Qwen90M step 790000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 785000,
+  "step": 790000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "9e9efa160bc5d9cdcfdc2ca5084b0841b10448f0013b892e4c614e0c0ff5502f",
+    "model.safetensors": "b05fbd69ca1e5cb4b7930464cb55d832e7fabd9331fc804eec985815869e02f6",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "da99d6bb0dd16c05dc0dcdcb4b5f86f1c1ca2eda0fbea8eec1ea8b1c34c0fc81",
+    "trainer.pt": "18b3061530ed9f14c62220eb7527367935619170cb668dc55e2f24995d4b01f2",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d",
-    "shard_scheduler.json": "70243b4426905888cedd0ae2bc3df6ea021387e3c4d1a15ff1b998d804d4625b",
-    "coverage.json": "2153e069f9b927c191b3d524d2beb884e93a0e0cf00c2acbc8c4eeeff4ac0602",
+    "shard_scheduler.json": "95b736366b05cdfc5520adebf2ae74583e9002f0f04d9c9db9b72e67289036a4",
+    "coverage.json": "2800913198ff53a2f8080d5558a4a2cd6bc636c3de0909e2fdff62c6a7b10ddc",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 785000,
+  "step": 790000,
   "stats": {
-    "climbmix/blocks": 28260000,
-    "climbmix/targets": 47241568234,
-    "climbmix/input_tokens": 47244562296,
-    "ultra_qa/blocks": 10597500,
-    "ultra_qa/targets": 17288525662,
-    "ultra_qa/input_tokens": 17288558958,
-    "cortex/blocks": 10597500,
-    "cortex/targets": 402969786,
-    "cortex/input_tokens": 16845487108,
-    "ultra_style/blocks": 10597500,
-    "ultra_style/targets": 16762084445,
-    "ultra_style/input_tokens": 16762964029,
-    "openmath/blocks": 7065000,
-    "openmath/targets": 10916644933,
-    "openmath/input_tokens": 10920282460,
-    "rewrite6/blocks": 3532500,
-    "rewrite6/targets": 5791040560,
-    "rewrite6/input_tokens": 5797389432
+    "climbmix/blocks": 28440000,
+    "climbmix/targets": 47542214483,
+    "climbmix/input_tokens": 47545227669,
+    "ultra_qa/blocks": 10665000,
+    "ultra_qa/targets": 17398813765,
+    "ultra_qa/input_tokens": 17398847287,
+    "cortex/blocks": 10665000,
+    "cortex/targets": 405531388,
+    "cortex/input_tokens": 16952860333,
+    "ultra_style/blocks": 10665000,
+    "ultra_style/targets": 16868950389,
+    "ultra_style/input_tokens": 16869835048,
+    "openmath/blocks": 7110000,
+    "openmath/targets": 10986171006,
+    "openmath/input_tokens": 10989831506,
+    "rewrite6/blocks": 3555000,
+    "rewrite6/targets": 5827916318,
+    "rewrite6/input_tokens": 5834305854
   },
   "sources": {
     "climbmix": {
-      "page": 3870,
-      "row": 2266,
+      "page": 3895,
+      "row": 1369,
       "start": {
-        "file": 187,
+        "file": 188,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 27,
+          "row_group": 44,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 15853786
+      "offset": 0,
+      "documents": 15955289
     },
     "rewrite6": {
-      "page": 1,
-      "row": 3122,
+      "page": 15,
+      "row": 3595,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 4096
+          "row": 61440
         }
       },
       "epoch": 44,
       "offset": 0,
-      "documents": 9044994
+      "documents": 9102811
     },
     "openmath": {
-      "page": 2418,
-      "row": 3723,
+      "page": 2434,
+      "row": 966,
       "start": {
         "file": 22,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 297,
-          "row": 828
+          "row_group": 363,
+          "row": 364
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 9907851
+      "offset": 2048,
+      "documents": 9970630
     },
     "ultra_qa": {
-      "page": 1045,
-      "row": 3516,
+      "page": 1052,
+      "row": 2140,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 58788
+          "row_group": 2,
+          "row": 22927
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 4283836
+      "offset": 2048,
+      "documents": 4311132
     },
     "ultra_style": {
-      "page": 1596,
-      "row": 2099,
+      "page": 1606,
+      "row": 1681,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 32666
+          "row": 73626
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6539315
+      "documents": 6579857
     },
     "cortex": {
-      "page": 205528,
-      "row": 6,
+      "page": 206838,
+      "row": 17,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 26307590
+      "documents": 26475281
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 785000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "9e9efa160bc5d9cdcfdc2ca5084b0841b10448f0013b892e4c614e0c0ff5502f", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "da99d6bb0dd16c05dc0dcdcb4b5f86f1c1ca2eda0fbea8eec1ea8b1c34c0fc81", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "70243b4426905888cedd0ae2bc3df6ea021387e3c4d1a15ff1b998d804d4625b", "coverage.json": "2153e069f9b927c191b3d524d2beb884e93a0e0cf00c2acbc8c4eeeff4ac0602", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000785000"}
+{"step": 790000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b05fbd69ca1e5cb4b7930464cb55d832e7fabd9331fc804eec985815869e02f6", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "18b3061530ed9f14c62220eb7527367935619170cb668dc55e2f24995d4b01f2", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "95b736366b05cdfc5520adebf2ae74583e9002f0f04d9c9db9b72e67289036a4", "coverage.json": "2800913198ff53a2f8080d5558a4a2cd6bc636c3de0909e2fdff62c6a7b10ddc", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000790000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e9efa160bc5d9cdcfdc2ca5084b0841b10448f0013b892e4c614e0c0ff5502f
+oid sha256:b05fbd69ca1e5cb4b7930464cb55d832e7fabd9331fc804eec985815869e02f6
 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": 77070.43158003519,
-      "required_pages_per_second_upper_estimate": 0.0053319112847457955
+      "required_token_positions_per_second": 76526.58374068562,
+      "required_pages_per_second_upper_estimate": 0.005294286629318749
     },
     {
       "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": 9633.803947504399,
-      "required_pages_per_second_upper_estimate": 0.004033472536117103
+      "required_token_positions_per_second": 9565.822967585702,
+      "required_pages_per_second_upper_estimate": 0.004005010319429433
     },
     {
       "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": 19267.607895008798,
-      "required_pages_per_second_upper_estimate": 0.004012934603540484
+      "required_token_positions_per_second": 19131.645935171404,
+      "required_pages_per_second_upper_estimate": 0.003984617312864354
     },
     {
       "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": 28901.4118425132,
-      "required_pages_per_second_upper_estimate": 0.001732595510057987
+      "required_token_positions_per_second": 28697.468902757108,
+      "required_pages_per_second_upper_estimate": 0.0017203694422224475
     },
     {
       "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": 28901.4118425132,
-      "required_pages_per_second_upper_estimate": 0.0022469576468795716
+      "required_token_positions_per_second": 28697.468902757108,
+      "required_pages_per_second_upper_estimate": 0.002231101980363724
     },
     {
       "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": 28901.4118425132,
-      "required_pages_per_second_upper_estimate": 0.37071627919746025
+      "required_token_positions_per_second": 28697.468902757108,
+      "required_pages_per_second_upper_estimate": 0.36810031814313704
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.09094359758591963,
-    "producer_p95_seconds": 0.24977909400331555,
-    "gpu_update_seconds": 0.9566314666790962,
-    "producer_batches_per_second": 10.995826276338393,
-    "gpu_batches_per_second": 1.0453346297205295,
-    "producer_headroom": 10.518953418082141,
+    "producer_batch_seconds": 0.12899730724237202,
+    "producer_p95_seconds": 0.2903622269986954,
+    "gpu_update_seconds": 0.9634299141045056,
+    "producer_batches_per_second": 7.752099802526172,
+    "gpu_batches_per_second": 1.0379582213092124,
+    "producer_headroom": 7.468604846877344,
     "sustainable": true,
-    "buffer_seconds": 60.267782400783055,
+    "buffer_seconds": 60.69608458858385,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 75,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 337241.253587939,
-    "upload_bytes_per_second": 4932306.221605968,
+    "generation_bytes_per_second": 339519.02450688067,
+    "upload_bytes_per_second": 5066127.074407039,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 72.36412244537952,
+    "estimated_full_shards_per_hour": 72.85287906079115,
     "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": 1.8041386129989405,
-      "bytes": 11898323,
-      "pages": 40,
+      "seconds": 1.181815096999344,
+      "bytes": 13310041,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.2189035779992992,
-      "bytes": 6954573,
-      "pages": 37,
+      "seconds": 1.608177831996727,
+      "bytes": 5937731,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.5599068500014255,
-      "bytes": 2415324,
-      "pages": 28,
+      "seconds": 1.4116140810001525,
+      "bytes": 4972441,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.996803005997208,
-      "bytes": 6353091,
+      "seconds": 2.1865444229988498,
+      "bytes": 7544891,
       "pages": 36,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1021929470007308,
+      "bytes": 6736997,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.149405411000771,
-      "bytes": 4840926,
-      "pages": 32,
+      "seconds": 1.0818150289996993,
+      "bytes": 2336300,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5049040789999708,
-      "bytes": 13497203,
-      "pages": 34,
+      "seconds": 1.2755690529993444,
+      "bytes": 16950690,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 2.2471259019985155,
-      "bytes": 6049971,
-      "pages": 37,
+      "seconds": 1.0866290939993632,
+      "bytes": 1216935,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1292916710008285,
-      "bytes": 2686677,
-      "pages": 37,
+      "seconds": 1.176057447999483,
+      "bytes": 7472528,
+      "pages": 31,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2658942909984034,
+      "bytes": 6109615,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3082805390004069,
-      "bytes": 10490550,
-      "pages": 32,
+      "seconds": 1.894670263001899,
+      "bytes": 6431784,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.171707494999282,
-      "bytes": 6154958,
+      "seconds": 1.2564982580006472,
+      "bytes": 6256940,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3786574580008164,
-      "bytes": 2401890,
-      "pages": 28,
+      "seconds": 1.1927087240001129,
+      "bytes": 6721845,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.469864372000302,
-      "bytes": 11763636,
-      "pages": 36,
+      "seconds": 1.142808556996897,
+      "bytes": 7320475,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.3772630089988525,
-      "bytes": 6008251,
-      "pages": 31,
+      "seconds": 1.333093295001163,
+      "bytes": 10300175,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.3044562169998244,
-      "bytes": 6256652,
-      "pages": 37,
+      "seconds": 1.2003908199985744,
+      "bytes": 2637875,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3504138789976423,
-      "bytes": 6609036,
-      "pages": 37,
+      "seconds": 2.519623430998763,
+      "bytes": 6025157,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9880936410008871,
-      "bytes": 2162409,
-      "pages": 24,
-      "shards": 2
+      "seconds": 1.0435017019990482,
+      "bytes": 1045686,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.2899976480002806,
-      "bytes": 11585751,
-      "pages": 45,
+      "seconds": 1.3271499310030777,
+      "bytes": 10074712,
+      "pages": 25,
       "shards": 3
     },
     {
-      "seconds": 1.4168642419972457,
-      "bytes": 6615057,
-      "pages": 28,
-      "shards": 3
+      "seconds": 2.2123866580004687,
+      "bytes": 13632293,
+      "pages": 49,
+      "shards": 4
     },
     {
-      "seconds": 1.1043323689991666,
-      "bytes": 7549149,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.2271638120000716,
+      "bytes": 936260,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.1122719240011065,
-      "bytes": 6287535,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1434297949999745,
+      "bytes": 7632762,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.1806445659967721,
-      "bytes": 6303271,
-      "pages": 33,
+      "seconds": 1.268450359999406,
+      "bytes": 10380594,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2656237250012055,
-      "bytes": 5785840,
-      "pages": 28,
+      "seconds": 1.1230385859998933,
+      "bytes": 790073,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2272778379992815,
+      "bytes": 1850276,
+      "pages": 12,
       "shards": 2
     },
     {
-      "seconds": 1.1340223679981136,
-      "bytes": 1097457,
+      "seconds": 1.1599697590027063,
+      "bytes": 11845234,
       "pages": 35,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 3.1717748339979153,
-      "bytes": 13032899,
-      "pages": 32,
-      "shards": 4
+      "seconds": 0.9193904499989003,
+      "bytes": 900332,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.520787964000192,
-      "bytes": 10348218,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.302292963002401,
+      "bytes": 11974442,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.0010391289979452,
-      "bytes": 2627126,
+      "seconds": 1.1877205400014645,
+      "bytes": 6194082,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0633560690002923,
-      "bytes": 6091672,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8510588639983325,
-      "bytes": 1121057,
-      "pages": 36,
+      "seconds": 1.0285773200012045,
+      "bytes": 923291,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.2576819179994345,
-      "bytes": 11872557,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.0239220590010518,
+      "bytes": 7655031,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.5458234039979288,
-      "bytes": 11492771,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.4461447649991896,
+      "bytes": 15647853,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.4007872010006395,
-      "bytes": 2588923,
+      "seconds": 2.7048667290000594,
+      "bytes": 2731260,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1760680010011129,
-      "bytes": 5785195,
-      "pages": 27,
+      "seconds": 1.060634581001068,
+      "bytes": 5780431,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2628126730014628,
-      "bytes": 9956766,
-      "pages": 33,
-      "shards": 3
+      "seconds": 0.836775318999571,
+      "bytes": 1105976,
+      "pages": 32,
+      "shards": 1
     },
     {
-      "seconds": 1.1540077350000502,
-      "bytes": 2614009,
+      "seconds": 1.2929117179992318,
+      "bytes": 11923823,
       "pages": 37,
-      "shards": 2
+      "shards": 4
     },
     {
-      "seconds": 1.9888146960001905,
-      "bytes": 6170248,
-      "pages": 34,
+      "seconds": 1.1343894960009493,
+      "bytes": 6177604,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.596614274996682,
-      "bytes": 11247081,
-      "pages": 38,
+      "seconds": 3.299507163999806,
+      "bytes": 8285983,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.2773266180010978,
-      "bytes": 7212495,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.3943405380014156,
+      "bytes": 6107726,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.216828146003536,
-      "bytes": 6096262,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2131986140011577,
+      "bytes": 10368948,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.38459085499926,
-      "bytes": 2407911,
+      "seconds": 0.9976378190003743,
+      "bytes": 2377649,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.9793150009973033,
-      "bytes": 10539605,
-      "pages": 41,
-      "shards": 3
+      "seconds": 1.268031460000202,
+      "bytes": 6225627,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.917002105001302,
-      "bytes": 895297,
-      "pages": 29,
+      "seconds": 1.058128027998464,
+      "bytes": 603020,
+      "pages": 18,
       "shards": 1
     },
     {
-      "seconds": 1.181815096999344,
-      "bytes": 13310041,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.6013217930012615,
+      "bytes": 18143583,
+      "pages": 46,
+      "shards": 5
     },
     {
-      "seconds": 1.608177831996727,
-      "bytes": 5937731,
-      "pages": 29,
+      "seconds": 1.0924186460033525,
+      "bytes": 6205338,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.4116140810001525,
-      "bytes": 4972441,
-      "pages": 35,
+      "seconds": 1.2924744820011256,
+      "bytes": 2413023,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 2.1865444229988498,
-      "bytes": 7544891,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1021929470007308,
-      "bytes": 6736997,
-      "pages": 37,
+      "seconds": 1.0274238659985713,
+      "bytes": 6279174,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.0818150289996993,
-      "bytes": 2336300,
-      "pages": 28,
+      "seconds": 1.6521351509982196,
+      "bytes": 5784763,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2755690529993444,
-      "bytes": 16950690,
+      "seconds": 1.7041611679997004,
+      "bytes": 7601822,
       "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0866290939993632,
-      "bytes": 1216935,
-      "pages": 36,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.176057447999483,
-      "bytes": 7472528,
-      "pages": 31,
+      "seconds": 1.273706002997642,
+      "bytes": 11750945,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.2658942909984034,
-      "bytes": 6109615,
-      "pages": 36,
+      "seconds": 1.1050900709997222,
+      "bytes": 2567501,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.894670263001899,
-      "bytes": 6431784,
-      "pages": 35,
+      "seconds": 1.2765829149975616,
+      "bytes": 9916034,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.2564982580006472,
-      "bytes": 6256940,
-      "pages": 37,
+      "seconds": 1.591446239999641,
+      "bytes": 5814050,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.1927087240001129,
-      "bytes": 6721845,
-      "pages": 31,
+      "seconds": 1.4009948579987395,
+      "bytes": 2472639,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.142808556996897,
-      "bytes": 7320475,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.2067919649998657,
+      "bytes": 5662165,
+      "pages": 21,
+      "shards": 2
     },
     {
-      "seconds": 1.333093295001163,
-      "bytes": 10300175,
-      "pages": 35,
+      "seconds": 1.381297531999735,
+      "bytes": 11693702,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.2003908199985744,
-      "bytes": 2637875,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.4175854469976912,
+      "bytes": 7625416,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 2.519623430998763,
-      "bytes": 6025157,
-      "pages": 28,
+      "seconds": 1.1141033280000556,
+      "bytes": 6156266,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0435017019990482,
-      "bytes": 1045686,
-      "pages": 30,
-      "shards": 1
+      "seconds": 0.90897045499878,
+      "bytes": 2264477,
+      "pages": 26,
+      "shards": 2
     },
     {
-      "seconds": 1.3271499310030777,
-      "bytes": 10074712,
-      "pages": 25,
+      "seconds": 1.6527253510030278,
+      "bytes": 10383315,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 2.2123866580004687,
-      "bytes": 13632293,
-      "pages": 49,
-      "shards": 4
+      "seconds": 1.1864270160003798,
+      "bytes": 6161998,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.2271638120000716,
-      "bytes": 936260,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.1574043389991857,
+      "bytes": 8408946,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.1434297949999745,
-      "bytes": 7632762,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.128104429000814,
+      "bytes": 5797209,
+      "pages": 26,
+      "shards": 2
     },
     {
-      "seconds": 1.268450359999406,
-      "bytes": 10380594,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.344746707000013,
+      "bytes": 12817447,
+      "pages": 46,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:da99d6bb0dd16c05dc0dcdcb4b5f86f1c1ca2eda0fbea8eec1ea8b1c34c0fc81
-size 721781183
+oid sha256:18b3061530ed9f14c62220eb7527367935619170cb668dc55e2f24995d4b01f2
+size 721770815