Qwen90M step 440000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 435000,
+  "step": 440000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "bf19dfbdafa63e001aa9aafbc90b1803081ed56a7003a3405b4bd251b7f12fb6",
+    "model.safetensors": "fe214f4ab1fb9964cf107eac88ddf367929c1be6f58e91da736242cfa8027d31",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "ef0d375c6273462fad4461943f3ee97eb93fad33803aa4f396c84603803639c6",
+    "trainer.pt": "8bb0a7558188a5dfb0a78d6ae95be1812ae85f92993c239c65c6dc7be7f1be7d",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18",
-    "shard_scheduler.json": "8aeff81a2ed0c0c9f85d391f6a9d59dff3e5b2336e287e0f785392dac9c5cb4d",
-    "coverage.json": "142a67e23a41b166cd9863617f189163392272c3e95892f505038ddd8f6114c9",
+    "shard_scheduler.json": "6724984473eb995a25775ee3318e6a2730c21a7c2160f6311680bbf3c31a3993",
+    "coverage.json": "3bc0c290712aa33916e32c37c9e507b57b125caed3a356f046223639dd1963d8",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 435000,
+  "step": 440000,
   "stats": {
-    "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
+    "climbmix/blocks": 15840000,
+    "climbmix/targets": 26479322066,
+    "climbmix/input_tokens": 26480999183,
+    "ultra_qa/blocks": 5940000,
+    "ultra_qa/targets": 9689929802,
+    "ultra_qa/input_tokens": 9689948398,
+    "cortex/blocks": 5940000,
+    "cortex/targets": 225912648,
+    "cortex/input_tokens": 9442402042,
+    "ultra_style/blocks": 5940000,
+    "ultra_style/targets": 9395030681,
+    "ultra_style/input_tokens": 9395523198,
+    "openmath/blocks": 3960000,
+    "openmath/targets": 6119068769,
+    "openmath/input_tokens": 6121107744,
+    "rewrite6/blocks": 1980000,
+    "rewrite6/targets": 3245883112,
+    "rewrite6/input_tokens": 3249442073
   },
   "sources": {
     "climbmix": {
-      "page": 2144,
-      "row": 2831,
+      "page": 2169,
+      "row": 730,
       "start": {
-        "file": 103,
+        "file": 105,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 65,
+          "row_group": 0,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 8784655
+      "documents": 8884954
     },
     "rewrite6": {
-      "page": 20,
-      "row": 892,
+      "page": 34,
+      "row": 1296,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 81920
+          "row": 139264
         }
       },
       "epoch": 24,
-      "offset": 2048,
-      "documents": 5012508
+      "offset": 0,
+      "documents": 5070256
     },
     "openmath": {
-      "page": 1340,
-      "row": 1711,
+      "page": 1355,
+      "row": 3306,
       "start": {
         "file": 12,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 248,
-          "row": 840
+          "row_group": 310,
+          "row": 280
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 5490351
+      "documents": 5553386
     },
     "ultra_qa": {
-      "page": 579,
-      "row": 2175,
+      "page": 586,
+      "row": 858,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 34798
+          "row": 63470
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 2373759
+      "offset": 4096,
+      "documents": 2401114
     },
     "ultra_style": {
-      "page": 884,
-      "row": 2600,
+      "page": 894,
+      "row": 3292,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 25169
+          "row": 66129
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 3623464
+      "offset": 2048,
+      "documents": 3665116
     },
     "cortex": {
-      "page": 113898,
-      "row": 103,
+      "page": 115207,
+      "row": 41,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 14579047
+      "documents": 14746537
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"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"}
+{"step": 440000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "fe214f4ab1fb9964cf107eac88ddf367929c1be6f58e91da736242cfa8027d31", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "8bb0a7558188a5dfb0a78d6ae95be1812ae85f92993c239c65c6dc7be7f1be7d", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "6724984473eb995a25775ee3318e6a2730c21a7c2160f6311680bbf3c31a3993", "coverage.json": "3bc0c290712aa33916e32c37c9e507b57b125caed3a356f046223639dd1963d8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000440000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bf19dfbdafa63e001aa9aafbc90b1803081ed56a7003a3405b4bd251b7f12fb6
+oid sha256:fe214f4ab1fb9964cf107eac88ddf367929c1be6f58e91da736242cfa8027d31
 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": 77090.63733768974,
-      "required_pages_per_second_upper_estimate": 0.005333309165944155
+      "required_token_positions_per_second": 76924.88308471063,
+      "required_pages_per_second_upper_estimate": 0.005321841902120205
     },
     {
       "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": 9636.329667211217,
-      "required_pages_per_second_upper_estimate": 0.004034530002215322
+      "required_token_positions_per_second": 9615.610385588829,
+      "required_pages_per_second_upper_estimate": 0.004025855271667829
     },
     {
       "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": 19272.659334422435,
-      "required_pages_per_second_upper_estimate": 0.004013986685154929
+      "required_token_positions_per_second": 19231.220771177657,
+      "required_pages_per_second_upper_estimate": 0.004005356125239444
     },
     {
       "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": 28908.989001633654,
-      "required_pages_per_second_upper_estimate": 0.0017330497491776062
+      "required_token_positions_per_second": 28846.831156766486,
+      "required_pages_per_second_upper_estimate": 0.0017293234812873531
     },
     {
       "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": 28908.989001633654,
-      "required_pages_per_second_upper_estimate": 0.002247546737672786
+      "required_token_positions_per_second": 28846.831156766486,
+      "required_pages_per_second_upper_estimate": 0.002242714238637829
     },
     {
       "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": 28908.989001633654,
-      "required_pages_per_second_upper_estimate": 0.37081347085893784
+      "required_token_positions_per_second": 28846.831156766486,
+      "required_pages_per_second_upper_estimate": 0.37001617676487586
     }
   ],
   "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": 0.14530138408622406,
-    "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,
+    "producer_batch_seconds": 0.5073274579607983,
+    "producer_p95_seconds": 0.34248151299834717,
+    "gpu_update_seconds": 0.9584414956966502,
+    "producer_batches_per_second": 1.9711134974233369,
+    "gpu_batches_per_second": 1.0433605019085102,
+    "producer_headroom": 1.8891969686582784,
     "sustainable": true,
-    "buffer_seconds": 60.25198597922498,
+    "buffer_seconds": 60.38181422888896,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
     "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 434186.3837777914,
-    "upload_bytes_per_second": 5398305.177243957,
+    "generation_bytes_per_second": 270603.4737486303,
+    "upload_bytes_per_second": 5111100.11035178,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 93.16629061699206,
+    "estimated_full_shards_per_hour": 58.06520613998587,
     "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": 1.1164300489981542,
-      "bytes": 9840833,
+      "seconds": 0.8855937030020868,
+      "bytes": 2335409,
       "pages": 29,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.1696946029987885,
-      "bytes": 13219787,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.0506481749980594,
+      "bytes": 6177700,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.8621246889997565,
-      "bytes": 1265428,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.1661089699991862,
+      "bytes": 11581137,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.4639245990001655,
-      "bytes": 10067426,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.0408890239996254,
+      "bytes": 7081099,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8352084129983268,
-      "bytes": 2708130,
-      "pages": 39,
+      "seconds": 1.0998482240029261,
+      "bytes": 5731633,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9732167729998764,
-      "bytes": 6207933,
-      "pages": 37,
+      "seconds": 1.2273955099954037,
+      "bytes": 7571144,
+      "pages": 41,
+      "shards": 3
+    },
+    {
+      "seconds": 1.115527177003969,
+      "bytes": 5389343,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 0.9747246619990619,
-      "bytes": 7297014,
-      "pages": 31,
-      "shards": 4
+      "seconds": 1.2195735799978138,
+      "bytes": 6166768,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.0759720290006953,
-      "bytes": 6305836,
-      "pages": 19,
+      "seconds": 0.954921390999516,
+      "bytes": 2330284,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0651401020004414,
-      "bytes": 11486452,
-      "pages": 47,
+      "seconds": 1.4671854679982061,
+      "bytes": 11708465,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0763056159994449,
-      "bytes": 2441249,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.1728415899997344,
+      "bytes": 11022627,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 4.102204398001049,
-      "bytes": 6262944,
-      "pages": 38,
+      "seconds": 0.9499405899987323,
+      "bytes": 1172384,
+      "pages": 36,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1931303429955733,
+      "bytes": 6041594,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1364409210000304,
-      "bytes": 7535674,
-      "pages": 31,
+      "seconds": 1.0443037180011743,
+      "bytes": 7667563,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9002165780002542,
-      "bytes": 5078039,
-      "pages": 38,
+      "seconds": 1.1781920779976645,
+      "bytes": 5828706,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.1330570440004522,
-      "bytes": 12010212,
+      "seconds": 1.4654938949970528,
+      "bytes": 13087295,
       "pages": 32,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 1.1789866869985417,
-      "bytes": 2676190,
+      "seconds": 1.2222818860027473,
+      "bytes": 6269271,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2169118900019384,
-      "bytes": 10000231,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.2005412010039436,
+      "bytes": 962816,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.287032372001704,
-      "bytes": 6056411,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.2577394740001182,
+      "bytes": 11361644,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.0081800109983305,
-      "bytes": 2487942,
-      "pages": 34,
-      "shards": 2
+      "seconds": 0.925975751997612,
+      "bytes": 1221763,
+      "pages": 37,
+      "shards": 1
     },
     {
-      "seconds": 1.1110187749982288,
-      "bytes": 6282739,
-      "pages": 38,
+      "seconds": 1.0596703449991765,
+      "bytes": 5842037,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.3052810369990766,
-      "bytes": 12708434,
-      "pages": 30,
+      "seconds": 1.226734026997292,
+      "bytes": 13551509,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.1760405560016807,
-      "bytes": 6277751,
-      "pages": 38,
+      "seconds": 1.065682766995451,
+      "bytes": 5277203,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.038220692997129,
-      "bytes": 6391937,
+      "seconds": 1.4707806820006226,
+      "bytes": 7704301,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.970617956998467,
-      "bytes": 2311552,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.0301591320021544,
+      "bytes": 1143669,
+      "pages": 32,
+      "shards": 1
     },
     {
-      "seconds": 1.171514493998984,
-      "bytes": 9915176,
-      "pages": 34,
+      "seconds": 2.9000736189991585,
+      "bytes": 10986050,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 3.413111547000881,
-      "bytes": 7579762,
-      "pages": 40,
+      "seconds": 1.3636105609984952,
+      "bytes": 7351661,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.0360288759984542,
-      "bytes": 6839680,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.2681014190020505,
+      "bytes": 6981210,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.1501934409971,
-      "bytes": 5967653,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.0006479639996542,
+      "bytes": 7455931,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0980839240000932,
-      "bytes": 6693759,
+      "seconds": 1.9897420429988415,
+      "bytes": 1194300,
       "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1884745839997777,
+      "bytes": 9577292,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.258701953003765,
-      "bytes": 6452551,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1139246599996113,
+      "bytes": 7438556,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 3.7520576210008585,
-      "bytes": 2441375,
+      "seconds": 2.768625452001288,
+      "bytes": 924425,
       "pages": 28,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.106896608005627,
-      "bytes": 5997924,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.3904317609994905,
+      "bytes": 18015559,
+      "pages": 40,
+      "shards": 5
     },
     {
-      "seconds": 1.6541905339981895,
-      "bytes": 16581039,
-      "pages": 37,
-      "shards": 4
+      "seconds": 0.9437920670025051,
+      "bytes": 911442,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.211611481005093,
-      "bytes": 2609258,
-      "pages": 36,
+      "seconds": 1.0681643999996595,
+      "bytes": 6436647,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.0486429719967418,
-      "bytes": 6039932,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.4995007219986292,
+      "bytes": 7398571,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.093474820998381,
-      "bytes": 5981075,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.022340563002217,
+      "bytes": 921929,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.4037001169999712,
-      "bytes": 6358641,
-      "pages": 36,
+      "seconds": 1.1844563459962956,
+      "bytes": 10179020,
+      "pages": 46,
       "shards": 3
     },
     {
-      "seconds": 1.1129142319987295,
-      "bytes": 6226935,
+      "seconds": 1.3179188790018088,
+      "bytes": 13975639,
       "pages": 37,
-      "shards": 2
+      "shards": 4
     },
     {
-      "seconds": 1.6370596409979044,
-      "bytes": 13312554,
+      "seconds": 1.3881760859949281,
+      "bytes": 920491,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.721910915999615,
+      "bytes": 12247927,
       "pages": 36,
-      "shards": 5
+      "shards": 4
     },
     {
-      "seconds": 0.9763937130046543,
-      "bytes": 4930611,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.029901437999797,
+      "bytes": 1106234,
+      "pages": 32,
+      "shards": 1
     },
     {
-      "seconds": 1.5403500989996246,
-      "bytes": 6045634,
-      "pages": 38,
+      "seconds": 0.968032641998434,
+      "bytes": 6353702,
+      "pages": 39,
       "shards": 2
     },
     {
-      "seconds": 0.8855937030020868,
-      "bytes": 2335409,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.354905072999827,
+      "bytes": 7189345,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0506481749980594,
-      "bytes": 6177700,
-      "pages": 37,
+      "seconds": 1.0060001309975632,
+      "bytes": 6436150,
+      "pages": 22,
       "shards": 2
     },
     {
-      "seconds": 1.1661089699991862,
-      "bytes": 11581137,
-      "pages": 34,
+      "seconds": 2.564948939005262,
+      "bytes": 12088456,
+      "pages": 50,
       "shards": 4
     },
     {
-      "seconds": 1.0408890239996254,
-      "bytes": 7081099,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.9611534259965993,
+      "bytes": 862575,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.0998482240029261,
-      "bytes": 5731633,
-      "pages": 29,
+      "seconds": 1.0102765199990245,
+      "bytes": 6370774,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2273955099954037,
-      "bytes": 7571144,
-      "pages": 41,
-      "shards": 3
+      "seconds": 1.36733477999951,
+      "bytes": 11825285,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.115527177003969,
-      "bytes": 5389343,
-      "pages": 30,
-      "shards": 2
+      "seconds": 0.7963763439984177,
+      "bytes": 1200164,
+      "pages": 36,
+      "shards": 1
     },
     {
-      "seconds": 1.2195735799978138,
-      "bytes": 6166768,
-      "pages": 38,
-      "shards": 2
+      "seconds": 4.367263744999946,
+      "bytes": 11457575,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 0.954921390999516,
-      "bytes": 2330284,
-      "pages": 28,
+      "seconds": 0.8959607119977591,
+      "bytes": 2404108,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.4671854679982061,
-      "bytes": 11708465,
-      "pages": 38,
+      "seconds": 1.3112088260022574,
+      "bytes": 10449012,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1728415899997344,
-      "bytes": 11022627,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.3342199069957132,
+      "bytes": 7580590,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 0.9499405899987323,
-      "bytes": 1172384,
+      "seconds": 1.0893147949973354,
+      "bytes": 1277647,
       "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 1.1931303429955733,
-      "bytes": 6041594,
+      "seconds": 1.0343618879996939,
+      "bytes": 5958933,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0443037180011743,
-      "bytes": 7667563,
-      "pages": 39,
+      "seconds": 1.130363664997276,
+      "bytes": 7892634,
+      "pages": 23,
       "shards": 3
     },
     {
-      "seconds": 1.1781920779976645,
-      "bytes": 5828706,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4654938949970528,
-      "bytes": 13087295,
-      "pages": 32,
+      "seconds": 1.3904960489962832,
+      "bytes": 10986012,
+      "pages": 47,
       "shards": 4
     },
     {
-      "seconds": 1.2222818860027473,
-      "bytes": 6269271,
-      "pages": 38,
+      "seconds": 1.050227845000336,
+      "bytes": 5898529,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2005412010039436,
-      "bytes": 962816,
-      "pages": 30,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2577394740001182,
-      "bytes": 11361644,
-      "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 0.925975751997612,
-      "bytes": 1221763,
-      "pages": 37,
-      "shards": 1
+      "seconds": 0.9206480690045282,
+      "bytes": 2670829,
+      "pages": 39,
+      "shards": 2
     },
     {
-      "seconds": 1.0596703449991765,
-      "bytes": 5842037,
-      "pages": 31,
+      "seconds": 1.0737715180002851,
+      "bytes": 5694268,
+      "pages": 23,
       "shards": 2
     },
     {
-      "seconds": 1.226734026997292,
-      "bytes": 13551509,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.1375886099995114,
+      "bytes": 7292178,
+      "pages": 43,
+      "shards": 3
     },
     {
-      "seconds": 1.065682766995451,
-      "bytes": 5277203,
-      "pages": 35,
+      "seconds": 1.6739402300008805,
+      "bytes": 5561171,
+      "pages": 22,
       "shards": 2
-    },
-    {
-      "seconds": 1.4707806820006226,
-      "bytes": 7704301,
-      "pages": 35,
-      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ef0d375c6273462fad4461943f3ee97eb93fad33803aa4f396c84603803639c6
-size 721768511
+oid sha256:8bb0a7558188a5dfb0a78d6ae95be1812ae85f92993c239c65c6dc7be7f1be7d
+size 721770431