Qwen90M step 195000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 190000,
+  "step": 195000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "5b29d7dbcb4b6dc66cbd141f4370adef39d9c5bf0d95a0cb23dbf0b3fbf77d71",
+    "model.safetensors": "e7124a2db2f213964f3cd249326d01e111d4002da26c586796a207bae4728dda",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "003d5c36c105e74cd2f29e7718b5727a4a7950ad08dd430bb87b3b0378f6d788",
+    "trainer.pt": "ba31b41d133fff78caf1b22d96d15106f826e2e03648c9dbff5ca60049e9ff2e",
     "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf",
     "performance.json": "d8fc0ab4ba9b75a14bc27c39b9cdaf8c3a3b38cccaa4c038b8370fdf2cd49c08",
-    "shard_scheduler.json": "6e7020fbb929d08abf5c36bb9ec83b6165c6ce2fb599b8042b1e42ee010831ac",
-    "coverage.json": "379675965a16132527a2a833bac7de4bb49e2af467d73478c93a42753679aff0",
+    "shard_scheduler.json": "9382ee0711b8a32284204139fa207dcc6bdbed89a4d1e96d772bcf51dbded624",
+    "coverage.json": "a52c19a6ab0dd4e5feeaf4d905f52d95523ae03ff4b83abba914a7c129d58cfb",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 190000,
+  "step": 195000,
   "stats": {
-    "climbmix/blocks": 6840000,
-    "climbmix/targets": 11431842491,
-    "climbmix/input_tokens": 11432567043,
-    "ultra_qa/blocks": 2565000,
-    "ultra_qa/targets": 4184969180,
-    "ultra_qa/input_tokens": 4184977239,
-    "cortex/blocks": 2565000,
-    "cortex/targets": 97573355,
-    "cortex/input_tokens": 4076931537,
-    "ultra_style/blocks": 2565000,
-    "ultra_style/targets": 4057172458,
-    "ultra_style/input_tokens": 4057385373,
-    "openmath/blocks": 1710000,
-    "openmath/targets": 2642162908,
-    "openmath/input_tokens": 2643043776,
-    "rewrite6/blocks": 855000,
-    "rewrite6/targets": 1401575108,
-    "rewrite6/input_tokens": 1403112022
+    "climbmix/blocks": 7020000,
+    "climbmix/targets": 11732742810,
+    "climbmix/input_tokens": 11733486473,
+    "ultra_qa/blocks": 2632500,
+    "ultra_qa/targets": 4295100300,
+    "ultra_qa/input_tokens": 4295108575,
+    "cortex/blocks": 2632500,
+    "cortex/targets": 100135715,
+    "cortex/input_tokens": 4184383493,
+    "ultra_style/blocks": 2632500,
+    "ultra_style/targets": 4163945713,
+    "ultra_style/input_tokens": 4164164467,
+    "openmath/blocks": 1755000,
+    "openmath/targets": 2711634953,
+    "openmath/input_tokens": 2712538826,
+    "rewrite6/blocks": 877500,
+    "rewrite6/targets": 1438575118,
+    "rewrite6/input_tokens": 1440152299
   },
   "sources": {
     "climbmix": {
-      "page": 937,
-      "row": 1183,
+      "page": 961,
+      "row": 3746,
       "start": {
-        "file": 45,
+        "file": 46,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 24,
+          "row_group": 39,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3839135
+      "documents": 3940002
     },
     "rewrite6": {
-      "page": 33,
-      "row": 359,
+      "page": 47,
+      "row": 360,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 135168
+          "row": 192512
         }
       },
       "epoch": 10,
       "offset": 0,
-      "documents": 2189567
+      "documents": 2246912
     },
     "openmath": {
-      "page": 585,
-      "row": 2658,
+      "page": 601,
+      "row": 98,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 212,
-          "row": 910
+          "row_group": 278,
+          "row": 446
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 2398818
+      "offset": 2048,
+      "documents": 2461794
     },
     "ultra_qa": {
-      "page": 253,
-      "row": 592,
+      "page": 259,
+      "row": 3323,
       "start": {
-        "file": 1,
+        "file": 2,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 8,
-          "row": 410
+          "row_group": 0,
+          "row": 21613
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1036880
+      "documents": 1064187
     },
     "ultra_style": {
-      "page": 386,
-      "row": 1768,
+      "page": 396,
+      "row": 2960,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 40788
+          "row": 81748
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1582824
+      "documents": 1624976
     },
     "cortex": {
-      "page": 49741,
-      "row": 11,
+      "page": 51053,
+      "row": 14,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 6366859
+      "documents": 6534798
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 190000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "5b29d7dbcb4b6dc66cbd141f4370adef39d9c5bf0d95a0cb23dbf0b3fbf77d71", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "003d5c36c105e74cd2f29e7718b5727a4a7950ad08dd430bb87b3b0378f6d788", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "d8fc0ab4ba9b75a14bc27c39b9cdaf8c3a3b38cccaa4c038b8370fdf2cd49c08", "shard_scheduler.json": "6e7020fbb929d08abf5c36bb9ec83b6165c6ce2fb599b8042b1e42ee010831ac", "coverage.json": "379675965a16132527a2a833bac7de4bb49e2af467d73478c93a42753679aff0", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000190000"}
+{"step": 195000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e7124a2db2f213964f3cd249326d01e111d4002da26c586796a207bae4728dda", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ba31b41d133fff78caf1b22d96d15106f826e2e03648c9dbff5ca60049e9ff2e", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "d8fc0ab4ba9b75a14bc27c39b9cdaf8c3a3b38cccaa4c038b8370fdf2cd49c08", "shard_scheduler.json": "9382ee0711b8a32284204139fa207dcc6bdbed89a4d1e96d772bcf51dbded624", "coverage.json": "a52c19a6ab0dd4e5feeaf4d905f52d95523ae03ff4b83abba914a7c129d58cfb", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000195000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5b29d7dbcb4b6dc66cbd141f4370adef39d9c5bf0d95a0cb23dbf0b3fbf77d71
+oid sha256:e7124a2db2f213964f3cd249326d01e111d4002da26c586796a207bae4728dda
 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": 74817.5457905918,
-      "required_pages_per_second_upper_estimate": 0.006484995245792312
+      "required_token_positions_per_second": 74454.29039832999,
+      "required_pages_per_second_upper_estimate": 0.006453509189053433
     },
     {
       "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": 9352.193223823975,
-      "required_pages_per_second_upper_estimate": 0.003730729704732717
+      "required_token_positions_per_second": 9306.786299791249,
+      "required_pages_per_second_upper_estimate": 0.0037126162038420493
     },
     {
       "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": 18704.38644764795,
-      "required_pages_per_second_upper_estimate": 0.0043269306187049715
+      "required_token_positions_per_second": 18613.572599582498,
+      "required_pages_per_second_upper_estimate": 0.004305922433224167
     },
     {
       "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": 28056.579671471925,
-      "required_pages_per_second_upper_estimate": 0.0017084985004954352
+      "required_token_positions_per_second": 27920.35889937375,
+      "required_pages_per_second_upper_estimate": 0.001700203370169813
     },
     {
       "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": 28056.579671471925,
-      "required_pages_per_second_upper_estimate": 0.002159873168693258
+      "required_token_positions_per_second": 27920.35889937375,
+      "required_pages_per_second_upper_estimate": 0.0021493865165739087
     },
     {
       "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": 21,
-      "required_token_positions_per_second": 28056.579671471925,
-      "required_pages_per_second_upper_estimate": 0.3483688201878879
+      "required_token_positions_per_second": 27920.35889937375,
+      "required_pages_per_second_upper_estimate": 0.34667741410002545
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 61,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.15711286308589933,
-    "producer_p95_seconds": 0.3503623950018664,
-    "gpu_update_seconds": 0.9854372957696134,
-    "producer_batches_per_second": 6.364851230884027,
-    "gpu_batches_per_second": 1.0147779105711778,
-    "producer_headroom": 6.272161784938251,
+    "producer_batch_seconds": 0.10249432328106423,
+    "producer_p95_seconds": 0.39124788300250657,
+    "gpu_update_seconds": 0.9902451504883822,
+    "producer_batches_per_second": 9.756637909182132,
+    "gpu_batches_per_second": 1.0098509439877659,
+    "producer_headroom": 9.661463374638716,
     "sustainable": true,
-    "buffer_seconds": 60.11167504194641,
+    "buffer_seconds": 60.40495417979132,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 84,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 374399.8448116255,
-    "upload_bytes_per_second": 4221736.650489915,
+    "generation_bytes_per_second": 328429.3339946892,
+    "upload_bytes_per_second": 4846781.735776416,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 80.33749111424993,
+    "estimated_full_shards_per_hour": 70.47328963165768,
     "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": 1.3773436970004695,
-      "bytes": 2708180,
-      "pages": 37,
+      "seconds": 1.207307631000731,
+      "bytes": 6036390,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.157878524998523,
-      "bytes": 6108198,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.257034783000563,
+      "bytes": 2415905,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.1510198380019574,
-      "bytes": 2279083,
+      "seconds": 1.2379260390007403,
+      "bytes": 10087351,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0402059740008553,
+      "bytes": 6293747,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 19.352378830000816,
-      "bytes": 15868354,
-      "pages": 40,
+      "seconds": 1.1847501439988264,
+      "bytes": 2572744,
+      "pages": 33,
+      "shards": 2
+    },
+    {
+      "seconds": 1.268939194000268,
+      "bytes": 16767391,
+      "pages": 42,
       "shards": 4
     },
     {
-      "seconds": 1.1481948339969676,
-      "bytes": 6679366,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2854500899993582,
+      "bytes": 7334524,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.3883708200010005,
-      "bytes": 2398547,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.0401804770008312,
+      "bytes": 1254870,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.0771264450013405,
-      "bytes": 6009796,
-      "pages": 32,
+      "seconds": 1.428139841002121,
+      "bytes": 6304931,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8690516380011104,
-      "bytes": 4841742,
+      "seconds": 0.987766936996195,
+      "bytes": 2378061,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.4456373529974371,
-      "bytes": 7578448,
-      "pages": 35,
+      "seconds": 1.0273188179999124,
+      "bytes": 9379394,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.2644371399983356,
-      "bytes": 12142497,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.4798621720037772,
+      "bytes": 13649971,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 0.8376236600015545,
-      "bytes": 2410849,
-      "pages": 28,
-      "shards": 3
+      "seconds": 0.8202600639997399,
+      "bytes": 1256174,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.585001501996885,
-      "bytes": 10678100,
-      "pages": 41,
+      "seconds": 1.0791088629994192,
+      "bytes": 10296860,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.8894862470006046,
-      "bytes": 957902,
-      "pages": 28,
-      "shards": 1
+      "seconds": 1.2926067090011202,
+      "bytes": 2391669,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.1445572449993051,
-      "bytes": 7733205,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.033323986994219,
+      "bytes": 6001182,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.1386810379990493,
-      "bytes": 6058898,
-      "pages": 29,
+      "seconds": 3.9172589509980753,
+      "bytes": 5890320,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.2615420420006558,
-      "bytes": 12864855,
-      "pages": 34,
+      "seconds": 1.3426188599987654,
+      "bytes": 12956908,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 0.9415042880027613,
-      "bytes": 5881567,
-      "pages": 28,
+      "seconds": 1.2346246180022717,
+      "bytes": 6315095,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8555704430000333,
-      "bytes": 1226801,
-      "pages": 35,
-      "shards": 1
+      "seconds": 3.684220068003924,
+      "bytes": 2396656,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 0.896007103001466,
-      "bytes": 7569788,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1850955770059954,
+      "bytes": 6553717,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1828385400003754,
-      "bytes": 9643641,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.714512132006348,
+      "bytes": 6167274,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 2.2667923209992296,
-      "bytes": 913952,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.2202410780009814,
+      "bytes": 6552255,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.163448106002761,
-      "bytes": 13378237,
-      "pages": 41,
-      "shards": 4
+      "seconds": 1.1842839309974806,
+      "bytes": 12073153,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.9504098300021724,
-      "bytes": 828896,
-      "pages": 27,
+      "seconds": 0.9938246039964724,
+      "bytes": 235889,
+      "pages": 8,
       "shards": 1
     },
     {
-      "seconds": 1.1922220630003721,
-      "bytes": 10349010,
-      "pages": 34,
+      "seconds": 1.3448229619971244,
+      "bytes": 7303500,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9792584309980157,
-      "bytes": 2231089,
-      "pages": 24,
+      "seconds": 1.1307224229967687,
+      "bytes": 5856127,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3213498320001236,
-      "bytes": 5953362,
-      "pages": 36,
+      "seconds": 1.2856477119930787,
+      "bytes": 6096134,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9344346500001848,
-      "bytes": 5832415,
-      "pages": 28,
+      "seconds": 1.163583995999943,
+      "bytes": 2620539,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.198589940999227,
-      "bytes": 12036533,
-      "pages": 42,
+      "seconds": 1.2240531869974802,
+      "bytes": 11650354,
+      "pages": 31,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2925304770033108,
+      "bytes": 12240192,
+      "pages": 41,
       "shards": 4
     },
     {
-      "seconds": 0.8354435640030715,
-      "bytes": 5772648,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.9206129219965078,
+      "bytes": 852268,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.2891920199981541,
-      "bytes": 7891753,
-      "pages": 40,
+      "seconds": 1.2395337650013971,
+      "bytes": 6262507,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9012639180000406,
-      "bytes": 5714275,
-      "pages": 30,
+      "seconds": 1.0192943180009024,
+      "bytes": 2301578,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.098721825001121,
-      "bytes": 5875494,
-      "pages": 37,
-      "shards": 2
+      "seconds": 2.074865057998977,
+      "bytes": 9907182,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.1078113340008713,
-      "bytes": 2283092,
+      "seconds": 1.498775358995772,
+      "bytes": 5765034,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9954401819995837,
-      "bytes": 6010708,
+      "seconds": 1.3661099450037,
+      "bytes": 8604864,
+      "pages": 41,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0984054020009353,
+      "bytes": 6085035,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1992983269992692,
-      "bytes": 13571269,
-      "pages": 40,
+      "seconds": 1.4394773240055656,
+      "bytes": 12207803,
+      "pages": 42,
       "shards": 4
     },
     {
-      "seconds": 1.1989007740012312,
-      "bytes": 5658429,
-      "pages": 39,
-      "shards": 2
-    },
-    {
-      "seconds": 1.222567180000624,
-      "bytes": 5708293,
+      "seconds": 0.8468366760062054,
+      "bytes": 871307,
       "pages": 27,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.0141177349978534,
-      "bytes": 2660846,
+      "seconds": 1.2028667299964582,
+      "bytes": 6129254,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.5187353149995033,
-      "bytes": 5879969,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.6012624290015083,
+      "bytes": 8291413,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.1908320559996355,
-      "bytes": 10104452,
-      "pages": 27,
+      "seconds": 1.3447834969992982,
+      "bytes": 10108218,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.954254078998929,
-      "bytes": 8371700,
-      "pages": 38,
+      "seconds": 1.1404380909953034,
+      "bytes": 7821336,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.207307631000731,
-      "bytes": 6036390,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2159917039971333,
+      "bytes": 870322,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 1.257034783000563,
-      "bytes": 2415905,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.047054365000804,
+      "bytes": 6014250,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.2379260390007403,
-      "bytes": 10087351,
-      "pages": 39,
+      "seconds": 1.1608161260010093,
+      "bytes": 6519078,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.0402059740008553,
-      "bytes": 6293747,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1847501439988264,
-      "bytes": 2572744,
-      "pages": 33,
+      "seconds": 1.5200698819971876,
+      "bytes": 6082599,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.268939194000268,
-      "bytes": 16767391,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.4744428199992399,
+      "bytes": 11695751,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.2854500899993582,
-      "bytes": 7334524,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.3584106329944916,
+      "bytes": 2483287,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0401804770008312,
-      "bytes": 1254870,
+      "seconds": 1.1772192700009327,
+      "bytes": 10757429,
       "pages": 34,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.428139841002121,
-      "bytes": 6304931,
-      "pages": 37,
+      "seconds": 2.1233109069944476,
+      "bytes": 2398655,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.987766936996195,
-      "bytes": 2378061,
-      "pages": 29,
+      "seconds": 1.0890797859974555,
+      "bytes": 6225655,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.0273188179999124,
-      "bytes": 9379394,
-      "pages": 33,
-      "shards": 3
+      "seconds": 3.324964125000406,
+      "bytes": 5816918,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.4798621720037772,
-      "bytes": 13649971,
+      "seconds": 1.304179352002393,
+      "bytes": 11867177,
       "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 0.8202600639997399,
-      "bytes": 1256174,
-      "pages": 35,
+      "seconds": 1.3306422289970214,
+      "bytes": 6277468,
+      "pages": 38,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9982111080025788,
+      "bytes": 877422,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.0791088629994192,
-      "bytes": 10296860,
-      "pages": 35,
+      "seconds": 1.6349642249988392,
+      "bytes": 7479267,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.2926067090011202,
-      "bytes": 2391669,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.1783448380010668,
+      "bytes": 10185058,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.033323986994219,
-      "bytes": 6001182,
+      "seconds": 0.9609743260007235,
+      "bytes": 2529461,
       "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 3.9172589509980753,
-      "bytes": 5890320,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3426188599987654,
-      "bytes": 12956908,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.5562968439990073,
+      "bytes": 11835464,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.2346246180022717,
-      "bytes": 6315095,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8645425069989869,
+      "bytes": 693988,
+      "pages": 20,
+      "shards": 1
     },
     {
-      "seconds": 3.684220068003924,
-      "bytes": 2396656,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.3442884890027926,
+      "bytes": 12581350,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.1850955770059954,
-      "bytes": 6553717,
+      "seconds": 1.4273276979947696,
+      "bytes": 6232998,
       "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.714512132006348,
-      "bytes": 6167274,
-      "pages": 33,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2202410780009814,
-      "bytes": 6552255,
-      "pages": 33,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:003d5c36c105e74cd2f29e7718b5727a4a7950ad08dd430bb87b3b0378f6d788
-size 721770879
+oid sha256:ba31b41d133fff78caf1b22d96d15106f826e2e03648c9dbff5ca60049e9ff2e
+size 721763199