Qwen90M step 435000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 430000,
+  "step": 435000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "55726e70d1f08038bb9239098dc07428f7dbb69a155603511a8d493d3b3d1bb8",
+    "model.safetensors": "bf19dfbdafa63e001aa9aafbc90b1803081ed56a7003a3405b4bd251b7f12fb6",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "ae57f71e3f57e49a3027c7cc6ec8a729344948717d72a54e4f742c154a0c26a4",
+    "trainer.pt": "ef0d375c6273462fad4461943f3ee97eb93fad33803aa4f396c84603803639c6",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18",
-    "shard_scheduler.json": "ea42642d1a5ca11fe1d3250a4c5ef108721b3514388692e82cb5ad01efa41e05",
-    "coverage.json": "5ff68de276024a1ae2e7817b08e75a238d264d40560ae7e48097661cad1ef6c8",
+    "shard_scheduler.json": "8aeff81a2ed0c0c9f85d391f6a9d59dff3e5b2336e287e0f785392dac9c5cb4d",
+    "coverage.json": "142a67e23a41b166cd9863617f189163392272c3e95892f505038ddd8f6114c9",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 430000,
+  "step": 435000,
   "stats": {
-    "climbmix/blocks": 15480000,
-    "climbmix/targets": 25877789085,
-    "climbmix/input_tokens": 25879428024,
-    "ultra_qa/blocks": 5805000,
-    "ultra_qa/targets": 9469588245,
-    "ultra_qa/input_tokens": 9469606395,
-    "cortex/blocks": 5805000,
-    "cortex/targets": 220770662,
-    "cortex/input_tokens": 9227785254,
-    "ultra_style/blocks": 5805000,
-    "ultra_style/targets": 9181956738,
-    "ultra_style/input_tokens": 9182438095,
-    "openmath/blocks": 3870000,
-    "openmath/targets": 5980155459,
-    "openmath/input_tokens": 5982148264,
-    "rewrite6/blocks": 1935000,
-    "rewrite6/targets": 3172177358,
-    "rewrite6/input_tokens": 3175655197
+    "climbmix/blocks": 15660000,
+    "climbmix/targets": 26178177994,
+    "climbmix/input_tokens": 26179836274,
+    "ultra_qa/blocks": 5872500,
+    "ultra_qa/targets": 9579803583,
+    "ultra_qa/input_tokens": 9579821954,
+    "cortex/blocks": 5872500,
+    "cortex/targets": 223336398,
+    "cortex/input_tokens": 9335114674,
+    "ultra_style/blocks": 5872500,
+    "ultra_style/targets": 9288626928,
+    "ultra_style/input_tokens": 9289113965,
+    "openmath/blocks": 3915000,
+    "openmath/targets": 6049599492,
+    "openmath/input_tokens": 6051615394,
+    "rewrite6/blocks": 1957500,
+    "rewrite6/targets": 3209011309,
+    "rewrite6/input_tokens": 3212529771
   },
   "sources": {
     "climbmix": {
-      "page": 2119,
-      "row": 3598,
+      "page": 2144,
+      "row": 2831,
       "start": {
-        "file": 102,
+        "file": 103,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 47,
+          "row_group": 65,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 8683022
+      "documents": 8784655
     },
     "rewrite6": {
-      "page": 6,
-      "row": 373,
+      "page": 20,
+      "row": 892,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 24576
+          "row": 81920
         }
       },
       "epoch": 24,
-      "offset": 4096,
-      "documents": 4954645
+      "offset": 2048,
+      "documents": 5012508
     },
     "openmath": {
-      "page": 1325,
-      "row": 120,
+      "page": 1340,
+      "row": 1711,
       "start": {
         "file": 12,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 187,
-          "row": 400
+          "row_group": 248,
+          "row": 840
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 5427320
+      "documents": 5490351
     },
     "ultra_qa": {
-      "page": 572,
-      "row": 3580,
+      "page": 579,
+      "row": 2175,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 6126
+          "row": 34798
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 2346492
+      "offset": 2048,
+      "documents": 2373759
     },
     "ultra_style": {
-      "page": 874,
-      "row": 1747,
+      "page": 884,
+      "row": 2600,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 70118
+          "row_group": 2,
+          "row": 25169
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 3581651
+      "offset": 0,
+      "documents": 3623464
     },
     "cortex": {
-      "page": 112589,
-      "row": 61,
+      "page": 113898,
+      "row": 103,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 14411453
+      "documents": 14579047
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 430000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "55726e70d1f08038bb9239098dc07428f7dbb69a155603511a8d493d3b3d1bb8", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ae57f71e3f57e49a3027c7cc6ec8a729344948717d72a54e4f742c154a0c26a4", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "ea42642d1a5ca11fe1d3250a4c5ef108721b3514388692e82cb5ad01efa41e05", "coverage.json": "5ff68de276024a1ae2e7817b08e75a238d264d40560ae7e48097661cad1ef6c8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000430000"}
+{"step": 435000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "bf19dfbdafa63e001aa9aafbc90b1803081ed56a7003a3405b4bd251b7f12fb6", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ef0d375c6273462fad4461943f3ee97eb93fad33803aa4f396c84603803639c6", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "8aeff81a2ed0c0c9f85d391f6a9d59dff3e5b2336e287e0f785392dac9c5cb4d", "coverage.json": "142a67e23a41b166cd9863617f189163392272c3e95892f505038ddd8f6114c9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000435000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:55726e70d1f08038bb9239098dc07428f7dbb69a155603511a8d493d3b3d1bb8
+oid sha256:bf19dfbdafa63e001aa9aafbc90b1803081ed56a7003a3405b4bd251b7f12fb6
 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": 76292.77819935197,
-      "required_pages_per_second_upper_estimate": 0.005278111419465689
+      "required_token_positions_per_second": 77090.63733768974,
+      "required_pages_per_second_upper_estimate": 0.005333309165944155
     },
     {
       "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": 9536.597274918997,
-      "required_pages_per_second_upper_estimate": 0.003992774132211747
+      "required_token_positions_per_second": 9636.329667211217,
+      "required_pages_per_second_upper_estimate": 0.004034530002215322
     },
     {
       "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": 19073.194549837994,
-      "required_pages_per_second_upper_estimate": 0.0039724434307660955
+      "required_token_positions_per_second": 19272.659334422435,
+      "required_pages_per_second_upper_estimate": 0.004013986685154929
     },
     {
       "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": 28609.79182475699,
-      "required_pages_per_second_upper_estimate": 0.0017151133352714869
+      "required_token_positions_per_second": 28908.989001633654,
+      "required_pages_per_second_upper_estimate": 0.0017330497491776062
     },
     {
       "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": 28609.79182475699,
-      "required_pages_per_second_upper_estimate": 0.0022242854731999244
+      "required_token_positions_per_second": 28908.989001633654,
+      "required_pages_per_second_upper_estimate": 0.002247546737672786
     },
     {
       "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": 28609.79182475699,
-      "required_pages_per_second_upper_estimate": 0.3669756907268633
+      "required_token_positions_per_second": 28908.989001633654,
+      "required_pages_per_second_upper_estimate": 0.37081347085893784
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 704566,
-    "producer_batch_seconds": 0.09154915285955667,
-    "producer_p95_seconds": 0.3261282250023214,
-    "gpu_update_seconds": 0.9663824249177262,
-    "producer_batches_per_second": 10.923093974818922,
-    "gpu_batches_per_second": 1.034787030698676,
-    "producer_headroom": 10.555886042989714,
+    "producer_batch_seconds": 0.14530138408622406,
+    "producer_p95_seconds": 0.3642332669987809,
+    "gpu_update_seconds": 0.9563807298289679,
+    "producer_batches_per_second": 6.882246898670867,
+    "gpu_batches_per_second": 1.0456086878484394,
+    "producer_headroom": 6.5820483118139945,
     "sustainable": true,
-    "buffer_seconds": 60.88209276981675,
+    "buffer_seconds": 60.25198597922498,
     "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": 336081.2992223133,
-    "upload_bytes_per_second": 5716296.3033110965,
+    "generation_bytes_per_second": 434186.3837777914,
+    "upload_bytes_per_second": 5398305.177243957,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 72.1152232408719,
+    "estimated_full_shards_per_hour": 93.16629061699206,
     "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": 0.8803229269979056,
-      "bytes": 4831371,
+      "seconds": 1.1164300489981542,
+      "bytes": 9840833,
       "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1803598539991071,
-      "bytes": 6473769,
-      "pages": 37,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.0211126419999346,
-      "bytes": 2557759,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.1696946029987885,
+      "bytes": 13219787,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.475009352001507,
-      "bytes": 6011684,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8621246889997565,
+      "bytes": 1265428,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.40803722399869,
-      "bytes": 10317954,
-      "pages": 35,
+      "seconds": 1.4639245990001655,
+      "bytes": 10067426,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.058006204002595,
-      "bytes": 8327609,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.8352084129983268,
+      "bytes": 2708130,
+      "pages": 39,
+      "shards": 2
     },
     {
-      "seconds": 1.1640828560011869,
-      "bytes": 6135465,
+      "seconds": 0.9732167729998764,
+      "bytes": 6207933,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8637312579994614,
-      "bytes": 2324435,
-      "pages": 28,
+      "seconds": 0.9747246619990619,
+      "bytes": 7297014,
+      "pages": 31,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0759720290006953,
+      "bytes": 6305836,
+      "pages": 19,
       "shards": 2
     },
     {
-      "seconds": 1.1669898459986143,
-      "bytes": 10131671,
-      "pages": 41,
+      "seconds": 1.0651401020004414,
+      "bytes": 11486452,
+      "pages": 47,
       "shards": 3
     },
     {
-      "seconds": 1.0318543029970897,
-      "bytes": 5757385,
-      "pages": 28,
+      "seconds": 1.0763056159994449,
+      "bytes": 2441249,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9449253849998058,
-      "bytes": 2649946,
-      "pages": 37,
+      "seconds": 4.102204398001049,
+      "bytes": 6262944,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2970311710014357,
-      "bytes": 11562430,
-      "pages": 33,
+      "seconds": 1.1364409210000304,
+      "bytes": 7535674,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.4141549599989958,
-      "bytes": 12336426,
-      "pages": 42,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8797184569993988,
-      "bytes": 870056,
-      "pages": 27,
-      "shards": 1
+      "seconds": 0.9002165780002542,
+      "bytes": 5078039,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.3180106779982452,
-      "bytes": 6077154,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1330570440004522,
+      "bytes": 12010212,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 0.8500619599981292,
-      "bytes": 2427201,
-      "pages": 31,
+      "seconds": 1.1789866869985417,
+      "bytes": 2676190,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.7313189029991918,
-      "bytes": 10259217,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.2169118900019384,
+      "bytes": 10000231,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.5030511410004692,
-      "bytes": 13228107,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.287032372001704,
+      "bytes": 6056411,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 0.8756032579985913,
-      "bytes": 1232264,
+      "seconds": 1.0081800109983305,
+      "bytes": 2487942,
       "pages": 34,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.2522269889996096,
-      "bytes": 10696457,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.1110187749982288,
+      "bytes": 6282739,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.1242637430004834,
-      "bytes": 7345030,
+      "seconds": 1.3052810369990766,
+      "bytes": 12708434,
       "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 0.8663610330004303,
-      "bytes": 1077995,
-      "pages": 35,
-      "shards": 1
+      "shards": 4
     },
     {
-      "seconds": 0.9089211409991549,
-      "bytes": 5857565,
-      "pages": 34,
+      "seconds": 1.1760405560016807,
+      "bytes": 6277751,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1926017650002905,
-      "bytes": 13608076,
+      "seconds": 1.038220692997129,
+      "bytes": 6391937,
       "pages": 35,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 0.8606232550009736,
-      "bytes": 4749519,
-      "pages": 26,
+      "seconds": 0.970617956998467,
+      "bytes": 2311552,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.311676024000917,
-      "bytes": 7627867,
-      "pages": 39,
+      "seconds": 1.171514493998984,
+      "bytes": 9915176,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.9741381140011072,
-      "bytes": 986302,
-      "pages": 29,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1853109039984702,
-      "bytes": 6001312,
-      "pages": 33,
-      "shards": 2
-    },
-    {
-      "seconds": 1.160805725998216,
-      "bytes": 12260097,
-      "pages": 42,
-      "shards": 4
+      "seconds": 3.413111547000881,
+      "bytes": 7579762,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9388873019997845,
-      "bytes": 6924644,
+      "seconds": 1.0360288759984542,
+      "bytes": 6839680,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.2151130479978747,
-      "bytes": 5819277,
-      "pages": 31,
+      "seconds": 1.1501934409971,
+      "bytes": 5967653,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1440478279982926,
-      "bytes": 2552467,
+      "seconds": 1.0980839240000932,
+      "bytes": 6693759,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.1846777530008694,
-      "bytes": 10218604,
+      "seconds": 1.258701953003765,
+      "bytes": 6452551,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.1278007279979647,
-      "bytes": 7339680,
-      "pages": 31,
-      "shards": 3
+      "seconds": 3.7520576210008585,
+      "bytes": 2441375,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.694962838999345,
-      "bytes": 1221938,
+      "seconds": 1.106896608005627,
+      "bytes": 5997924,
       "pages": 35,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.4893900960014435,
-      "bytes": 11508997,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.6541905339981895,
+      "bytes": 16581039,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 1.178049504000228,
-      "bytes": 7502276,
-      "pages": 41,
-      "shards": 3
+      "seconds": 1.211611481005093,
+      "bytes": 2609258,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.9951034849982534,
-      "bytes": 5795502,
-      "pages": 28,
+      "seconds": 1.0486429719967418,
+      "bytes": 6039932,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9211728519985627,
-      "bytes": 7602156,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.093474820998381,
+      "bytes": 5981075,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.7880418419990747,
-      "bytes": 1256366,
+      "seconds": 1.4037001169999712,
+      "bytes": 6358641,
       "pages": 36,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.1164300489981542,
-      "bytes": 9840833,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.1129142319987295,
+      "bytes": 6226935,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1696946029987885,
-      "bytes": 13219787,
+      "seconds": 1.6370596409979044,
+      "bytes": 13312554,
       "pages": 36,
-      "shards": 4
+      "shards": 5
     },
     {
-      "seconds": 0.8621246889997565,
-      "bytes": 1265428,
-      "pages": 35,
-      "shards": 1
+      "seconds": 0.9763937130046543,
+      "bytes": 4930611,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.4639245990001655,
-      "bytes": 10067426,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.5403500989996246,
+      "bytes": 6045634,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.8352084129983268,
-      "bytes": 2708130,
-      "pages": 39,
+      "seconds": 0.8855937030020868,
+      "bytes": 2335409,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9732167729998764,
-      "bytes": 6207933,
+      "seconds": 1.0506481749980594,
+      "bytes": 6177700,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9747246619990619,
-      "bytes": 7297014,
-      "pages": 31,
+      "seconds": 1.1661089699991862,
+      "bytes": 11581137,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.0759720290006953,
-      "bytes": 6305836,
-      "pages": 19,
+      "seconds": 1.0408890239996254,
+      "bytes": 7081099,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0651401020004414,
-      "bytes": 11486452,
-      "pages": 47,
+      "seconds": 1.0998482240029261,
+      "bytes": 5731633,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2273955099954037,
+      "bytes": 7571144,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.0763056159994449,
-      "bytes": 2441249,
-      "pages": 29,
+      "seconds": 1.115527177003969,
+      "bytes": 5389343,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 4.102204398001049,
-      "bytes": 6262944,
+      "seconds": 1.2195735799978138,
+      "bytes": 6166768,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1364409210000304,
-      "bytes": 7535674,
-      "pages": 31,
-      "shards": 3
+      "seconds": 0.954921390999516,
+      "bytes": 2330284,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.9002165780002542,
-      "bytes": 5078039,
+      "seconds": 1.4671854679982061,
+      "bytes": 11708465,
       "pages": 38,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.1330570440004522,
-      "bytes": 12010212,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.1728415899997344,
+      "bytes": 11022627,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.1789866869985417,
-      "bytes": 2676190,
-      "pages": 38,
+      "seconds": 0.9499405899987323,
+      "bytes": 1172384,
+      "pages": 36,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1931303429955733,
+      "bytes": 6041594,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2169118900019384,
-      "bytes": 10000231,
-      "pages": 33,
+      "seconds": 1.0443037180011743,
+      "bytes": 7667563,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.287032372001704,
-      "bytes": 6056411,
-      "pages": 32,
+      "seconds": 1.1781920779976645,
+      "bytes": 5828706,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.0081800109983305,
-      "bytes": 2487942,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.4654938949970528,
+      "bytes": 13087295,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.1110187749982288,
-      "bytes": 6282739,
+      "seconds": 1.2222818860027473,
+      "bytes": 6269271,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3052810369990766,
-      "bytes": 12708434,
+      "seconds": 1.2005412010039436,
+      "bytes": 962816,
       "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2577394740001182,
+      "bytes": 11361644,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.1760405560016807,
-      "bytes": 6277751,
-      "pages": 38,
-      "shards": 2
+      "seconds": 0.925975751997612,
+      "bytes": 1221763,
+      "pages": 37,
+      "shards": 1
     },
     {
-      "seconds": 1.038220692997129,
-      "bytes": 6391937,
-      "pages": 35,
+      "seconds": 1.0596703449991765,
+      "bytes": 5842037,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 0.970617956998467,
-      "bytes": 2311552,
-      "pages": 30,
+      "seconds": 1.226734026997292,
+      "bytes": 13551509,
+      "pages": 37,
+      "shards": 4
+    },
+    {
+      "seconds": 1.065682766995451,
+      "bytes": 5277203,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.171514493998984,
-      "bytes": 9915176,
-      "pages": 34,
+      "seconds": 1.4707806820006226,
+      "bytes": 7704301,
+      "pages": 35,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae57f71e3f57e49a3027c7cc6ec8a729344948717d72a54e4f742c154a0c26a4
-size 721760319
+oid sha256:ef0d375c6273462fad4461943f3ee97eb93fad33803aa4f396c84603803639c6
+size 721768511