Qwen90M step 225000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 220000,
+  "step": 225000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "498fb63be761dda04e4ae3ce457299bd3d432b3bdea8411d8a28b5ef0b7d194b",
+    "model.safetensors": "8026e56ede4b6be87fe5a0a8df3ee3a51ac084bd9dae00c82e2b3e41fd9a844c",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "796925d7192e7f222c8c88ea2d27a48542f3cbdadf89ca004b7822e1a82d3d36",
+    "trainer.pt": "c3941f290352b013b70ffc29d16e39df4aacbd1a763c6261ae4bbca696edc601",
     "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf",
     "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522",
-    "shard_scheduler.json": "7a64512af661fdaf88001f0c4d3af7a439cbeb5dc84c5e6b0d6fd966ff4d9a43",
-    "coverage.json": "2672288392f9e38924ab0573a5d641849c393b2a73175abb1c4c6167c4492336",
+    "shard_scheduler.json": "2e841a817820d99fd0c9bb30efa85bd8f3246df3a2689387bff17be63f08f2a2",
+    "coverage.json": "2e342a84546df2ccad31948e71307f1c47ca262201e1551430aefb1fb67fb347",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 220000,
+  "step": 225000,
   "stats": {
-    "climbmix/blocks": 7920000,
-    "climbmix/targets": 13238818073,
-    "climbmix/input_tokens": 13239656864,
-    "ultra_qa/blocks": 2970000,
-    "ultra_qa/targets": 4845561521,
-    "ultra_qa/input_tokens": 4845570884,
-    "cortex/blocks": 2970000,
-    "cortex/targets": 113003359,
-    "cortex/input_tokens": 4720874700,
-    "ultra_style/blocks": 2970000,
-    "ultra_style/targets": 4697944483,
-    "ultra_style/input_tokens": 4698190714,
-    "openmath/blocks": 1980000,
-    "openmath/targets": 3059241505,
-    "openmath/input_tokens": 3060261058,
-    "rewrite6/blocks": 990000,
-    "rewrite6/targets": 1622934859,
-    "rewrite6/input_tokens": 1624714314
+    "climbmix/blocks": 8100000,
+    "climbmix/targets": 13540352549,
+    "climbmix/input_tokens": 13541210224,
+    "ultra_qa/blocks": 3037500,
+    "ultra_qa/targets": 4955605158,
+    "ultra_qa/input_tokens": 4955614747,
+    "cortex/blocks": 3037500,
+    "cortex/targets": 115559566,
+    "cortex/input_tokens": 4828157787,
+    "ultra_style/blocks": 3037500,
+    "ultra_style/targets": 4804534476,
+    "ultra_style/input_tokens": 4804786332,
+    "openmath/blocks": 2025000,
+    "openmath/targets": 3128842388,
+    "openmath/input_tokens": 3129885291,
+    "rewrite6/blocks": 1012500,
+    "rewrite6/targets": 1659779143,
+    "rewrite6/input_tokens": 1661599123
   },
   "sources": {
     "climbmix": {
-      "page": 1084,
-      "row": 2938,
+      "page": 1109,
+      "row": 563,
       "start": {
-        "file": 52,
+        "file": 53,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 37,
+          "row_group": 55,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4443002
+      "documents": 4543027
     },
     "rewrite6": {
-      "page": 17,
-      "row": 543,
+      "page": 31,
+      "row": 995,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 69632
+          "row": 126976
         }
       },
       "epoch": 12,
       "offset": 0,
-      "documents": 2535023
+      "documents": 2592819
     },
     "openmath": {
-      "page": 677,
-      "row": 4056,
+      "page": 693,
+      "row": 1819,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 153,
-          "row": 92
+          "row_group": 218,
+          "row": 628
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 2777048
+      "offset": 4096,
+      "documents": 2840347
     },
     "ultra_qa": {
-      "page": 293,
-      "row": 482,
+      "page": 299,
+      "row": 3200,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 31730
+          "row": 56306
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 1200610
+      "offset": 2048,
+      "documents": 1227904
     },
     "ultra_style": {
-      "page": 447,
-      "row": 1254,
+      "page": 457,
+      "row": 2213,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 36808
+          "row": 77768
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1832166
+      "documents": 1874085
     },
     "cortex": {
-      "page": 57598,
-      "row": 126,
+      "page": 58907,
+      "row": 79,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 7372670
+      "documents": 7540175
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 220000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "498fb63be761dda04e4ae3ce457299bd3d432b3bdea8411d8a28b5ef0b7d194b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "796925d7192e7f222c8c88ea2d27a48542f3cbdadf89ca004b7822e1a82d3d36", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522", "shard_scheduler.json": "7a64512af661fdaf88001f0c4d3af7a439cbeb5dc84c5e6b0d6fd966ff4d9a43", "coverage.json": "2672288392f9e38924ab0573a5d641849c393b2a73175abb1c4c6167c4492336", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000220000"}
+{"step": 225000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "8026e56ede4b6be87fe5a0a8df3ee3a51ac084bd9dae00c82e2b3e41fd9a844c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "c3941f290352b013b70ffc29d16e39df4aacbd1a763c6261ae4bbca696edc601", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522", "shard_scheduler.json": "2e841a817820d99fd0c9bb30efa85bd8f3246df3a2689387bff17be63f08f2a2", "coverage.json": "2e342a84546df2ccad31948e71307f1c47ca262201e1551430aefb1fb67fb347", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000225000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:498fb63be761dda04e4ae3ce457299bd3d432b3bdea8411d8a28b5ef0b7d194b
+oid sha256:8026e56ede4b6be87fe5a0a8df3ee3a51ac084bd9dae00c82e2b3e41fd9a844c
 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": 74218.2968781337,
-      "required_pages_per_second_upper_estimate": 0.006433053868842927
+      "required_token_positions_per_second": 74741.43834005215,
+      "required_pages_per_second_upper_estimate": 0.006478398444872104
     },
     {
       "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": 9277.287109766712,
-      "required_pages_per_second_upper_estimate": 0.0037008485358890667
+      "required_token_positions_per_second": 9342.679792506518,
+      "required_pages_per_second_upper_estimate": 0.00372693465474171
     },
     {
       "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": 18554.574219533424,
-      "required_pages_per_second_upper_estimate": 0.0042922741963358395
+      "required_token_positions_per_second": 18685.359585013037,
+      "required_pages_per_second_upper_estimate": 0.00432252908889573
     },
     {
       "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": 27831.861329300136,
-      "required_pages_per_second_upper_estimate": 0.0016948143324631963
+      "required_token_positions_per_second": 28028.03937751956,
+      "required_pages_per_second_upper_estimate": 0.0017067605463330859
     },
     {
       "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": 27831.861329300136,
-      "required_pages_per_second_upper_estimate": 0.002142573728652689
+      "required_token_positions_per_second": 28028.03937751956,
+      "required_pages_per_second_upper_estimate": 0.0021576760578602147
     },
     {
       "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": 27831.861329300136,
-      "required_pages_per_second_upper_estimate": 0.34557857046202534
+      "required_token_positions_per_second": 28028.03937751956,
+      "required_pages_per_second_upper_estimate": 0.34801444525521885
     }
   ],
   "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": 0.3495573840859265,
-    "producer_p95_seconds": 0.33007302099940716,
-    "gpu_update_seconds": 0.993393854362641,
-    "producer_batches_per_second": 2.8607606233664478,
-    "gpu_batches_per_second": 1.0066500770146172,
-    "producer_headroom": 2.8418620220548676,
+    "producer_batch_seconds": 0.11895504780454758,
+    "producer_p95_seconds": 0.35881434900147724,
+    "gpu_update_seconds": 0.9864407434140978,
+    "producer_batches_per_second": 8.406536910002155,
+    "gpu_batches_per_second": 1.0137456372077387,
+    "producer_headroom": 8.292550519040578,
     "sustainable": true,
-    "buffer_seconds": 60.5970251161211,
+    "buffer_seconds": 60.17288534825997,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 81,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 384535.0366229114,
-    "upload_bytes_per_second": 2874108.1431070296,
+    "generation_bytes_per_second": 252760.82975083197,
+    "upload_bytes_per_second": 2930376.808544498,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 82.51226734176166,
+    "estimated_full_shards_per_hour": 54.23659009355277,
     "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": 1.1491692820018216,
-      "bytes": 11269048,
-      "pages": 45,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0205925500013109,
-      "bytes": 2657105,
+      "seconds": 1.1927491759997793,
+      "bytes": 6167341,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0683662000010372,
-      "bytes": 6645182,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0938033080019522,
-      "bytes": 6719573,
-      "pages": 31,
-      "shards": 2
+      "seconds": 0.8386335059985868,
+      "bytes": 1124320,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.3891484439991473,
-      "bytes": 11561860,
-      "pages": 34,
+      "seconds": 34.149081805000606,
+      "bytes": 11274982,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 1.268800834997819,
-      "bytes": 6327049,
+      "seconds": 0.9564227560003928,
+      "bytes": 6333046,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.941673843000899,
-      "bytes": 1008328,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.5801954900016426,
+      "bytes": 8475194,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.119837071000802,
-      "bytes": 7224601,
+      "seconds": 1.4765983690012945,
+      "bytes": 10921563,
       "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.2717098209977848,
-      "bytes": 11010766,
-      "pages": 39,
-      "shards": 3
+      "seconds": 0.8757540010010416,
+      "bytes": 1242501,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.1728442359999462,
-      "bytes": 8097014,
-      "pages": 29,
+      "seconds": 1.202817702996981,
+      "bytes": 7607811,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.3505811920003907,
-      "bytes": 5904996,
-      "pages": 27,
+      "seconds": 1.1735024220033665,
+      "bytes": 5786171,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9558858569980657,
-      "bytes": 1199021,
+      "seconds": 1.022158355001011,
+      "bytes": 1083307,
       "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.896848256001249,
-      "bytes": 11337881,
-      "pages": 42,
-      "shards": 4
+      "seconds": 33.35392404899903,
+      "bytes": 16956987,
+      "pages": 37,
+      "shards": 5
     },
     {
-      "seconds": 1.1246725650016742,
-      "bytes": 878351,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.4096410159982042,
+      "bytes": 6047627,
+      "pages": 45,
+      "shards": 2
     },
     {
-      "seconds": 1.1035908939993533,
-      "bytes": 6360663,
-      "pages": 35,
+      "seconds": 1.516586848996667,
+      "bytes": 2356968,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.9343603500019526,
-      "bytes": 2109863,
-      "pages": 20,
+      "seconds": 1.1747013890017115,
+      "bytes": 10632168,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.3331431920014438,
-      "bytes": 16665888,
-      "pages": 43,
-      "shards": 4
+      "seconds": 0.9600440670001262,
+      "bytes": 1151300,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.675639938999666,
-      "bytes": 7647558,
-      "pages": 37,
+      "seconds": 1.3565457900003821,
+      "bytes": 7973344,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.968150565000542,
-      "bytes": 1141925,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0775915699996403,
-      "bytes": 5849379,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.262600800000655,
+      "bytes": 11885196,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.3228814079993754,
-      "bytes": 11625452,
+      "seconds": 0.9526699870002631,
+      "bytes": 6951757,
       "pages": 35,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.1623286959984398,
-      "bytes": 7130450,
-      "pages": 38,
+      "seconds": 1.125239683999098,
+      "bytes": 6131144,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.6525030769989826,
-      "bytes": 8063307,
-      "pages": 38,
+      "seconds": 0.9603508889995283,
+      "bytes": 1054047,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.33804497500023,
+      "bytes": 7455487,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.0988706719981565,
-      "bytes": 771997,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.2401924379992124,
+      "bytes": 10747728,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.993406634999701,
-      "bytes": 10476570,
-      "pages": 39,
+      "seconds": 1.4399759389998508,
+      "bytes": 8283696,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.8121612659997481,
-      "bytes": 2238306,
-      "pages": 27,
+      "seconds": 1.058994241997425,
+      "bytes": 5472606,
+      "pages": 16,
       "shards": 2
     },
     {
-      "seconds": 1.133708954002941,
-      "bytes": 6148950,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.7334500290016877,
+      "bytes": 861667,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.2174438740003097,
-      "bytes": 5854390,
-      "pages": 27,
+      "seconds": 1.2821004600009474,
+      "bytes": 6367474,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2782237920000625,
-      "bytes": 7965202,
-      "pages": 25,
+      "seconds": 1.168987963003019,
+      "bytes": 6662663,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.4521252480008116,
-      "bytes": 10591031,
-      "pages": 45,
-      "shards": 3
+      "seconds": 3.650391375998879,
+      "bytes": 6163332,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 0.8998503880029602,
-      "bytes": 1160065,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9980511270005081,
+      "bytes": 2638272,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 2.0495857230016554,
-      "bytes": 7463969,
-      "pages": 30,
+      "seconds": 1.6268621890012582,
+      "bytes": 11807082,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.0575754780002171,
-      "bytes": 10536411,
-      "pages": 39,
+      "seconds": 3.1277496100010467,
+      "bytes": 10196563,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 3.8299359349985025,
-      "bytes": 2330234,
-      "pages": 28,
+      "seconds": 0.9736092290004308,
+      "bytes": 2727259,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1309446140003274,
-      "bytes": 12087341,
-      "pages": 41,
-      "shards": 3
+      "seconds": 1.154159282999899,
+      "bytes": 5704653,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 2.7144779629998084,
-      "bytes": 851325,
-      "pages": 25,
+      "seconds": 1.143835795999621,
+      "bytes": 1177702,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.4230423680019157,
-      "bytes": 11644215,
-      "pages": 41,
+      "seconds": 1.4552070569989155,
+      "bytes": 11480592,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.8787968369979353,
-      "bytes": 5636855,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.264268914997956,
+      "bytes": 11538285,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 0.8648194890010927,
-      "bytes": 1161752,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9019468660007988,
+      "bytes": 1029860,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.3708638380012417,
-      "bytes": 7781670,
-      "pages": 39,
+      "seconds": 1.031183188999421,
+      "bytes": 7489997,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 2.350306128999364,
-      "bytes": 11141333,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.171489565000229,
+      "bytes": 5465773,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.3496742860006634,
-      "bytes": 7704664,
-      "pages": 35,
+      "seconds": 1.024480873002176,
+      "bytes": 7386490,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1927491759997793,
-      "bytes": 6167341,
+      "seconds": 1.0989981620004983,
+      "bytes": 6127153,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8386335059985868,
-      "bytes": 1124320,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.2221524419983325,
+      "bytes": 6743871,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 34.149081805000606,
-      "bytes": 11274982,
-      "pages": 35,
+      "seconds": 1.1651974790001987,
+      "bytes": 11511318,
+      "pages": 32,
       "shards": 4
     },
     {
-      "seconds": 0.9564227560003928,
-      "bytes": 6333046,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8884790899974178,
+      "bytes": 1189196,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.5801954900016426,
-      "bytes": 8475194,
+      "seconds": 1.1859263639998971,
+      "bytes": 7316047,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2110801790004189,
+      "bytes": 6933944,
       "pages": 35,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 1.4765983690012945,
-      "bytes": 10921563,
+      "seconds": 1.1756158520001918,
+      "bytes": 5451054,
       "pages": 32,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.8757540010010416,
-      "bytes": 1242501,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.2875092799986305,
+      "bytes": 13122964,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.202817702996981,
-      "bytes": 7607811,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.006686715001706,
+      "bytes": 1247241,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.1735024220033665,
-      "bytes": 5786171,
-      "pages": 28,
+      "seconds": 1.2502870179996535,
+      "bytes": 6125090,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.022158355001011,
-      "bytes": 1083307,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1863440980014275,
+      "bytes": 11209339,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 33.35392404899903,
-      "bytes": 16956987,
-      "pages": 37,
-      "shards": 5
+      "seconds": 0.8518053779989714,
+      "bytes": 1243189,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.4096410159982042,
-      "bytes": 6047627,
-      "pages": 45,
-      "shards": 2
+      "seconds": 1.3384703369993076,
+      "bytes": 7381158,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.516586848996667,
-      "bytes": 2356968,
-      "pages": 27,
+      "seconds": 0.9186418360004609,
+      "bytes": 6519513,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.1747013890017115,
-      "bytes": 10632168,
-      "pages": 34,
+      "seconds": 1.2462770620004449,
+      "bytes": 11030793,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9600440670001262,
-      "bytes": 1151300,
-      "pages": 35,
-      "shards": 1
+      "seconds": 0.8822051489987643,
+      "bytes": 2343640,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.3565457900003821,
-      "bytes": 7973344,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0508419269972364,
+      "bytes": 6132224,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.262600800000655,
-      "bytes": 11885196,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.237680772999738,
+      "bytes": 6534213,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.9526699870002631,
-      "bytes": 6951757,
-      "pages": 35,
+      "seconds": 1.075308318999305,
+      "bytes": 6271877,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.125239683999098,
-      "bytes": 6131144,
-      "pages": 36,
+      "seconds": 1.1617618540003605,
+      "bytes": 5816168,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9603508889995283,
-      "bytes": 1054047,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.0311880699991889,
+      "bytes": 6926958,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.33804497500023,
-      "bytes": 7455487,
-      "pages": 29,
+      "seconds": 1.0975180309978896,
+      "bytes": 7653776,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.2401924379992124,
-      "bytes": 10747728,
-      "pages": 36,
+      "seconds": 0.9772678519984765,
+      "bytes": 5704498,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.346531524999591,
+      "bytes": 7336381,
+      "pages": 29,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:796925d7192e7f222c8c88ea2d27a48542f3cbdadf89ca004b7822e1a82d3d36
-size 721771839
+oid sha256:c3941f290352b013b70ffc29d16e39df4aacbd1a763c6261ae4bbca696edc601
+size 721759807