Qwen90M step 470000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 465000,
+  "step": 470000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "a36c516961ecbdb4d135f7aec065aa33ba895b425e41adf7a7130615b0e7c534",
+    "model.safetensors": "e476e5303342f155469c3f56a579382f525ddf8789c7fe638747144789d71895",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "5d3c17f33f3432629d704e2ea2debd635921d926d1bb7e925d71c43aa06fd061",
+    "trainer.pt": "629577df31a18ce0968102b0e3f539c35489cebe6a6ddc046c8ab547af131faa",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "58f8b6f79b191aa11b2bedb4594ca2897b6fdad2419ea7c9210688ac8022e7cb",
-    "shard_scheduler.json": "c18cd8a1a78ebd1e9db154aa776adb8bac1bfd7543f18628291873cd4bfa08f0",
-    "coverage.json": "f26a841a3097414f1b1263238f23d6076c33c6431b047f162e8bf7038c932c5f",
+    "shard_scheduler.json": "0d21f56fc9331c0f8678684332a3f56087d5f2a22aa6813b19d0321679297611",
+    "coverage.json": "e5989284b0b1414d692537fe449a887c54c77be9ca79006a1b7f013bdd941e1d",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 465000,
+  "step": 470000,
   "stats": {
-    "climbmix/blocks": 16740000,
-    "climbmix/targets": 27982214264,
-    "climbmix/input_tokens": 27983987016,
-    "ultra_qa/blocks": 6277500,
-    "ultra_qa/targets": 10240630112,
-    "ultra_qa/input_tokens": 10240649816,
-    "cortex/blocks": 6277500,
-    "cortex/targets": 238761321,
-    "cortex/input_tokens": 9979079341,
-    "ultra_style/blocks": 6277500,
-    "ultra_style/targets": 9929669455,
-    "ultra_style/input_tokens": 9930189433,
-    "openmath/blocks": 4185000,
-    "openmath/targets": 6466843653,
-    "openmath/input_tokens": 6468998513,
-    "rewrite6/blocks": 2092500,
-    "rewrite6/targets": 3430367981,
-    "rewrite6/input_tokens": 3434128818
+    "climbmix/blocks": 16920000,
+    "climbmix/targets": 28282840249,
+    "climbmix/input_tokens": 28284632204,
+    "ultra_qa/blocks": 6345000,
+    "ultra_qa/targets": 10350703632,
+    "ultra_qa/input_tokens": 10350723557,
+    "cortex/blocks": 6345000,
+    "cortex/targets": 241332160,
+    "cortex/input_tokens": 10086409319,
+    "ultra_style/blocks": 6345000,
+    "ultra_style/targets": 10036596659,
+    "ultra_style/input_tokens": 10037122085,
+    "openmath/blocks": 4230000,
+    "openmath/targets": 6536344953,
+    "openmath/input_tokens": 6538523066,
+    "rewrite6/blocks": 2115000,
+    "rewrite6/targets": 3467211486,
+    "rewrite6/input_tokens": 3471013006
   },
   "sources": {
     "climbmix": {
-      "page": 2292,
-      "row": 3018,
+      "page": 2317,
+      "row": 1972,
       "start": {
-        "file": 110,
+        "file": 112,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 77,
+          "row_group": 11,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 9391050
+      "offset": 2048,
+      "documents": 9492404
     },
     "rewrite6": {
-      "page": 4,
-      "row": 985,
+      "page": 18,
+      "row": 1545,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 16384
+          "row": 73728
         }
       },
       "epoch": 26,
       "offset": 2048,
-      "documents": 5357873
+      "documents": 5415777
     },
     "openmath": {
-      "page": 1432,
-      "row": 3738,
+      "page": 1448,
+      "row": 1399,
       "start": {
         "file": 13,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 189,
-          "row": 22
+          "row_group": 254,
+          "row": 558
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 5869210
+      "offset": 0,
+      "documents": 5932407
     },
     "ultra_qa": {
-      "page": 619,
-      "row": 2226,
+      "page": 626,
+      "row": 855,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 4726
+          "row": 33398
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 2537650
+      "offset": 4096,
+      "documents": 2564951
     },
     "ultra_style": {
-      "page": 945,
-      "row": 653,
+      "page": 955,
+      "row": 1084,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 21316
+          "row": 62276
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 3871373
+      "offset": 0,
+      "documents": 3912764
     },
     "cortex": {
-      "page": 121754,
-      "row": 94,
+      "page": 123063,
+      "row": 96,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 15584606
+      "documents": 15752160
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 465000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "a36c516961ecbdb4d135f7aec065aa33ba895b425e41adf7a7130615b0e7c534", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "5d3c17f33f3432629d704e2ea2debd635921d926d1bb7e925d71c43aa06fd061", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "58f8b6f79b191aa11b2bedb4594ca2897b6fdad2419ea7c9210688ac8022e7cb", "shard_scheduler.json": "c18cd8a1a78ebd1e9db154aa776adb8bac1bfd7543f18628291873cd4bfa08f0", "coverage.json": "f26a841a3097414f1b1263238f23d6076c33c6431b047f162e8bf7038c932c5f", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000465000"}
+{"step": 470000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e476e5303342f155469c3f56a579382f525ddf8789c7fe638747144789d71895", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "629577df31a18ce0968102b0e3f539c35489cebe6a6ddc046c8ab547af131faa", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "58f8b6f79b191aa11b2bedb4594ca2897b6fdad2419ea7c9210688ac8022e7cb", "shard_scheduler.json": "0d21f56fc9331c0f8678684332a3f56087d5f2a22aa6813b19d0321679297611", "coverage.json": "e5989284b0b1414d692537fe449a887c54c77be9ca79006a1b7f013bdd941e1d", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000470000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a36c516961ecbdb4d135f7aec065aa33ba895b425e41adf7a7130615b0e7c534
+oid sha256:e476e5303342f155469c3f56a579382f525ddf8789c7fe638747144789d71895
 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": 76285.71196533572,
-      "required_pages_per_second_upper_estimate": 0.005277622561000523
+      "required_token_positions_per_second": 77031.49068797404,
+      "required_pages_per_second_upper_estimate": 0.005329217263477687
     },
     {
       "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": 9535.713995666965,
-      "required_pages_per_second_upper_estimate": 0.003992404321633889
+      "required_token_positions_per_second": 9628.936335996756,
+      "required_pages_per_second_upper_estimate": 0.004031434568826139
     },
     {
       "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": 19071.42799133393,
-      "required_pages_per_second_upper_estimate": 0.00397207550321698
+      "required_token_positions_per_second": 19257.87267199351,
+      "required_pages_per_second_upper_estimate": 0.004010907013321496
     },
     {
       "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": 28607.141987000894,
-      "required_pages_per_second_upper_estimate": 0.0017149544815440768
+      "required_token_positions_per_second": 28886.809007990265,
+      "required_pages_per_second_upper_estimate": 0.0017317200924255743
     },
     {
       "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": 28607.141987000894,
-      "required_pages_per_second_upper_estimate": 0.002224079459969792
+      "required_token_positions_per_second": 28886.809007990265,
+      "required_pages_per_second_upper_estimate": 0.002245822340726494
     },
     {
       "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": 28607.141987000894,
-      "required_pages_per_second_upper_estimate": 0.36694170145330224
+      "required_token_positions_per_second": 28886.809007990265,
+      "required_pages_per_second_upper_estimate": 0.3705289697155022
     }
   ],
   "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": 0.21675075976563107,
-    "producer_p95_seconds": 0.2752362049977819,
-    "gpu_update_seconds": 0.9664719395094858,
-    "producer_batches_per_second": 4.613593978084705,
-    "gpu_batches_per_second": 1.0346911887659467,
-    "producer_headroom": 4.4589091201088085,
+    "producer_batch_seconds": 0.10171361467956785,
+    "producer_p95_seconds": 0.3008065889989666,
+    "gpu_update_seconds": 0.9571150621846947,
+    "producer_batches_per_second": 9.831525535203294,
+    "gpu_batches_per_second": 1.0448064600690923,
+    "producer_headroom": 9.409901173996515,
     "sustainable": true,
-    "buffer_seconds": 60.887732189097605,
+    "buffer_seconds": 60.29824891763576,
     "buffer_drain_seconds": null,
-    "shard_target_mib": 32,
-    "upload_pages_per_commit": 169,
+    "shard_target_mib": 16,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 614227.4902624867,
-    "upload_bytes_per_second": 4817176.597645194,
+    "generation_bytes_per_second": 320066.6025585662,
+    "upload_bytes_per_second": 5515946.4320974825,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 65.89946046307541,
+    "estimated_full_shards_per_hour": 68.6788421398901,
     "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": 1.1430494310006907,
-      "bytes": 10900083,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.3281948670010024,
+      "bytes": 5971602,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.591609420000168,
-      "bytes": 11822162,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.442292184001417,
+      "bytes": 7645104,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 0.9329451240009803,
-      "bytes": 2485686,
-      "pages": 29,
+      "seconds": 1.229033948999131,
+      "bytes": 6359372,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 4.417064211000252,
-      "bytes": 6221614,
-      "pages": 37,
+      "seconds": 0.9523748860010528,
+      "bytes": 6784780,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.0521548180004174,
-      "bytes": 6267595,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.1795219960004033,
+      "bytes": 10882665,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 0.9623302439977124,
-      "bytes": 6111937,
-      "pages": 28,
+      "seconds": 1.0230853959983506,
+      "bytes": 6145347,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1251186529989354,
-      "bytes": 6276648,
-      "pages": 36,
+      "seconds": 1.223426783002651,
+      "bytes": 2322033,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.272974397001235,
-      "bytes": 8064738,
-      "pages": 28,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1465204630003427,
-      "bytes": 10636216,
-      "pages": 42,
+      "seconds": 1.2023178439994808,
+      "bytes": 11142768,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.9501123180016293,
-      "bytes": 1078327,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.175540739997814,
+      "bytes": 6085764,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 3.8270818399978452,
-      "bytes": 7309721,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.1963462499988964,
+      "bytes": 8278514,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.1971988570003305,
-      "bytes": 10569382,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.9674637549978797,
+      "bytes": 6111470,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8849704640015261,
-      "bytes": 2587633,
-      "pages": 35,
+      "seconds": 1.2372247309976956,
+      "bytes": 2619840,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 2.1376686900002824,
-      "bytes": 11886724,
+      "seconds": 2.7946544099977473,
+      "bytes": 9699006,
       "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.1407382979996328,
-      "bytes": 5980776,
-      "pages": 35,
+      "seconds": 1.143838866999431,
+      "bytes": 6205051,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9373651969981438,
-      "bytes": 2246840,
-      "pages": 25,
+      "seconds": 3.739576465002756,
+      "bytes": 2400719,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 2.0274646729994856,
-      "bytes": 10144494,
-      "pages": 44,
-      "shards": 3
+      "seconds": 1.58055241799957,
+      "bytes": 15825794,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 2.5764055170002393,
-      "bytes": 7638759,
+      "seconds": 1.061300241999561,
+      "bytes": 7669220,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.8547688260005089,
-      "bytes": 858247,
-      "pages": 26,
+      "seconds": 1.3649325709993718,
+      "bytes": 1043801,
+      "pages": 32,
       "shards": 1
     },
     {
-      "seconds": 1.4054743929991673,
-      "bytes": 16596361,
+      "seconds": 1.102126538000448,
+      "bytes": 5859413,
       "pages": 32,
-      "shards": 4
-    },
-    {
-      "seconds": 3.107019715000206,
-      "bytes": 2583987,
-      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.155031209000299,
-      "bytes": 6268449,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2403170809993753,
+      "bytes": 7520295,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.8907613930023217,
-      "bytes": 6205142,
+      "seconds": 1.3349149159985245,
+      "bytes": 6151647,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1056236590011395,
-      "bytes": 6648792,
-      "pages": 34,
+      "seconds": 1.3665102609993482,
+      "bytes": 11866902,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.084016504999454,
-      "bytes": 5907243,
-      "pages": 27,
+      "seconds": 1.1544846530014183,
+      "bytes": 2585533,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0631858130000182,
-      "bytes": 6774777,
-      "pages": 30,
+      "seconds": 1.3021024750014476,
+      "bytes": 5911875,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0792735680006444,
-      "bytes": 7390991,
-      "pages": 38,
+      "seconds": 1.0596985519987356,
+      "bytes": 6289450,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.0645652359999076,
-      "bytes": 834051,
-      "pages": 23,
-      "shards": 1
-    },
-    {
-      "seconds": 1.34710112600078,
-      "bytes": 12646439,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.1195497589978913,
+      "bytes": 6194218,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.121954108999489,
-      "bytes": 5958053,
-      "pages": 34,
+      "seconds": 1.2359527980006533,
+      "bytes": 6034339,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.8286497560002317,
-      "bytes": 1212322,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.2140417409973452,
+      "bytes": 13269645,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.1948913660016842,
-      "bytes": 13257800,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.3757806130015524,
+      "bytes": 5911161,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.0794521610005177,
-      "bytes": 9877213,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.0065856220026035,
+      "bytes": 6244496,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.2257234370008518,
-      "bytes": 2680179,
+      "seconds": 0.9367298329998448,
+      "bytes": 2789154,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3023044059991662,
-      "bytes": 5788392,
-      "pages": 27,
+      "seconds": 1.218474695997429,
+      "bytes": 10329696,
+      "pages": 35,
+      "shards": 4
+    },
+    {
+      "seconds": 0.9623371359994053,
+      "bytes": 2372191,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0160050430022238,
-      "bytes": 1159604,
-      "pages": 32,
-      "shards": 1
+      "seconds": 1.2589823859998432,
+      "bytes": 12076255,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.8482585339988873,
-      "bytes": 12054418,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.1457863180003187,
+      "bytes": 6106429,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 2.080120522998186,
-      "bytes": 11568040,
-      "pages": 30,
+      "seconds": 0.9636820249979792,
+      "bytes": 6954429,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9286353159986902,
-      "bytes": 2598650,
-      "pages": 37,
+      "seconds": 1.3229683749996184,
+      "bytes": 6064094,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.0528518130013254,
-      "bytes": 5770251,
-      "pages": 27,
+      "seconds": 0.8354571769996255,
+      "bytes": 2427677,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3281948670010024,
-      "bytes": 5971602,
-      "pages": 35,
+      "seconds": 1.1609949890007556,
+      "bytes": 6207129,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.442292184001417,
-      "bytes": 7645104,
-      "pages": 40,
+      "seconds": 1.1704257509991294,
+      "bytes": 15475495,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 1.229033948999131,
-      "bytes": 6359372,
-      "pages": 38,
+      "seconds": 1.1073925720011175,
+      "bytes": 2592720,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9523748860010528,
-      "bytes": 6784780,
-      "pages": 33,
+      "seconds": 0.9348203119989194,
+      "bytes": 5613969,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1795219960004033,
-      "bytes": 10882665,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.1180898200000229,
+      "bytes": 7687731,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.0230853959983506,
-      "bytes": 6145347,
+      "seconds": 1.1014808299987635,
+      "bytes": 5892940,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.223426783002651,
-      "bytes": 2322033,
-      "pages": 29,
+      "seconds": 0.8753045230005227,
+      "bytes": 5867523,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2023178439994808,
-      "bytes": 11142768,
-      "pages": 40,
+      "seconds": 1.8846877780015348,
+      "bytes": 8419016,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.175540739997814,
-      "bytes": 6085764,
-      "pages": 28,
+      "seconds": 0.9963944320006703,
+      "bytes": 6337215,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1963462499988964,
-      "bytes": 8278514,
-      "pages": 33,
+      "seconds": 1.0649026120008784,
+      "bytes": 9997467,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9674637549978797,
-      "bytes": 6111470,
-      "pages": 36,
+      "seconds": 1.0817448720008542,
+      "bytes": 2541266,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2372247309976956,
-      "bytes": 2619840,
-      "pages": 37,
+      "seconds": 1.4285129730014887,
+      "bytes": 6270013,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 2.7946544099977473,
-      "bytes": 9699006,
-      "pages": 33,
+      "seconds": 1.485799860998668,
+      "bytes": 7350955,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.143838866999431,
-      "bytes": 6205051,
-      "pages": 37,
+      "seconds": 1.3855582350006443,
+      "bytes": 11926342,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1727449809986865,
+      "bytes": 5685962,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 3.739576465002756,
-      "bytes": 2400719,
-      "pages": 29,
+      "seconds": 1.2225738800007093,
+      "bytes": 2613097,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.58055241799957,
-      "bytes": 15825794,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.0333214540005429,
+      "bytes": 6358222,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.061300241999561,
-      "bytes": 7669220,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.2548035990002973,
+      "bytes": 11160297,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.3649325709993718,
-      "bytes": 1043801,
-      "pages": 32,
+      "seconds": 1.0187819040002069,
+      "bytes": 1294212,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.102126538000448,
-      "bytes": 5859413,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.2491005060001044,
+      "bytes": 11541986,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.2403170809993753,
-      "bytes": 7520295,
-      "pages": 35,
+      "seconds": 1.6100285119973705,
+      "bytes": 7404435,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3349149159985245,
-      "bytes": 6151647,
-      "pages": 37,
+      "seconds": 1.136316104002617,
+      "bytes": 5061136,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3665102609993482,
-      "bytes": 11866902,
-      "pages": 32,
+      "seconds": 1.2786440780000703,
+      "bytes": 7516874,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1544846530014183,
-      "bytes": 2585533,
-      "pages": 37,
+      "seconds": 0.9931387870019535,
+      "bytes": 6287159,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3021024750014476,
-      "bytes": 5911875,
-      "pages": 28,
+      "seconds": 0.9191857780024293,
+      "bytes": 859128,
+      "pages": 26,
       "shards": 2
+    },
+    {
+      "seconds": 1.4843631869989622,
+      "bytes": 18084349,
+      "pages": 33,
+      "shards": 5
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5d3c17f33f3432629d704e2ea2debd635921d926d1bb7e925d71c43aa06fd061
-size 721776127
+oid sha256:629577df31a18ce0968102b0e3f539c35489cebe6a6ddc046c8ab547af131faa
+size 721758463