate": 0.0017339569254339446
     },
     {
       "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": 28650.263425747125,
-      "required_pages_per_second_upper_estimate": 0.002227431962161148
+      "required_token_positions_per_second": 28924.121601508225,
+      "required_pages_per_second_upper_estimate": 0.0022487232307517617
     },
     {
       "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": 28650.263425747125,
-      "required_pages_per_second_upper_estimate": 0.3674948169693453
+      "required_token_positions_per_second": 28924.121601508225,
+      "required_pages_per_second_upper_estimate": 0.37100757560200903
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,338 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.0745995468515872,
-    "producer_p95_seconds": 0.19039384700045048,
-    "gpu_update_seconds": 0.9650173050469608,
-    "producer_batches_per_second": 13.40490716370516,
-    "gpu_batches_per_second": 1.036250847285414,
-    "producer_headroom": 12.935967385523451,
+    "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,
     "sustainable": true,
-    "buffer_seconds": 60.79609021795853,
+    "buffer_seconds": 60.22046318285337,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 80,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 277550.1790140697,
-    "upload_bytes_per_second": 5458592.488142992,
+    "generation_bytes_per_second": 462776.96354896243,
+    "upload_bytes_per_second": 5621658.650253022,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 59.555807378926936,
+    "estimated_full_shards_per_hour": 99.30116348125128,
     "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": 0.9713157099999989,
-      "bytes": 23380,
-      "pages": 1,
+      "seconds": 1.0387209649998113,
+      "bytes": 1064900,
+      "pages": 32,
       "shards": 1
     },
     {
-      "seconds": 1.497052147999966,
-      "bytes": 11817101,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.1239516590003404,
+      "bytes": 11388095,
+      "pages": 24,
+      "shards": 3
     },
     {
-      "seconds": 1.0985135669998272,
-      "bytes": 6249223,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.270762184999512,
+      "bytes": 12213784,
+      "pages": 48,
+      "shards": 4
     },
     {
-      "seconds": 1.3139188940003805,
-      "bytes": 2690437,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.305295600000136,
+      "bytes": 1053111,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.4740382569998474,
-      "bytes": 11539917,
-      "pages": 32,
+      "seconds": 1.13043969499995,
+      "bytes": 7715241,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.530777683999986,
-      "bytes": 10014604,
-      "pages": 34,
+      "seconds": 1.1387970109999515,
+      "bytes": 10453428,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.4347790020001412,
-      "bytes": 2652487,
-      "pages": 37,
+      "seconds": 1.3777619700003925,
+      "bytes": 819264,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.0014089890000832,
-      "bytes": 6005067,
+      "seconds": 1.2043475289992784,
+      "bytes": 13122126,
+      "pages": 40,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1971378379994349,
+      "bytes": 5971358,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9762548340004287,
-      "bytes": 2290276,
-      "pages": 28,
+      "seconds": 0.9979249650004931,
+      "bytes": 6639528,
+      "pages": 32,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0738694510000641,
+      "bytes": 6116466,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4867014489991561,
-      "bytes": 11031744,
-      "pages": 35,
+      "seconds": 0.993601779000528,
+      "bytes": 1093689,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0732885580000584,
+      "bytes": 7199351,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 2.084590124000897,
-      "bytes": 12097526,
+      "seconds": 1.3326483809996716,
+      "bytes": 16444186,
       "pages": 37,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 0.8632086769994203,
-      "bytes": 2602865,
-      "pages": 34,
+      "seconds": 0.9647154859994771,
+      "bytes": 2573191,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1280111480000414,
-      "bytes": 298657,
-      "pages": 9,
+      "seconds": 1.2671719289992325,
+      "bytes": 6070484,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9431054610013234,
+      "bytes": 842654,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.3505803890002426,
-      "bytes": 6226670,
-      "pages": 36,
+      "seconds": 1.3556453979999787,
+      "bytes": 11128099,
+      "pages": 32,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1991763389996777,
+      "bytes": 5989028,
+      "pages": 39,
       "shards": 2
     },
     {
-      "seconds": 1.6309672779998436,
-      "bytes": 11574403,
-      "pages": 34,
+      "seconds": 1.483354228999815,
+      "bytes": 13516314,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.1955943939992721,
-      "bytes": 921426,
-      "pages": 30,
+      "seconds": 0.855230520999612,
+      "bytes": 838138,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.0380561770007262,
-      "bytes": 6122044,
+      "seconds": 1.1489447340009065,
+      "bytes": 10403004,
+      "pages": 36,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9511843500004034,
+      "bytes": 2374318,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.678345239999544,
+      "bytes": 6141170,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.4052477490004094,
-      "bytes": 13455700,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.0750775129999965,
+      "bytes": 1235188,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2133236099998612,
+      "bytes": 11087663,
+      "pages": 9,
+      "shards": 3
+    },
+    {
+      "seconds": 1.060693833998812,
+      "bytes": 7602919,
+      "pages": 45,
+      "shards": 3
     },
     {
-      "seconds": 0.9797629530003178,
-      "bytes": 4961107,
+      "seconds": 1.202748789000907,
+      "bytes": 6204281,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0490742649999447,
-      "bytes": 7239835,
+      "seconds": 1.4798138580008526,
+      "bytes": 7393031,
       "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.0533643100006884,
-      "bytes": 1118487,
+      "seconds": 0.8912479129994608,
+      "bytes": 1116344,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.5156246349997673,
-      "bytes": 10630930,
-      "pages": 34,
+      "seconds": 1.331704049000109,
+      "bytes": 10313766,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1701987970000118,
-      "bytes": 7151255,
+      "seconds": 1.5044031500001438,
+      "bytes": 8199304,
       "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.1764370720002262,
-      "bytes": 6881233,
-      "pages": 35,
+      "seconds": 1.4808365860008053,
+      "bytes": 6083362,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.185791879000135,
-      "bytes": 7588973,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.2456822759995703,
+      "bytes": 12102575,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 0.8400346700000227,
-      "bytes": 806070,
-      "pages": 24,
+      "seconds": 0.872640274999867,
+      "bytes": 1185922,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.260675972000172,
-      "bytes": 10735591,
-      "pages": 41,
-      "shards": 3
+      "seconds": 1.1524197270009608,
+      "bytes": 6081827,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.1190538959999685,
-      "bytes": 7646532,
-      "pages": 38,
+      "seconds": 1.1447629150006833,
+      "bytes": 7321171,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.0387209649998113,
-      "bytes": 1064900,
+      "seconds": 1.0708932610014017,
+      "bytes": 6812845,
       "pages": 32,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1239516590003404,
-      "bytes": 11388095,
-      "pages": 24,
       "shards": 3
     },
     {
-      "seconds": 1.270762184999512,
-      "bytes": 12213784,
-      "pages": 48,
+      "seconds": 1.235942547000377,
+      "bytes": 11771801,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.305295600000136,
-      "bytes": 1053111,
-      "pages": 30,
+      "seconds": 1.195207675000347,
+      "bytes": 1134238,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.13043969499995,
-      "bytes": 7715241,
+      "seconds": 1.0769621990002634,
+      "bytes": 6468284,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 2.3685029429998394,
+      "bytes": 11410659,
       "pages": 34,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 1.1387970109999515,
-      "bytes": 10453428,
-      "pages": 40,
-      "shards": 3
+      "seconds": 0.8315555410008528,
+      "bytes": 1017212,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "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,
+      "seconds": 1.084104961999401,
+      "bytes": 947822,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1121349749992078,
+      "bytes": 6275461,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.993601779000528,
-      "bytes": 1093689,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.2975029250010266,
+      "bytes": 6264701,
+      "pages": 26,
+      "shards": 3
     },
     {
-      "seconds": 1.0732885580000584,
-      "bytes": 7199351,
-      "pages": 29,
+      "seconds": 1.4136529820007127,
+      "bytes": 11858220,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.3326483809996716,
-      "bytes": 16444186,
+      "seconds": 1.0470604990005086,
+      "bytes": 6246532,
       "pages": 37,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 0.9647154859994771,
-      "bytes": 2573191,
+      "seconds": 1.0089250809996884,
+      "bytes": 2580280,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2671719289992325,
-      "bytes": 6070484,
-      "pages": 36,
+      "seconds": 1.1785594149987446,
+      "bytes": 10228291,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9172461540001677,
+      "bytes": 2354642,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9431054610013234,
-      "bytes": 842654,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.1485085579988663,
+      "bytes": 12167168,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.3556453979999787,
-      "bytes": 11128099,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.0151017729986052,
+      "bytes": 6037243,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.1991763389996777,
-      "bytes": 5989028,
-      "pages": 39,
+      "seconds": 1.2390309599995817,
+      "bytes": 6886865,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3525706070013257,
+      "bytes": 5675921,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.483354228999815,
-      "bytes": 13516314,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.2551854249995813,
+      "bytes": 7559192,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.855230520999612,
-      "bytes": 838138,
-      "pages": 29,
+      "seconds": 0.825367109000581,
+      "bytes": 1141341,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.1489447340009065,
-      "bytes": 10403004,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.5828043500005151,
+      "bytes": 15814359,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 0.9511843500004034,
-      "bytes": 2374318,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.3585308149995399,
+      "bytes": 7486305,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.678345239999544,
-      "bytes": 6141170,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8041142309994029,
+      "bytes": 1148050,
+      "pages": 34,
+      "shards": 1
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a498688babdf2cae771028b6c89d36575e2ed244953a97468586e9e5595fe80e
-size 721763327
+oid sha256:3b0cdf7569ff3f756bba96adbbd77bced2a88ef4f965a14cbe15078a1fd95894
+size 721780543