Qwen90M step 410000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 405000,
+  "step": 410000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "971a9eeb4c3937f1740c96e56a608fda61840fb0477a14bea2f2a2fb4b6f0d33",
+    "model.safetensors": "6c6f9da68cb642941532a5ace8e8fec462efe7e94eb3fbc9d86a49411d6d84d5",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "6dc22034db3d8da70156cf3616a8261ff7bce22ec2f1aaee4adaf8904f0c628d",
+    "trainer.pt": "2683a5571b60dc0874cc06273234cc7f53125ac1227bf84bc08e5b65eefaa4bc",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18",
-    "shard_scheduler.json": "8466082ba94a8cec381b73bc76af24a887c21c79f8a3381f5bbf4c50f884eaa8",
-    "coverage.json": "869317a753fc4e9b828a15de2e9ed753b83822bd9b2bdfac8fc380952e35f060",
+    "shard_scheduler.json": "dcd8a9cb4cda262678652187e892a303cca733af30618ef17353af63ba53c942",
+    "coverage.json": "551c8b6a0d7c9ec77f8c7a79a16a6436443df8715ce046c0113917bd6f5dc3f4",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 405000,
+  "step": 410000,
   "stats": {
-    "climbmix/blocks": 14580000,
-    "climbmix/targets": 24373460669,
-    "climbmix/input_tokens": 24375004213,
-    "ultra_qa/blocks": 5467500,
-    "ultra_qa/targets": 8919175872,
-    "ultra_qa/input_tokens": 8919193035,
-    "cortex/blocks": 5467500,
-    "cortex/targets": 207967668,
-    "cortex/input_tokens": 8691301141,
-    "ultra_style/blocks": 5467500,
-    "ultra_style/targets": 8648596018,
-    "ultra_style/input_tokens": 8649049645,
-    "openmath/blocks": 3645000,
-    "openmath/targets": 5632655091,
-    "openmath/input_tokens": 5634532171,
-    "rewrite6/blocks": 1822500,
-    "rewrite6/targets": 2987706438,
-    "rewrite6/input_tokens": 2990982166
+    "climbmix/blocks": 14760000,
+    "climbmix/targets": 24673977216,
+    "climbmix/input_tokens": 24675540036,
+    "ultra_qa/blocks": 5535000,
+    "ultra_qa/targets": 9029200380,
+    "ultra_qa/input_tokens": 9029217755,
+    "cortex/blocks": 5535000,
+    "cortex/targets": 210523889,
+    "cortex/input_tokens": 8798719668,
+    "ultra_style/blocks": 5535000,
+    "ultra_style/targets": 8755304288,
+    "ultra_style/input_tokens": 8755763253,
+    "openmath/blocks": 3690000,
+    "openmath/targets": 5702199677,
+    "openmath/input_tokens": 5704099937,
+    "rewrite6/blocks": 1845000,
+    "rewrite6/targets": 3024643728,
+    "rewrite6/input_tokens": 3027959756
   },
   "sources": {
     "climbmix": {
-      "page": 1996,
-      "row": 2155,
+      "page": 2021,
+      "row": 1375,
       "start": {
-        "file": 96,
+        "file": 97,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 52,
+          "row_group": 69,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 8177771
+      "documents": 8279391
     },
     "rewrite6": {
-      "page": 36,
-      "row": 435,
+      "page": 50,
+      "row": 510,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 147456
+          "row": 204800
         }
       },
       "epoch": 22,
-      "offset": 2048,
-      "documents": 4666779
+      "offset": 0,
+      "documents": 4724198
     },
     "openmath": {
-      "page": 1247,
-      "row": 3991,
+      "page": 1263,
+      "row": 1659,
       "start": {
         "file": 11,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 304,
-          "row": 562
+          "row_group": 370,
+          "row": 98
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 5111703
+      "documents": 5174907
     },
     "ultra_qa": {
-      "page": 539,
-      "row": 2341,
+      "page": 546,
+      "row": 940,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 48
+          "row": 28720
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2210085
+      "documents": 2237356
     },
     "ultra_style": {
-      "page": 823,
-      "row": 2388,
+      "page": 833,
+      "row": 2801,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 35437
+          "row": 76397
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 3373396
+      "documents": 3414769
     },
     "cortex": {
-      "page": 106043,
-      "row": 21,
+      "page": 107353,
+      "row": 26,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 13573525
+      "documents": 13741210
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 405000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "971a9eeb4c3937f1740c96e56a608fda61840fb0477a14bea2f2a2fb4b6f0d33", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "6dc22034db3d8da70156cf3616a8261ff7bce22ec2f1aaee4adaf8904f0c628d", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "8466082ba94a8cec381b73bc76af24a887c21c79f8a3381f5bbf4c50f884eaa8", "coverage.json": "869317a753fc4e9b828a15de2e9ed753b83822bd9b2bdfac8fc380952e35f060", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000405000"}
+{"step": 410000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "6c6f9da68cb642941532a5ace8e8fec462efe7e94eb3fbc9d86a49411d6d84d5", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "2683a5571b60dc0874cc06273234cc7f53125ac1227bf84bc08e5b65eefaa4bc", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "dcd8a9cb4cda262678652187e892a303cca733af30618ef17353af63ba53c942", "coverage.json": "551c8b6a0d7c9ec77f8c7a79a16a6436443df8715ce046c0113917bd6f5dc3f4", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000410000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:971a9eeb4c3937f1740c96e56a608fda61840fb0477a14bea2f2a2fb4b6f0d33
+oid sha256:6c6f9da68cb642941532a5ace8e8fec462efe7e94eb3fbc9d86a49411d6d84d5
 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": 76765.55632890557,
-      "required_pages_per_second_upper_estimate": 0.005310819307464605
+      "required_token_positions_per_second": 76595.74770061958,
+      "required_pages_per_second_upper_estimate": 0.005299071552549478
     },
     {
       "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": 9595.694541113196,
-      "required_pages_per_second_upper_estimate": 0.004017516923476006
+      "required_token_positions_per_second": 9574.468462577448,
+      "required_pages_per_second_upper_estimate": 0.004008630007644012
     },
     {
       "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": 19191.389082226393,
-      "required_pages_per_second_upper_estimate": 0.003997060234863171
+      "required_token_positions_per_second": 19148.936925154896,
+      "required_pages_per_second_upper_estimate": 0.0039882185700838456
     },
     {
       "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": 28787.08362333959,
-      "required_pages_per_second_upper_estimate": 0.0017257417078876118
+      "required_token_positions_per_second": 28723.405387732342,
+      "required_pages_per_second_upper_estimate": 0.0017219242948939932
     },
     {
       "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": 28787.08362333959,
-      "required_pages_per_second_upper_estimate": 0.0022380691307155136
+      "required_token_positions_per_second": 28723.405387732342,
+      "required_pages_per_second_upper_estimate": 0.002233118427988008
     },
     {
       "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": 28787.08362333959,
-      "required_pages_per_second_upper_estimate": 0.3692497995579789
+      "required_token_positions_per_second": 28723.405387732342,
+      "required_pages_per_second_upper_estimate": 0.3684330035239715
     }
   ],
   "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": 704566,
-    "producer_batch_seconds": 0.06708011378908907,
-    "producer_p95_seconds": 0.1769912210002076,
-    "gpu_update_seconds": 0.9604307390688211,
-    "producer_batches_per_second": 14.907547759149079,
-    "gpu_batches_per_second": 1.0411994944784284,
-    "producer_headroom": 14.3176671120233,
+    "producer_batch_seconds": 0.09953904679677805,
+    "producer_p95_seconds": 0.22590240799945605,
+    "gpu_update_seconds": 0.9625599620513087,
+    "producer_batches_per_second": 10.046308782136828,
+    "gpu_batches_per_second": 1.0388963175539765,
+    "producer_headroom": 9.670174600089354,
     "sustainable": true,
-    "buffer_seconds": 60.507136561335734,
+    "buffer_seconds": 60.641277609232446,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 83,
+    "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 432985.9895226683,
-    "upload_bytes_per_second": 4974423.660555432,
+    "generation_bytes_per_second": 352991.2228158892,
+    "upload_bytes_per_second": 5469736.43334308,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 92.90871395359075,
+    "estimated_full_shards_per_hour": 75.74369920117863,
     "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
   },
   "upload_measurements": [
     {
-      "seconds": 0.8249086619999844,
-      "bytes": 105910,
-      "pages": 3,
-      "shards": 1
+      "seconds": 1.0353907819999222,
+      "bytes": 2349590,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.08548185799998,
-      "bytes": 2013983,
-      "pages": 18,
+      "seconds": 1.3265869809993092,
+      "bytes": 11233813,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.0607301400000324,
-      "bytes": 873319,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4327509930000133,
-      "bytes": 9634876,
-      "pages": 38,
+      "seconds": 1.9781427679999979,
+      "bytes": 8201204,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0578138549999494,
-      "bytes": 7567895,
-      "pages": 37,
-      "shards": 3
+      "seconds": 0.9253323299999465,
+      "bytes": 6468324,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8248641469999711,
-      "bytes": 1159359,
-      "pages": 34,
+      "seconds": 1.0003412309997657,
+      "bytes": 1020572,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 0.9606663819999994,
-      "bytes": 10228939,
-      "pages": 23,
-      "shards": 3
+      "seconds": 1.2060352290000083,
+      "bytes": 11046742,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.4356544900001609,
-      "bytes": 13537180,
-      "pages": 46,
-      "shards": 4
+      "seconds": 1.0752750450001258,
+      "bytes": 6242403,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 0.9362600690001273,
-      "bytes": 1025985,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.8482705019996502,
+      "bytes": 2583323,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.7697618709999006,
-      "bytes": 7336838,
-      "pages": 30,
+      "seconds": 1.3519573450003008,
+      "bytes": 11517949,
+      "pages": 22,
       "shards": 3
     },
     {
-      "seconds": 1.264844782000182,
-      "bytes": 9996183,
-      "pages": 40,
+      "seconds": 1.325917474000562,
+      "bytes": 11191562,
+      "pages": 43,
       "shards": 3
     },
     {
-      "seconds": 0.9787746179999885,
-      "bytes": 1132423,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1523795839993909,
+      "bytes": 2573171,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.8866051049999442,
-      "bytes": 7112636,
-      "pages": 23,
-      "shards": 3
+      "seconds": 1.718842794000011,
+      "bytes": 5798754,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.117540433999693,
-      "bytes": 7055686,
-      "pages": 40,
-      "shards": 2
+      "seconds": 1.0535618110006908,
+      "bytes": 1088694,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.0181373959999291,
-      "bytes": 10411775,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.3156409709999934,
+      "bytes": 11593993,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.183017940999889,
-      "bytes": 7545162,
-      "pages": 32,
+      "seconds": 2.0578567419997853,
+      "bytes": 11618530,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9269013600001017,
-      "bytes": 1114116,
+      "seconds": 1.2497664069996972,
+      "bytes": 1138477,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.037127201000203,
-      "bytes": 7612192,
-      "pages": 37,
+      "seconds": 1.3184527379999054,
+      "bytes": 7724848,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.9605875500001275,
-      "bytes": 5330342,
-      "pages": 25,
+      "seconds": 0.9524561609996454,
+      "bytes": 5747185,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2280645959999674,
-      "bytes": 12018597,
-      "pages": 45,
+      "seconds": 1.0773643330003324,
+      "bytes": 7156662,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.1235430140000062,
-      "bytes": 7362856,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.3278628770003706,
+      "bytes": 6286061,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8067839350001123,
-      "bytes": 1162221,
-      "pages": 35,
-      "shards": 1
+      "seconds": 0.9090305769996121,
+      "bytes": 7087614,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.3463661700002376,
-      "bytes": 11675037,
-      "pages": 32,
+      "seconds": 1.3239770070003942,
+      "bytes": 12583619,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 0.8115306640002018,
-      "bytes": 1206731,
-      "pages": 34,
+      "seconds": 0.9289323200000581,
+      "bytes": 883918,
+      "pages": 25,
+      "shards": 2
+    },
+    {
+      "seconds": 1.464534527000069,
+      "bytes": 7593844,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0184807959994941,
+      "bytes": 301114,
+      "pages": 8,
       "shards": 1
     },
     {
-      "seconds": 1.1671833450000122,
-      "bytes": 5654442,
-      "pages": 19,
+      "seconds": 1.0933929989987519,
+      "bytes": 5793673,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1527475369998683,
-      "bytes": 8325445,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0935950399998546,
+      "bytes": 4381259,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.049984759000381,
-      "bytes": 5490081,
-      "pages": 21,
+      "seconds": 1.3099932099994476,
+      "bytes": 13324819,
+      "pages": 42,
+      "shards": 4
+    },
+    {
+      "seconds": 0.9060626449991105,
+      "bytes": 6034341,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.3486151890001565,
-      "bytes": 10508121,
-      "pages": 37,
+      "seconds": 1.1673664749996533,
+      "bytes": 6863070,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.566957560999981,
-      "bytes": 2658357,
-      "pages": 37,
+      "seconds": 2.7079190219992597,
+      "bytes": 5909001,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1752439570000206,
-      "bytes": 5996670,
+      "seconds": 1.1004524960007984,
+      "bytes": 6145895,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1166780060002566,
-      "bytes": 4727222,
-      "pages": 30,
+      "seconds": 0.9040275779989315,
+      "bytes": 2336556,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 2.7728241000004346,
-      "bytes": 13625242,
-      "pages": 41,
+      "seconds": 1.34454566200111,
+      "bytes": 16421464,
+      "pages": 42,
       "shards": 4
     },
     {
-      "seconds": 3.518534920000093,
-      "bytes": 6664316,
-      "pages": 36,
+      "seconds": 0.9113021489993116,
+      "bytes": 908789,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 0.963736620000418,
+      "bytes": 7684006,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9569276100000934,
-      "bytes": 2410711,
-      "pages": 28,
+      "seconds": 1.1074004389993206,
+      "bytes": 6213956,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1522871959996337,
-      "bytes": 11018272,
-      "pages": 39,
+      "seconds": 1.0808812039995246,
+      "bytes": 6459806,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 3.0842820179996124,
-      "bytes": 6556297,
-      "pages": 37,
+      "seconds": 1.1259296549997089,
+      "bytes": 5894764,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 3.363522853000177,
-      "bytes": 2328587,
-      "pages": 27,
+      "seconds": 0.93207276099929,
+      "bytes": 6831941,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.091435188000105,
-      "bytes": 12252514,
-      "pages": 41,
+      "seconds": 1.0320983010005875,
+      "bytes": 7553870,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0064982160001819,
-      "bytes": 6233391,
-      "pages": 31,
+      "seconds": 1.168341065000277,
+      "bytes": 9926928,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9982179830003588,
-      "bytes": 6432855,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8547535799989419,
+      "bytes": 1160518,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 2.2747711609999897,
-      "bytes": 6302843,
-      "pages": 37,
-      "shards": 2
+      "seconds": 0.9846142869992036,
+      "bytes": 7709639,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.0353907819999222,
-      "bytes": 2349590,
-      "pages": 28,
+      "seconds": 1.0955427600001713,
+      "bytes": 6199869,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3265869809993092,
-      "bytes": 11233813,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.9781427679999979,
-      "bytes": 8201204,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.6065525880003406,
+      "bytes": 12580162,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 0.9253323299999465,
-      "bytes": 6468324,
-      "pages": 36,
+      "seconds": 0.995714869000949,
+      "bytes": 5809499,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.0003412309997657,
-      "bytes": 1020572,
-      "pages": 29,
+      "seconds": 0.7974183579990495,
+      "bytes": 1155225,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.2060352290000083,
-      "bytes": 11046742,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0752750450001258,
-      "bytes": 6242403,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.29576844300027,
+      "bytes": 7786718,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.8482705019996502,
-      "bytes": 2583323,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0071225869996852,
+      "bytes": 672738,
+      "pages": 21,
+      "shards": 1
     },
     {
-      "seconds": 1.3519573450003008,
-      "bytes": 11517949,
-      "pages": 22,
-      "shards": 3
+      "seconds": 1.3492830719987978,
+      "bytes": 12380695,
+      "pages": 48,
+      "shards": 4
     },
     {
-      "seconds": 1.325917474000562,
-      "bytes": 11191562,
-      "pages": 43,
+      "seconds": 1.1991475859995262,
+      "bytes": 11620692,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.1523795839993909,
-      "bytes": 2573171,
+      "seconds": 0.9693775400010054,
+      "bytes": 6039517,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.718842794000011,
-      "bytes": 5798754,
-      "pages": 27,
+      "seconds": 0.8906193579987303,
+      "bytes": 2228503,
+      "pages": 22,
       "shards": 2
     },
     {
-      "seconds": 1.0535618110006908,
-      "bytes": 1088694,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.1276412609986437,
+      "bytes": 11020284,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.3156409709999934,
-      "bytes": 11593993,
+      "seconds": 0.8274413319995801,
+      "bytes": 1262070,
       "pages": 35,
-      "shards": 4
+      "shards": 1
     },
     {
-      "seconds": 2.0578567419997853,
-      "bytes": 11618530,
-      "pages": 30,
+      "seconds": 1.0886287620014627,
+      "bytes": 7651274,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2497664069996972,
-      "bytes": 1138477,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1375053659994592,
+      "bytes": 11652818,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.3184527379999054,
-      "bytes": 7724848,
-      "pages": 34,
+      "seconds": 2.7600006199991185,
+      "bytes": 6232730,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9524561609996454,
-      "bytes": 5747185,
-      "pages": 38,
+      "seconds": 1.0662739279996458,
+      "bytes": 6304947,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0773643330003324,
-      "bytes": 7156662,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3278628770003706,
-      "bytes": 6286061,
+      "seconds": 1.1643560489992524,
+      "bytes": 6319425,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9090305769996121,
-      "bytes": 7087614,
-      "pages": 37,
+      "seconds": 2.530195315999663,
+      "bytes": 2235467,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.3239770070003942,
-      "bytes": 12583619,
+      "seconds": 1.0548529980005696,
+      "bytes": 10543433,
       "pages": 38,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.9289323200000581,
-      "bytes": 883918,
-      "pages": 25,
-      "shards": 2
+      "seconds": 1.3102050620000227,
+      "bytes": 8381851,
+      "pages": 37,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6dc22034db3d8da70156cf3616a8261ff7bce22ec2f1aaee4adaf8904f0c628d
-size 721764287
+oid sha256:2683a5571b60dc0874cc06273234cc7f53125ac1227bf84bc08e5b65eefaa4bc
+size 721789695