Qwen90M step 390000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 385000,
+  "step": 390000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "2f1fd8a77492c63de0166886d8fade646f0af3c2be8d3d896db134f80a50f80e",
+    "model.safetensors": "3445e25637ff28b766a77ae05bc59ec15b294ce669d092bc5e88156d9adba6ce",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "b7abb72e2e8b959c46cdb1904d9eef803368467bb28b5507e8fe5fed6f8eafc8",
+    "trainer.pt": "62981a06b7db55eac80b0820e301dd2522b6f1ca97872b5e04d9d00e4779099f",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15",
-    "shard_scheduler.json": "e2b097f0d0e81ecf9bee6574b648aca8b7332ccc9611e53ad70065a29ab32a3b",
-    "coverage.json": "48a9ce486999758fcda827d9b250e83153d30945a6565c051cf5e98a5544de4b",
+    "shard_scheduler.json": "a3cecf4f99b4babbedb1f1e008b99c2bf6e52201ec4ee5a9670c2a4034e567de",
+    "coverage.json": "eb0c2309e40705779a35e534d2ee49b824e6be0073c4849921791384d1631876",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 385000,
+  "step": 390000,
   "stats": {
-    "climbmix/blocks": 13860000,
-    "climbmix/targets": 23168651278,
-    "climbmix/input_tokens": 23170119193,
-    "ultra_qa/blocks": 5197500,
-    "ultra_qa/targets": 8479023486,
-    "ultra_qa/input_tokens": 8479039844,
-    "cortex/blocks": 5197500,
-    "cortex/targets": 197711285,
-    "cortex/input_tokens": 8262348867,
-    "ultra_style/blocks": 5197500,
-    "ultra_style/targets": 8221351338,
-    "ultra_style/input_tokens": 8221782488,
-    "openmath/blocks": 3465000,
-    "openmath/targets": 5354364862,
-    "openmath/input_tokens": 5356149156,
-    "rewrite6/blocks": 1732500,
-    "rewrite6/targets": 2840128064,
-    "rewrite6/input_tokens": 2843242116
+    "climbmix/blocks": 14040000,
+    "climbmix/targets": 23469768992,
+    "climbmix/input_tokens": 23471255758,
+    "ultra_qa/blocks": 5265000,
+    "ultra_qa/targets": 8589103233,
+    "ultra_qa/input_tokens": 8589119791,
+    "cortex/blocks": 5265000,
+    "cortex/targets": 200266960,
+    "cortex/input_tokens": 8369485865,
+    "ultra_style/blocks": 5265000,
+    "ultra_style/targets": 8328115889,
+    "ultra_style/input_tokens": 8328552449,
+    "openmath/blocks": 3510000,
+    "openmath/targets": 5423887511,
+    "openmath/input_tokens": 5425694857,
+    "rewrite6/blocks": 1755000,
+    "rewrite6/targets": 2877105386,
+    "rewrite6/input_tokens": 2880259808
   },
   "sources": {
     "climbmix": {
-      "page": 1898,
-      "row": 1930,
+      "page": 1922,
+      "row": 3973,
       "start": {
-        "file": 91,
+        "file": 93,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 71,
+          "row_group": 3,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 7776138
+      "offset": 6144,
+      "documents": 7876485
     },
     "rewrite6": {
-      "page": 30,
-      "row": 44,
+      "page": 44,
+      "row": 214,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 122880
+          "row": 180224
         }
       },
       "epoch": 21,
-      "offset": 6144,
-      "documents": 4436408
+      "offset": 0,
+      "documents": 4493922
     },
     "openmath": {
-      "page": 1186,
-      "row": 1451,
+      "page": 1201,
+      "row": 3048,
       "start": {
         "file": 11,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 54,
-          "row": 706
+          "row_group": 116,
+          "row": 146
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 4859307
+      "documents": 4922344
     },
     "ultra_qa": {
-      "page": 512,
-      "row": 3792,
+      "page": 519,
+      "row": 2369,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 0,
-          "row": 18651
+          "row": 47323
         }
       },
       "epoch": 0,
-      "offset": 6144,
-      "documents": 2100944
+      "offset": 2048,
+      "documents": 2128193
     },
     "ultra_style": {
-      "page": 782,
-      "row": 3875,
+      "page": 793,
+      "row": 45,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 36959
+          "row": 82015
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 3206947
+      "documents": 3248173
     },
     "cortex": {
-      "page": 100809,
-      "row": 66,
+      "page": 102116,
+      "row": 47,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 12903618
+      "documents": 13070895
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 385000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "2f1fd8a77492c63de0166886d8fade646f0af3c2be8d3d896db134f80a50f80e", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b7abb72e2e8b959c46cdb1904d9eef803368467bb28b5507e8fe5fed6f8eafc8", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15", "shard_scheduler.json": "e2b097f0d0e81ecf9bee6574b648aca8b7332ccc9611e53ad70065a29ab32a3b", "coverage.json": "48a9ce486999758fcda827d9b250e83153d30945a6565c051cf5e98a5544de4b", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000385000"}
+{"step": 390000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "3445e25637ff28b766a77ae05bc59ec15b294ce669d092bc5e88156d9adba6ce", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "62981a06b7db55eac80b0820e301dd2522b6f1ca97872b5e04d9d00e4779099f", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15", "shard_scheduler.json": "a3cecf4f99b4babbedb1f1e008b99c2bf6e52201ec4ee5a9670c2a4034e567de", "coverage.json": "eb0c2309e40705779a35e534d2ee49b824e6be0073c4849921791384d1631876", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000390000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f1fd8a77492c63de0166886d8fade646f0af3c2be8d3d896db134f80a50f80e
+oid sha256:3445e25637ff28b766a77ae05bc59ec15b294ce669d092bc5e88156d9adba6ce
 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": 75567.00649204037,
-      "required_pages_per_second_upper_estimate": 0.006549956599903092
+      "required_token_positions_per_second": 75353.21624803833,
+      "required_pages_per_second_upper_estimate": 0.006531425803399414
     },
     {
       "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": 9445.875811505046,
-      "required_pages_per_second_upper_estimate": 0.0037681010896382025
+      "required_token_positions_per_second": 9419.15203100479,
+      "required_pages_per_second_upper_estimate": 0.003757440574040526
     },
     {
       "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": 18891.751623010092,
-      "required_pages_per_second_upper_estimate": 0.004370274254510541
+      "required_token_positions_per_second": 18838.30406200958,
+      "required_pages_per_second_upper_estimate": 0.004357910101918019
     },
     {
       "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": 28337.627434515136,
-      "required_pages_per_second_upper_estimate": 0.0017256128347211128
+      "required_token_positions_per_second": 28257.456093014374,
+      "required_pages_per_second_upper_estimate": 0.0017207308206502374
     },
     {
       "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": 28337.627434515136,
-      "required_pages_per_second_upper_estimate": 0.002181509003482326
+      "required_token_positions_per_second": 28257.456093014374,
+      "required_pages_per_second_upper_estimate": 0.00217533719168512
     },
     {
       "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": 22,
-      "required_token_positions_per_second": 28337.627434515136,
-      "required_pages_per_second_upper_estimate": 0.3518584927985291
+      "required_token_positions_per_second": 28257.456093014374,
+      "required_pages_per_second_upper_estimate": 0.35086303305330935
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 717798,
-    "producer_batch_seconds": 0.12316156856283555,
-    "producer_p95_seconds": 0.3393745300018054,
-    "gpu_update_seconds": 0.9756638964885546,
-    "producer_batches_per_second": 8.119415915767686,
-    "gpu_batches_per_second": 1.0249431219080996,
-    "producer_headroom": 7.921820969589085,
+    "producer_batch_seconds": 0.09481285885919988,
+    "producer_p95_seconds": 0.3795729499979643,
+    "gpu_update_seconds": 0.9784320254799922,
+    "producer_batches_per_second": 10.547092578286579,
+    "gpu_batches_per_second": 1.022043406142013,
+    "producer_headroom": 10.319613154297931,
     "sustainable": true,
-    "buffer_seconds": 60.49116158229038,
+    "buffer_seconds": 60.662785579759515,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 85,
+    "upload_pages_per_commit": 86,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 503469.10088892095,
-    "upload_bytes_per_second": 5051464.40603483,
+    "generation_bytes_per_second": 285047.90833474684,
+    "upload_bytes_per_second": 5287078.859485369,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 108.03274888992998,
+    "estimated_full_shards_per_hour": 61.164645552938495,
     "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": 1.0170475559971237,
-      "bytes": 6061997,
-      "pages": 35,
+      "seconds": 1.0007863089995226,
+      "bytes": 6027863,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9019806840005913,
-      "bytes": 1136025,
-      "pages": 35,
-      "shards": 1
+      "seconds": 0.9187128839985235,
+      "bytes": 2707013,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.5857687079987954,
-      "bytes": 7492427,
-      "pages": 32,
+      "seconds": 0.9819577199996274,
+      "bytes": 10903211,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.18117357400115,
-      "bytes": 5405796,
-      "pages": 19,
-      "shards": 2
+      "seconds": 0.860417497002345,
+      "bytes": 864723,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 2.5150336740007333,
-      "bytes": 12238900,
-      "pages": 44,
+      "seconds": 1.116861483998946,
+      "bytes": 7838954,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 2.105870221999794,
-      "bytes": 7597429,
-      "pages": 38,
+      "seconds": 1.1656662089990277,
+      "bytes": 11880694,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 4.179794984000182,
-      "bytes": 6264644,
-      "pages": 36,
+      "seconds": 1.2683707119977043,
+      "bytes": 2325676,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 0.9406074870021257,
-      "bytes": 762678,
-      "pages": 21,
-      "shards": 1
-    },
-    {
-      "seconds": 1.7635130470007425,
-      "bytes": 11843733,
-      "pages": 46,
-      "shards": 5
-    },
-    {
-      "seconds": 0.9203152180016332,
-      "bytes": 582089,
-      "pages": 18,
-      "shards": 1
-    },
-    {
-      "seconds": 1.208456311996997,
-      "bytes": 13456782,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 0.7929018860013457,
-      "bytes": 738445,
-      "pages": 21,
-      "shards": 1
-    },
-    {
-      "seconds": 1.138339361998078,
-      "bytes": 10270327,
-      "pages": 46,
+      "seconds": 1.4917878320011368,
+      "bytes": 10275338,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.5007521030020143,
-      "bytes": 7605566,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.1128048520004086,
+      "bytes": 6259334,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8186183350007923,
-      "bytes": 971258,
+      "seconds": 1.2836710900010075,
+      "bytes": 2351070,
       "pages": 28,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5642075370014936,
-      "bytes": 10184682,
-      "pages": 31,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.1651681650000683,
-      "bytes": 8287753,
+      "seconds": 1.2001916299996083,
+      "bytes": 10729681,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1439757010011817,
-      "bytes": 6251248,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.0578519230002712,
+      "bytes": 8294494,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.9833170429992606,
-      "bytes": 2688996,
+      "seconds": 1.1691894300020067,
+      "bytes": 6431968,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.4986307230028615,
-      "bytes": 10640144,
+      "seconds": 0.9079114560008748,
+      "bytes": 1141999,
       "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1828810689985403,
+      "bytes": 7222759,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 3.6133480419994157,
-      "bytes": 5871337,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.2303334819989686,
+      "bytes": 10801054,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.06491817099959,
-      "bytes": 2604937,
+      "seconds": 0.9937779290012259,
+      "bytes": 2662341,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2623599489998014,
-      "bytes": 11727417,
-      "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4179154230005224,
-      "bytes": 11130810,
+      "seconds": 1.1579458060004981,
+      "bytes": 11572043,
       "pages": 34,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.9252583010020317,
-      "bytes": 184838,
-      "pages": 5,
+      "seconds": 0.8931559880002169,
+      "bytes": 949478,
+      "pages": 28,
       "shards": 1
     },
     {
-      "seconds": 0.9666055019988562,
-      "bytes": 1234734,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.1832785620026698,
+      "bytes": 12385340,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.0084878679990652,
-      "bytes": 5849421,
-      "pages": 27,
+      "seconds": 1.0688240559975384,
+      "bytes": 6186249,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.7869264700020722,
-      "bytes": 2542185,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.054510723999556,
+      "bytes": 1093266,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.5899441080000543,
-      "bytes": 10502441,
-      "pages": 32,
+      "seconds": 1.209129931001371,
+      "bytes": 7450282,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2662762959989777,
-      "bytes": 11950298,
-      "pages": 38,
-      "shards": 3
+      "seconds": 2.1295198230000096,
+      "bytes": 324019,
+      "pages": 10,
+      "shards": 1
     },
     {
-      "seconds": 1.1647725229995558,
-      "bytes": 2632325,
+      "seconds": 1.2268210219990578,
+      "bytes": 15731128,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 0.8699136540017207,
+      "bytes": 2665528,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.06261416399866,
-      "bytes": 5980251,
-      "pages": 27,
+      "seconds": 1.3528436140004487,
+      "bytes": 6109175,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.218213135998667,
-      "bytes": 11903355,
-      "pages": 42,
+      "seconds": 1.3453164509992348,
+      "bytes": 11716150,
+      "pages": 31,
       "shards": 4
     },
     {
-      "seconds": 1.0320353019997128,
-      "bytes": 1021485,
-      "pages": 27,
+      "seconds": 0.8494216189974395,
+      "bytes": 1181935,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 0.9136627450025117,
-      "bytes": 6235523,
-      "pages": 35,
-      "shards": 2
+      "seconds": 3.5161270839998906,
+      "bytes": 6159420,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.3885049839991552,
-      "bytes": 8289862,
-      "pages": 33,
+      "seconds": 1.2511098670001957,
+      "bytes": 8306730,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0629600510001183,
-      "bytes": 10683433,
-      "pages": 39,
+      "seconds": 1.168453648002469,
+      "bytes": 10300928,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2693092910012638,
-      "bytes": 6119669,
-      "pages": 31,
+      "seconds": 4.622547731996747,
+      "bytes": 6328963,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.0424042099984945,
-      "bytes": 2508707,
+      "seconds": 0.9812373249988013,
+      "bytes": 2474379,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.2800720170016575,
-      "bytes": 6017394,
+      "seconds": 1.035560734999308,
+      "bytes": 6144689,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0338583490010933,
-      "bytes": 6530808,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.3160131810000166,
+      "bytes": 6821427,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.5235254969993548,
-      "bytes": 11921244,
-      "pages": 32,
+      "seconds": 1.3006846259995655,
+      "bytes": 12006518,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0007863089995226,
-      "bytes": 6027863,
-      "pages": 36,
+      "seconds": 0.8710538570012432,
+      "bytes": 5900566,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9187128839985235,
-      "bytes": 2707013,
+      "seconds": 1.0173681279993616,
+      "bytes": 2612410,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9819577199996274,
-      "bytes": 10903211,
+      "seconds": 1.2024259139980131,
+      "bytes": 11156038,
       "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.860417497002345,
-      "bytes": 864723,
-      "pages": 26,
+      "seconds": 0.9720597020022979,
+      "bytes": 1226766,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.116861483998946,
-      "bytes": 7838954,
-      "pages": 38,
+      "seconds": 1.4056276539995451,
+      "bytes": 7428876,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1656662089990277,
-      "bytes": 11880694,
-      "pages": 33,
+      "seconds": 1.1535106260016619,
+      "bytes": 12418046,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2683707119977043,
-      "bytes": 2325676,
-      "pages": 30,
+      "seconds": 0.8009545050008455,
+      "bytes": 2263071,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.4917878320011368,
-      "bytes": 10275338,
-      "pages": 40,
+      "seconds": 2.506807755002228,
+      "bytes": 10223439,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.1128048520004086,
-      "bytes": 6259334,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.890203859002213,
+      "bytes": 916282,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.2836710900010075,
-      "bytes": 2351070,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.0841833719969145,
+      "bytes": 7541614,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.2001916299996083,
-      "bytes": 10729681,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.8043565109983319,
+      "bytes": 540912,
+      "pages": 18,
+      "shards": 1
     },
     {
-      "seconds": 1.0578519230002712,
-      "bytes": 8294494,
-      "pages": 34,
+      "seconds": 1.4004500939990976,
+      "bytes": 16241287,
+      "pages": 47,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0899500639934558,
+      "bytes": 7493063,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1691894300020067,
-      "bytes": 6431968,
+      "seconds": 0.8779368290051934,
+      "bytes": 1248675,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2020556749994284,
+      "bytes": 12102958,
       "pages": 36,
-      "shards": 2
+      "shards": 4
     },
     {
-      "seconds": 0.9079114560008748,
-      "bytes": 1141999,
-      "pages": 35,
+      "seconds": 1.048562693002168,
+      "bytes": 883869,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.1828810689985403,
-      "bytes": 7222759,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.072776924003847,
+      "bytes": 6196482,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.2303334819989686,
-      "bytes": 10801054,
-      "pages": 35,
+      "seconds": 0.9420911959969089,
+      "bytes": 7110283,
+      "pages": 25,
       "shards": 3
     },
     {
-      "seconds": 0.9937779290012259,
-      "bytes": 2662341,
+      "seconds": 1.256377692996466,
+      "bytes": 6950847,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1579458060004981,
-      "bytes": 11572043,
-      "pages": 34,
+      "seconds": 1.3214142249998986,
+      "bytes": 10934190,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.8931559880002169,
-      "bytes": 949478,
+      "seconds": 1.2156976079932065,
+      "bytes": 2375462,
       "pages": 28,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.1832785620026698,
-      "bytes": 12385340,
-      "pages": 36,
-      "shards": 4
+      "seconds": 0.9770713670004625,
+      "bytes": 6199366,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0688240559975384,
-      "bytes": 6186249,
+      "seconds": 1.0861811289942125,
+      "bytes": 7695547,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0140557359991362,
+      "bytes": 5030935,
+      "pages": 25,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2526760300024762,
+      "bytes": 11691054,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9398190409992822,
+      "bytes": 2819009,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.054510723999556,
-      "bytes": 1093266,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1499773260002257,
+      "bytes": 6017503,
+      "pages": 37,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7abb72e2e8b959c46cdb1904d9eef803368467bb28b5507e8fe5fed6f8eafc8
-size 721762175
+oid sha256:62981a06b7db55eac80b0820e301dd2522b6f1ca97872b5e04d9d00e4779099f
+size 721770367