Qwen90M step 680000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 675000,
+  "step": 680000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "74dc22051f83b4c8984329a3666317ee194d1224a37d93e4e1a34b0174c4bbf8",
+    "model.safetensors": "4a1d373cfe63b63f51c7756f3587c44e326964e25d196bd6fd7f16ad403624e6",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "3557acc152a7dfbc059cc3cb10ac7d0ee8716631de59030b922ca4ad23b76a00",
+    "trainer.pt": "def47c4f3c010dce6d7624c71b7cffcba72fe3c499691cd19b43ed26143caae4",
     "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb",
     "performance.json": "5949473856ba60c6ea6e09958b880f6c16bf1db77e55259c9de9f88123633edf",
-    "shard_scheduler.json": "5cf47e165a37561b98deded2063e10e26d280446908c77d7d22b1dd0b6ca3f1e",
-    "coverage.json": "2965b7663d35170c3297939a7eeaa8e936903ecb9f013cfad5133eed9394b125",
+    "shard_scheduler.json": "c51fe95c8d0f826af2d22730226b70169bd01d52e0fa35e70b3c271ed878f995",
+    "coverage.json": "1fa3ed346a97a26d94a789368927da0f3d9343e3c8dde50979494db8173cd5df",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 675000,
+  "step": 680000,
   "stats": {
-    "climbmix/blocks": 24300000,
-    "climbmix/targets": 40622544143,
-    "climbmix/input_tokens": 40625118554,
-    "ultra_qa/blocks": 9112500,
-    "ultra_qa/targets": 14866485854,
-    "ultra_qa/input_tokens": 14866514573,
-    "cortex/blocks": 9112500,
-    "cortex/targets": 346482518,
-    "cortex/input_tokens": 14484924008,
-    "ultra_style/blocks": 9112500,
-    "ultra_style/targets": 14413351865,
-    "ultra_style/input_tokens": 14414108313,
-    "openmath/blocks": 6075000,
-    "openmath/targets": 9387040123,
-    "openmath/input_tokens": 9390167294,
-    "rewrite6/blocks": 3037500,
-    "rewrite6/targets": 4979579505,
-    "rewrite6/input_tokens": 4985038925
+    "climbmix/blocks": 24480000,
+    "climbmix/targets": 40923747216,
+    "climbmix/input_tokens": 40926340688,
+    "ultra_qa/blocks": 9180000,
+    "ultra_qa/targets": 14976547101,
+    "ultra_qa/input_tokens": 14976576055,
+    "cortex/blocks": 9180000,
+    "cortex/targets": 349045275,
+    "cortex/input_tokens": 14592186628,
+    "ultra_style/blocks": 9180000,
+    "ultra_style/targets": 14519965635,
+    "ultra_style/input_tokens": 14520727794,
+    "openmath/blocks": 6120000,
+    "openmath/targets": 9456596693,
+    "openmath/input_tokens": 9459747141,
+    "rewrite6/blocks": 3060000,
+    "rewrite6/targets": 5016477122,
+    "rewrite6/input_tokens": 5021976869
   },
   "sources": {
     "climbmix": {
-      "page": 3327,
-      "row": 3996,
+      "page": 3352,
+      "row": 2082,
       "start": {
-        "file": 161,
+        "file": 162,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
+          "row_group": 20,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 13631388
+      "documents": 13731874
     },
     "rewrite6": {
-      "page": 43,
-      "row": 1704,
+      "page": 7,
+      "row": 1154,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 176128
+          "row": 28672
         }
       },
-      "epoch": 37,
-      "offset": 2048,
-      "documents": 7777780
+      "epoch": 38,
+      "offset": 0,
+      "documents": 7835178
     },
     "openmath": {
-      "page": 2079,
-      "row": 3284,
+      "page": 2095,
+      "row": 988,
       "start": {
         "file": 19,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 219,
-          "row": 234
+          "row_group": 284,
+          "row": 770
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 8518868
+      "offset": 2048,
+      "documents": 8582108
     },
     "ultra_qa": {
-      "page": 899,
-      "row": 1468,
+      "page": 906,
+      "row": 0,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 44920
+          "row_group": 1,
+          "row": 9040
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3683772
+      "documents": 3710976
     },
     "ultra_style": {
-      "page": 1372,
-      "row": 3093,
+      "page": 1382,
+      "row": 3920,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 55620
+          "row_group": 2,
+          "row": 11087
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5622805
+      "documents": 5664592
     },
     "cortex": {
-      "page": 176729,
-      "row": 46,
+      "page": 178037,
+      "row": 110,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 22621358
+      "documents": 22788846
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 675000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "74dc22051f83b4c8984329a3666317ee194d1224a37d93e4e1a34b0174c4bbf8", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "3557acc152a7dfbc059cc3cb10ac7d0ee8716631de59030b922ca4ad23b76a00", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "5949473856ba60c6ea6e09958b880f6c16bf1db77e55259c9de9f88123633edf", "shard_scheduler.json": "5cf47e165a37561b98deded2063e10e26d280446908c77d7d22b1dd0b6ca3f1e", "coverage.json": "2965b7663d35170c3297939a7eeaa8e936903ecb9f013cfad5133eed9394b125", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000675000"}
+{"step": 680000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "4a1d373cfe63b63f51c7756f3587c44e326964e25d196bd6fd7f16ad403624e6", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "def47c4f3c010dce6d7624c71b7cffcba72fe3c499691cd19b43ed26143caae4", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "5949473856ba60c6ea6e09958b880f6c16bf1db77e55259c9de9f88123633edf", "shard_scheduler.json": "c51fe95c8d0f826af2d22730226b70169bd01d52e0fa35e70b3c271ed878f995", "coverage.json": "1fa3ed346a97a26d94a789368927da0f3d9343e3c8dde50979494db8173cd5df", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000680000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:74dc22051f83b4c8984329a3666317ee194d1224a37d93e4e1a34b0174c4bbf8
+oid sha256:4a1d373cfe63b63f51c7756f3587c44e326964e25d196bd6fd7f16ad403624e6
 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": 76357.51005666674,
-      "required_pages_per_second_upper_estimate": 0.005282589719553327
+      "required_token_positions_per_second": 76877.10649082456,
+      "required_pages_per_second_upper_estimate": 0.005318536606498196
     },
     {
       "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": 9544.688757083342,
-      "required_pages_per_second_upper_estimate": 0.003996161866824596
+      "required_token_positions_per_second": 9609.63831135307,
+      "required_pages_per_second_upper_estimate": 0.004023354888896408
     },
     {
       "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": 19089.377514166685,
-      "required_pages_per_second_upper_estimate": 0.0039758139154622936
+      "required_token_positions_per_second": 19219.27662270614,
+      "required_pages_per_second_upper_estimate": 0.004002868474101203
     },
     {
       "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": 28634.066271250027,
-      "required_pages_per_second_upper_estimate": 0.0017165685512738108
+      "required_token_positions_per_second": 28828.91493405921,
+      "required_pages_per_second_upper_estimate": 0.0017282494310925398
     },
     {
       "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": 28634.066271250027,
-      "required_pages_per_second_upper_estimate": 0.002226172704642752
+      "required_token_positions_per_second": 28828.91493405921,
+      "required_pages_per_second_upper_estimate": 0.0022413213311274736
     },
     {
       "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": 28634.066271250027,
-      "required_pages_per_second_upper_estimate": 0.367287057262606
+      "required_token_positions_per_second": 28828.91493405921,
+      "required_pages_per_second_upper_estimate": 0.36978636669692805
     }
   ],
   "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": 0.08469342544537994,
-    "producer_p95_seconds": 0.2953102920000674,
-    "gpu_update_seconds": 0.965563177024561,
-    "producer_batches_per_second": 11.80729194434242,
-    "gpu_batches_per_second": 1.0356650127043556,
-    "producer_headroom": 11.400686321835773,
+    "producer_batch_seconds": 0.13814311317139527,
+    "producer_p95_seconds": 0.3269497630026308,
+    "gpu_update_seconds": 0.9590371355716879,
+    "producer_batches_per_second": 7.238869727507096,
+    "gpu_batches_per_second": 1.0427124903811924,
+    "producer_headroom": 6.94234488824501,
     "sustainable": true,
-    "buffer_seconds": 60.83048015254734,
+    "buffer_seconds": 60.419339541016335,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 331883.15068931517,
-    "upload_bytes_per_second": 5613899.857854061,
+    "generation_bytes_per_second": 548419.5468379183,
+    "upload_bytes_per_second": 5461558.934611099,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 71.21439829358665,
+    "estimated_full_shards_per_hour": 117.67806819775737,
     "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": 1.3863541229984548,
-      "bytes": 10474994,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2548714610020397,
-      "bytes": 2378266,
+      "seconds": 1.07342461799999,
+      "bytes": 5972281,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0437052470006165,
-      "bytes": 6102982,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.201682383001753,
-      "bytes": 8258960,
-      "pages": 32,
+      "seconds": 1.2648427659987647,
+      "bytes": 7742666,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.4597207000006165,
-      "bytes": 10914765,
-      "pages": 43,
+      "seconds": 1.1688129319991276,
+      "bytes": 5401350,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.293525116998353,
-      "bytes": 5848497,
-      "pages": 28,
+      "seconds": 1.2462927969972952,
+      "bytes": 6148998,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.0607272609995562,
-      "bytes": 2667610,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.1124823669997568,
+      "bytes": 8182028,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.5769093039998552,
-      "bytes": 5962126,
-      "pages": 29,
+      "seconds": 1.1143527310014179,
+      "bytes": 6164100,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0608517170003324,
-      "bytes": 4607080,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.3281419349987118,
+      "bytes": 11547797,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.4511087869977928,
-      "bytes": 13183448,
-      "pages": 39,
-      "shards": 4
+      "seconds": 0.8457415819975722,
+      "bytes": 905737,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.3023088439986168,
-      "bytes": 6313993,
+      "seconds": 1.6407027059976826,
+      "bytes": 6090920,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0238163379981415,
-      "bytes": 2362328,
-      "pages": 29,
+      "seconds": 1.0669106919995102,
+      "bytes": 2526136,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.3659524169997894,
-      "bytes": 10315345,
-      "pages": 40,
+      "seconds": 1.477937416999339,
+      "bytes": 16111480,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.0596161510002275,
-      "bytes": 5840926,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.2388989300015965,
+      "bytes": 7565588,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9319052459977684,
-      "bytes": 2836860,
-      "pages": 38,
+      "seconds": 0.9064518010018219,
+      "bytes": 1123131,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0333158499997808,
+      "bytes": 5705052,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.3332925520007848,
-      "bytes": 16563668,
-      "pages": 36,
+      "seconds": 1.4537645320015145,
+      "bytes": 12446484,
+      "pages": 45,
       "shards": 4
     },
     {
-      "seconds": 1.482665261002694,
-      "bytes": 2527538,
+      "seconds": 1.027312460999383,
+      "bytes": 1095320,
       "pages": 33,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.213385584000207,
-      "bytes": 6122336,
+      "seconds": 1.1021007999988797,
+      "bytes": 11824886,
       "pages": 33,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0116616440027428,
+      "bytes": 2398524,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3804073209976195,
-      "bytes": 6371391,
+      "seconds": 1.1044605290007894,
+      "bytes": 9770514,
       "pages": 37,
-      "shards": 2
+      "shards": 3
+    },
+    {
+      "seconds": 1.1610946500004502,
+      "bytes": 7427357,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.1654375459984294,
-      "bytes": 2648981,
+      "seconds": 1.3267378080017806,
+      "bytes": 1199966,
       "pages": 36,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.4223995709980954,
-      "bytes": 9800135,
-      "pages": 33,
+      "seconds": 1.69540837599925,
+      "bytes": 9973657,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.239405350999732,
-      "bytes": 13306132,
-      "pages": 35,
+      "seconds": 1.1572326720015553,
+      "bytes": 13287961,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.4411270890013839,
-      "bytes": 1188235,
+      "seconds": 0.7943182929993782,
+      "bytes": 1190422,
       "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 1.200595816000714,
-      "bytes": 10503607,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.1334048999997322,
+      "bytes": 5665943,
+      "pages": 19,
+      "shards": 2
     },
     {
-      "seconds": 1.0193097230003332,
-      "bytes": 2594227,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1813513000015519,
+      "bytes": 7573359,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.1054332999992766,
-      "bytes": 5601825,
-      "pages": 19,
+      "seconds": 1.4761009860012564,
+      "bytes": 5975966,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0161830090000876,
-      "bytes": 1229232,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.119130936000147,
+      "bytes": 7848522,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.4130291219989886,
-      "bytes": 17178237,
-      "pages": 40,
-      "shards": 5
+      "seconds": 1.0408579100003408,
+      "bytes": 6882526,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.2973765460010327,
-      "bytes": 5824332,
-      "pages": 28,
+      "seconds": 1.0372045380026975,
+      "bytes": 6042585,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.3192006099998252,
-      "bytes": 2683646,
+      "seconds": 1.5961815709997609,
+      "bytes": 6520796,
       "pages": 38,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.3172165699979814,
-      "bytes": 5876805,
+      "seconds": 0.9840364880001289,
+      "bytes": 6376962,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2478048169978138,
-      "bytes": 10304832,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.2160047779980232,
+      "bytes": 7548487,
+      "pages": 31,
+      "shards": 4
     },
     {
-      "seconds": 0.9087454849977803,
-      "bytes": 2381627,
-      "pages": 29,
+      "seconds": 1.0057136780014844,
+      "bytes": 5529362,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.198183002001315,
-      "bytes": 11915570,
-      "pages": 35,
+      "seconds": 1.374610462000419,
+      "bytes": 11892118,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.2237985769970692,
-      "bytes": 7820030,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.2482218100012688,
+      "bytes": 2427794,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.4434572989994194,
-      "bytes": 5751342,
-      "pages": 32,
+      "seconds": 1.1882967479978106,
+      "bytes": 6156483,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.5010855370019272,
-      "bytes": 6022741,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.4572524590003013,
+      "bytes": 10153840,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0192723729996942,
-      "bytes": 2390933,
-      "pages": 30,
+      "seconds": 0.8581832199997734,
+      "bytes": 2650389,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0420342719990003,
-      "bytes": 6189478,
+      "seconds": 1.6456713849984226,
+      "bytes": 6071875,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2739770709995355,
-      "bytes": 15491147,
-      "pages": 33,
+      "seconds": 1.52896361799867,
+      "bytes": 13388189,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 1.0505857130010554,
-      "bytes": 2636972,
+      "seconds": 1.3202310080014286,
+      "bytes": 5863529,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 1.5651336049995734,
+      "bytes": 6044120,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.07342461799999,
-      "bytes": 5972281,
+      "seconds": 1.1207633359990723,
+      "bytes": 2404167,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2648427659987647,
-      "bytes": 7742666,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.4597645020003256,
+      "bytes": 6256363,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1688129319991276,
-      "bytes": 5401350,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.2800013509986456,
+      "bytes": 667279,
+      "pages": 18,
+      "shards": 1
     },
     {
-      "seconds": 1.2462927969972952,
-      "bytes": 6148998,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.4021912890020758,
+      "bytes": 11567150,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.1124823669997568,
-      "bytes": 8182028,
-      "pages": 32,
+      "seconds": 1.280043183000089,
+      "bytes": 12206708,
+      "pages": 48,
       "shards": 3
     },
     {
-      "seconds": 1.1143527310014179,
-      "bytes": 6164100,
-      "pages": 37,
+      "seconds": 0.8664777250014595,
+      "bytes": 2384518,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.3281419349987118,
-      "bytes": 11547797,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.027337190000253,
+      "bytes": 5987324,
+      "pages": 24,
+      "shards": 2
     },
     {
-      "seconds": 0.8457415819975722,
-      "bytes": 905737,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.2340771259987378,
+      "bytes": 10655993,
+      "pages": 48,
+      "shards": 3
     },
     {
-      "seconds": 1.6407027059976826,
-      "bytes": 6090920,
-      "pages": 36,
+      "seconds": 1.3521600980020594,
+      "bytes": 2290704,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0669106919995102,
-      "bytes": 2526136,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.766581726995355,
+      "bytes": 12101436,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.477937416999339,
-      "bytes": 16111480,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.0664826619977248,
+      "bytes": 6396203,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.2388989300015965,
-      "bytes": 7565588,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.070789342003991,
+      "bytes": 6236244,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.9064518010018219,
-      "bytes": 1123131,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.5840121689980151,
+      "bytes": 6547762,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0333158499997808,
-      "bytes": 5705052,
-      "pages": 24,
+      "seconds": 1.2987593140060198,
+      "bytes": 2393188,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.4537645320015145,
-      "bytes": 12446484,
-      "pages": 45,
-      "shards": 4
+      "seconds": 1.808516741002677,
+      "bytes": 10401411,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.027312460999383,
-      "bytes": 1095320,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1967187990012462,
+      "bytes": 13416312,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.1021007999988797,
-      "bytes": 11824886,
-      "pages": 33,
-      "shards": 3
+      "seconds": 0.995916217994818,
+      "bytes": 1003302,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.0116616440027428,
-      "bytes": 2398524,
-      "pages": 30,
+      "seconds": 0.9674175760010257,
+      "bytes": 6098431,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1044605290007894,
-      "bytes": 9770514,
-      "pages": 37,
+      "seconds": 1.0489661490064464,
+      "bytes": 6157279,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.1610946500004502,
-      "bytes": 7427357,
+      "seconds": 1.4413105299972813,
+      "bytes": 6425996,
       "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3267378080017806,
-      "bytes": 1199966,
-      "pages": 36,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.69540837599925,
-      "bytes": 9973657,
-      "pages": 31,
+      "seconds": 1.128546699997969,
+      "bytes": 5952550,
+      "pages": 28,
       "shards": 3
-    },
-    {
-      "seconds": 1.1572326720015553,
-      "bytes": 13287961,
-      "pages": 34,
-      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3557acc152a7dfbc059cc3cb10ac7d0ee8716631de59030b922ca4ad23b76a00
-size 721778687
+oid sha256:def47c4f3c010dce6d7624c71b7cffcba72fe3c499691cd19b43ed26143caae4
+size 721786943