Qwen90M step 570000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 565000,
+  "step": 570000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "95a5b58e2b26fc5a9ee97e43501f78f2f22c159d31689ff3bf684a9cdf3dee89",
+    "model.safetensors": "7d4140ba71fe97e1e82990744e5a0ec1dbc7c23b51309036d4c2230272e12e25",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "f2026403cd8252eb8fbef1272eb9f1d5d4b596cf0488b73b1b78201830ca2e97",
+    "trainer.pt": "a3bc7b8df78537b2475ab2dd3ebc2aba12dbb1b97f2867ccda2914ec9a3004e2",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667",
-    "shard_scheduler.json": "e3f3e619a117b2827f3e52efa7e755833ae25de54c4ea382eb05fbf3b86130d8",
-    "coverage.json": "6d935be2426b023c8381c6f1092c50faa4e567b7afaa6fdf07207b78a26ef97d",
+    "shard_scheduler.json": "01987d1d04369b6d0c27ef5a7d5f76afa9290cfe2789b0c2414b1dca44f86adb",
+    "coverage.json": "ea81a0c5669ee210b85f1ffe803c00bba540b014a73ee1c73b43e81dbc6770b8",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 565000,
+  "step": 570000,
   "stats": {
-    "climbmix/blocks": 20340000,
-    "climbmix/targets": 34000795378,
-    "climbmix/input_tokens": 34002950085,
-    "ultra_qa/blocks": 7627500,
-    "ultra_qa/targets": 12443347733,
-    "ultra_qa/input_tokens": 12443371739,
-    "cortex/blocks": 7627500,
-    "cortex/targets": 290056072,
-    "cortex/input_tokens": 12124570582,
-    "ultra_style/blocks": 7627500,
-    "ultra_style/targets": 12064354329,
-    "ultra_style/input_tokens": 12064986794,
-    "openmath/blocks": 5085000,
-    "openmath/targets": 7857224925,
-    "openmath/input_tokens": 7859841919,
-    "rewrite6/blocks": 2542500,
-    "rewrite6/targets": 4168025058,
-    "rewrite6/input_tokens": 4172594909
+    "climbmix/blocks": 20520000,
+    "climbmix/targets": 34301849164,
+    "climbmix/input_tokens": 34304022931,
+    "ultra_qa/blocks": 7695000,
+    "ultra_qa/targets": 12553627897,
+    "ultra_qa/input_tokens": 12553652133,
+    "cortex/blocks": 7695000,
+    "cortex/targets": 292622737,
+    "cortex/input_tokens": 12231895676,
+    "ultra_style/blocks": 7695000,
+    "ultra_style/targets": 12171363052,
+    "ultra_style/input_tokens": 12172001166,
+    "openmath/blocks": 5130000,
+    "openmath/targets": 7926745772,
+    "openmath/input_tokens": 7929385871,
+    "rewrite6/blocks": 2565000,
+    "rewrite6/targets": 4205011866,
+    "rewrite6/input_tokens": 4209622064
   },
   "sources": {
     "climbmix": {
-      "page": 2785,
-      "row": 3702,
+      "page": 2810,
+      "row": 1988,
       "start": {
-        "file": 134,
+        "file": 135,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 64,
+          "row_group": 82,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 8192,
-      "documents": 11411062
+      "offset": 2048,
+      "documents": 11511748
     },
     "rewrite6": {
-      "page": 34,
-      "row": 3660,
+      "page": 48,
+      "row": 3760,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 139264
+          "row": 196608
         }
       },
       "epoch": 31,
       "offset": 0,
-      "documents": 6510448
+      "documents": 6567892
     },
     "openmath": {
-      "page": 1740,
-      "row": 2763,
+      "page": 1756,
+      "row": 343,
       "start": {
         "file": 16,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 140,
-          "row": 640
+          "row_group": 206,
+          "row": 176
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 7129803
+      "documents": 7192919
     },
     "ultra_qa": {
-      "page": 752,
-      "row": 3016,
+      "page": 759,
+      "row": 1682,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 30558
+          "row": 59230
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3083208
+      "documents": 3110546
     },
     "ultra_style": {
-      "page": 1148,
-      "row": 3198,
+      "page": 1158,
+      "row": 3739,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 6,
-          "row": 81384
+          "row_group": 7,
+          "row": 37605
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 4705406
+      "offset": 0,
+      "documents": 4746907
     },
     "cortex": {
-      "page": 147929,
-      "row": 71,
+      "page": 149240,
+      "row": 26,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 18934983
+      "documents": 19102746
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 565000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "95a5b58e2b26fc5a9ee97e43501f78f2f22c159d31689ff3bf684a9cdf3dee89", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "f2026403cd8252eb8fbef1272eb9f1d5d4b596cf0488b73b1b78201830ca2e97", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667", "shard_scheduler.json": "e3f3e619a117b2827f3e52efa7e755833ae25de54c4ea382eb05fbf3b86130d8", "coverage.json": "6d935be2426b023c8381c6f1092c50faa4e567b7afaa6fdf07207b78a26ef97d", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000565000"}
+{"step": 570000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "7d4140ba71fe97e1e82990744e5a0ec1dbc7c23b51309036d4c2230272e12e25", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a3bc7b8df78537b2475ab2dd3ebc2aba12dbb1b97f2867ccda2914ec9a3004e2", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667", "shard_scheduler.json": "01987d1d04369b6d0c27ef5a7d5f76afa9290cfe2789b0c2414b1dca44f86adb", "coverage.json": "ea81a0c5669ee210b85f1ffe803c00bba540b014a73ee1c73b43e81dbc6770b8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000570000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:95a5b58e2b26fc5a9ee97e43501f78f2f22c159d31689ff3bf684a9cdf3dee89
+oid sha256:7d4140ba71fe97e1e82990744e5a0ec1dbc7c23b51309036d4c2230272e12e25
 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": 77200.01258345156,
-      "required_pages_per_second_upper_estimate": 0.005340875999231493
+      "required_token_positions_per_second": 76299.61100662185,
+      "required_pages_per_second_upper_estimate": 0.005278584128926917
     },
     {
       "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": 9650.001572931445,
-      "required_pages_per_second_upper_estimate": 0.004040254143638525
+      "required_token_positions_per_second": 9537.451375827732,
+      "required_pages_per_second_upper_estimate": 0.003993131726426578
     },
     {
       "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": 19300.00314586289,
-      "required_pages_per_second_upper_estimate": 0.0040196816799731765
+      "required_token_positions_per_second": 19074.902751655463,
+      "required_pages_per_second_upper_estimate": 0.003972799204156364
     },
     {
       "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": 28950.004718794335,
-      "required_pages_per_second_upper_estimate": 0.001735508578794015
+      "required_token_positions_per_second": 28612.354127483195,
+      "required_pages_per_second_upper_estimate": 0.0017152669414075132
     },
     {
       "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": 28950.004718794335,
-      "required_pages_per_second_upper_estimate": 0.0022507355292729556
+      "required_token_positions_per_second": 28612.354127483195,
+      "required_pages_per_second_upper_estimate": 0.002224484680966508
     },
     {
       "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": 28950.004718794335,
-      "required_pages_per_second_upper_estimate": 0.3713395764394291
+      "required_token_positions_per_second": 28612.354127483195,
+      "required_pages_per_second_upper_estimate": 0.36700855719504877
     }
   ],
   "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": 0.04976373940626644,
-    "producer_p95_seconds": 0.1702074619997802,
-    "gpu_update_seconds": 0.9550257510683902,
-    "producer_batches_per_second": 20.094952910111818,
-    "gpu_batches_per_second": 1.0470921845628738,
-    "producer_headroom": 19.191197495663474,
+    "producer_batch_seconds": 0.11649422678902965,
+    "producer_p95_seconds": 0.21004571299999952,
+    "gpu_update_seconds": 0.9662958831284386,
+    "producer_batches_per_second": 8.5841163769514,
+    "gpu_batches_per_second": 1.03487970657853,
+    "producer_headroom": 8.294796315343547,
     "sustainable": true,
-    "buffer_seconds": 60.16662231730858,
+    "buffer_seconds": 60.87664063709163,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 86,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 317438.5499692344,
-    "upload_bytes_per_second": 5504378.909206007,
+    "generation_bytes_per_second": 322997.3148886,
+    "upload_bytes_per_second": 5468898.378129647,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 68.11492323215269,
+    "estimated_full_shards_per_hour": 69.30770478242398,
     "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": 0.8258607079999933,
-      "bytes": 145464,
-      "pages": 4,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1176107269999989,
-      "bytes": 5541224,
-      "pages": 18,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2054241769999408,
-      "bytes": 2385611,
-      "pages": 27,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8888624019999725,
-      "bytes": 6324210,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1897545489999857,
-      "bytes": 1226596,
-      "pages": 34,
+      "seconds": 1.035186543999771,
+      "bytes": 6290854,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2994146069999033,
-      "bytes": 17142614,
-      "pages": 34,
-      "shards": 5
+      "seconds": 1.1961185629998,
+      "bytes": 11182870,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.1055204159999903,
-      "bytes": 6016330,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.2023984399993424,
+      "bytes": 8470279,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 0.9128079020001678,
-      "bytes": 2402677,
-      "pages": 27,
+      "seconds": 0.9793791169995529,
+      "bytes": 5849196,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0666157699999985,
-      "bytes": 6099291,
-      "pages": 35,
+      "seconds": 0.8912132279992875,
+      "bytes": 2690813,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0668149449998054,
-      "bytes": 5027383,
-      "pages": 32,
+      "seconds": 1.3103161509998245,
+      "bytes": 5739919,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0457134279999991,
-      "bytes": 7722680,
-      "pages": 38,
+      "seconds": 1.3039567450005052,
+      "bytes": 9869427,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1198342459999822,
-      "bytes": 518226,
-      "pages": 18,
-      "shards": 1
+      "seconds": 0.9875589590001255,
+      "bytes": 2599753,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.3007597939999869,
-      "bytes": 12269409,
-      "pages": 44,
+      "seconds": 1.266042872000071,
+      "bytes": 11864568,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.437565898999992,
-      "bytes": 12411381,
-      "pages": 38,
+      "seconds": 1.21107310699972,
+      "bytes": 11765224,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 0.765915884000151,
-      "bytes": 1219645,
-      "pages": 34,
+      "seconds": 2.659455514000001,
+      "bytes": 1230518,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.470795656000064,
-      "bytes": 7604342,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.1675947069998074,
+      "bytes": 6176188,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.9404902840001341,
-      "bytes": 6153110,
-      "pages": 37,
+      "seconds": 0.943889709999894,
+      "bytes": 2411720,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0696395039999516,
-      "bytes": 10812914,
-      "pages": 37,
+      "seconds": 1.251686485999926,
+      "bytes": 10374348,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.5107661739998548,
-      "bytes": 7696973,
-      "pages": 34,
+      "seconds": 1.5644413750005697,
+      "bytes": 11913928,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.9512165999999525,
-      "bytes": 1125687,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.3241815069995937,
+      "bytes": 2678696,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0864234489999944,
-      "bytes": 7369290,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1117148500006806,
+      "bytes": 6299752,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.1581361090002247,
-      "bytes": 10067921,
-      "pages": 37,
-      "shards": 3
+      "seconds": 3.5441821820004407,
+      "bytes": 11982351,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 0.7976953110000977,
-      "bytes": 1131427,
+      "seconds": 1.0069952920002834,
+      "bytes": 1220170,
       "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1826321499997903,
-      "bytes": 12996691,
-      "pages": 33,
-      "shards": 4
+      "seconds": 0.9136476189996756,
+      "bytes": 5828433,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.7847848540000086,
-      "bytes": 603387,
-      "pages": 17,
-      "shards": 1
+      "seconds": 1.4721885819999443,
+      "bytes": 8185407,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.5595179319998351,
-      "bytes": 10763416,
-      "pages": 29,
+      "seconds": 1.3622697179998795,
+      "bytes": 10825600,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.8392939139998816,
-      "bytes": 2549967,
+      "seconds": 1.2178643509996618,
+      "bytes": 7619131,
       "pages": 36,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.9600205120000282,
-      "bytes": 6407325,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8663683109998601,
+      "bytes": 939323,
+      "pages": 28,
+      "shards": 1
     },
     {
-      "seconds": 1.0153757429998223,
-      "bytes": 6078141,
+      "seconds": 0.8838954740003828,
+      "bytes": 5887982,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.8920637669998541,
-      "bytes": 2452797,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.2055649489993812,
+      "bytes": 6199136,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.1147250890003306,
-      "bytes": 15773840,
-      "pages": 37,
+      "seconds": 1.1954915420010366,
+      "bytes": 12042262,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 0.9467493070001183,
-      "bytes": 2673204,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2374154409990297,
+      "bytes": 7885146,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.3979449759999625,
-      "bytes": 6163668,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.751855661999798,
+      "bytes": 918929,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.1962734589997126,
-      "bytes": 10535589,
-      "pages": 38,
+      "seconds": 1.077764785999534,
+      "bytes": 11149283,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.9773600910002642,
-      "bytes": 2300498,
-      "pages": 27,
+      "seconds": 1.0994711280000047,
+      "bytes": 2636989,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3482739699993544,
-      "bytes": 6309981,
-      "pages": 37,
+      "seconds": 0.9980228669992357,
+      "bytes": 5628518,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1507453160002115,
-      "bytes": 8207384,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.4070167600002605,
+      "bytes": 12110556,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.092882823999389,
-      "bytes": 10744820,
-      "pages": 39,
+      "seconds": 1.1031485080002312,
+      "bytes": 6510962,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.0739986469998257,
-      "bytes": 6490657,
+      "seconds": 1.2347332800000004,
+      "bytes": 6132473,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8932356570003321,
-      "bytes": 2377409,
-      "pages": 28,
+      "seconds": 0.8308523640007479,
+      "bytes": 2627309,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.035186543999771,
-      "bytes": 6290854,
-      "pages": 36,
+      "seconds": 2.325367698998889,
+      "bytes": 6081267,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1961185629998,
-      "bytes": 11182870,
-      "pages": 37,
+      "seconds": 1.1514169719994243,
+      "bytes": 10850299,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.2023984399993424,
-      "bytes": 8470279,
-      "pages": 37,
+      "seconds": 1.1272267000003922,
+      "bytes": 8136270,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.9793791169995529,
-      "bytes": 5849196,
-      "pages": 28,
+      "seconds": 1.007810305000021,
+      "bytes": 6126457,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8912132279992875,
-      "bytes": 2690813,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.4154219219999504,
+      "bytes": 7407761,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.3103161509998245,
-      "bytes": 5739919,
-      "pages": 29,
+      "seconds": 1.1865251550007088,
+      "bytes": 5346992,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.3039567450005052,
-      "bytes": 9869427,
+      "seconds": 1.0492023780007003,
+      "bytes": 6288876,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.9875589590001255,
-      "bytes": 2599753,
-      "pages": 37,
+      "seconds": 1.7196641009995801,
+      "bytes": 2291853,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.266042872000071,
-      "bytes": 11864568,
-      "pages": 36,
+      "seconds": 1.0487870670003758,
+      "bytes": 11950097,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.21107310699972,
-      "bytes": 11765224,
-      "pages": 34,
+      "seconds": 1.2208401020016026,
+      "bytes": 11777262,
+      "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 2.659455514000001,
-      "bytes": 1230518,
-      "pages": 35,
+      "seconds": 0.9635194839993346,
+      "bytes": 1029009,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.1675947069998074,
-      "bytes": 6176188,
-      "pages": 36,
+      "seconds": 1.0835146409990557,
+      "bytes": 5946250,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 0.943889709999894,
-      "bytes": 2411720,
-      "pages": 28,
+      "seconds": 0.9391229799985013,
+      "bytes": 2632765,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.251686485999926,
-      "bytes": 10374348,
-      "pages": 34,
+      "seconds": 1.0898778940008924,
+      "bytes": 10697456,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.5644413750005697,
-      "bytes": 11913928,
+      "seconds": 1.3169437420001486,
+      "bytes": 11722554,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.3241815069995937,
-      "bytes": 2678696,
+      "seconds": 1.2027081380001619,
+      "bytes": 2601682,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1117148500006806,
-      "bytes": 6299752,
+      "seconds": 1.0705139929996221,
+      "bytes": 6406862,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 3.5441821820004407,
-      "bytes": 11982351,
-      "pages": 39,
+      "seconds": 3.478892785999051,
+      "bytes": 11471341,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.0069952920002834,
-      "bytes": 1220170,
-      "pages": 35,
+      "seconds": 0.9223496029990201,
+      "bytes": 897417,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 0.9136476189996756,
-      "bytes": 5828433,
-      "pages": 27,
+      "seconds": 1.0840730929994606,
+      "bytes": 6199927,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.4721885819999443,
-      "bytes": 8185407,
-      "pages": 36,
+      "seconds": 1.112097929999436,
+      "bytes": 8142857,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.3622697179998795,
-      "bytes": 10825600,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.17854872099997,
+      "bytes": 10442242,
+      "pages": 43,
+      "shards": 4
     },
     {
-      "seconds": 1.2178643509996618,
-      "bytes": 7619131,
-      "pages": 36,
+      "seconds": 1.163782216999607,
+      "bytes": 7671664,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.8663683109998601,
-      "bytes": 939323,
-      "pages": 28,
+      "seconds": 1.1290235629985546,
+      "bytes": 1186173,
+      "pages": 34,
       "shards": 1
+    },
+    {
+      "seconds": 1.2037071409995406,
+      "bytes": 10277758,
+      "pages": 28,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9222849990001123,
+      "bytes": 2428641,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1589227389995358,
+      "bytes": 11976819,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0021623419997923,
+      "bytes": 6273488,
+      "pages": 37,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f2026403cd8252eb8fbef1272eb9f1d5d4b596cf0488b73b1b78201830ca2e97
-size 721761919
+oid sha256:a3bc7b8df78537b2475ab2dd3ebc2aba12dbb1b97f2867ccda2914ec9a3004e2
+size 721777983