Qwen90M step 295000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 290000,
+  "step": 295000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "0a0aa696d2def99e7d22de3de4aeff668d43be50c970daf9230586d60febe595",
+    "model.safetensors": "59f58f5c4bdb2a0576a3b78dd8f8666a1dd1d6108b3210d2d71cfcc0ff980702",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "00b52798ce4bbc979b3cc619c785cf5f2e3b8c46472773cb746b19a780d676ff",
-    "training_config.json": "65390fe09f12d0188f44175607f7a9d1665082a136656274cb5d054eb998e80b",
+    "trainer.pt": "9182b42958ff73eae94225e7d961ee0aec9363f3f5f387fc8e3072607f9dc6fc",
+    "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf",
     "performance.json": "f89f82c3f4c44a52fcd532346ea71580902d757e59e1961e9b2bed0030259dba",
-    "shard_scheduler.json": "6b224639b01f5768d2f11ee38fffbf310fdb30edc5736dacb2cee65fc42a8162",
-    "coverage.json": "244fa16ffdbcd8c9dc4f95fdb90318ed6782c1c785fc2b0b5b817be4e931ecb6",
+    "shard_scheduler.json": "aeba209266c0f57fa8c60a462539562d488605c9131398cabd5bb4cec88efffe",
+    "coverage.json": "517abe535f2de9e8ba9121e0877123b28aeea27b9303155337162e9d0163f11e",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 290000,
+  "step": 295000,
   "stats": {
-    "climbmix/blocks": 10440000,
-    "climbmix/targets": 17452321234,
-    "climbmix/input_tokens": 17453427448,
-    "ultra_qa/blocks": 3915000,
-    "ultra_qa/targets": 6386778194,
-    "ultra_qa/input_tokens": 6386790465,
-    "cortex/blocks": 3915000,
-    "cortex/targets": 148927427,
-    "cortex/input_tokens": 6223329188,
-    "ultra_style/blocks": 3915000,
-    "ultra_style/targets": 6192313215,
-    "ultra_style/input_tokens": 6192638355,
-    "openmath/blocks": 2610000,
-    "openmath/targets": 4033044745,
-    "openmath/input_tokens": 4034389394,
-    "rewrite6/blocks": 1305000,
-    "rewrite6/targets": 2139357551,
-    "rewrite6/input_tokens": 2141703031
+    "climbmix/blocks": 10620000,
+    "climbmix/targets": 17753092470,
+    "climbmix/input_tokens": 17754217730,
+    "ultra_qa/blocks": 3982500,
+    "ultra_qa/targets": 6496902054,
+    "ultra_qa/input_tokens": 6496914542,
+    "cortex/blocks": 3982500,
+    "cortex/targets": 151499631,
+    "cortex/input_tokens": 6330619515,
+    "ultra_style/blocks": 3982500,
+    "ultra_style/targets": 6299348778,
+    "ultra_style/input_tokens": 6299679365,
+    "openmath/blocks": 2655000,
+    "openmath/targets": 4102640002,
+    "openmath/input_tokens": 4104007922,
+    "rewrite6/blocks": 1327500,
+    "rewrite6/targets": 2176202421,
+    "rewrite6/input_tokens": 2178588636
   },
   "sources": {
     "climbmix": {
-      "page": 1429,
-      "row": 3549,
+      "page": 1454,
+      "row": 2314,
       "start": {
-        "file": 69,
+        "file": 70,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 14,
+          "row_group": 31,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5856733
+      "documents": 5957898
     },
     "rewrite6": {
-      "page": 13,
-      "row": 1729,
+      "page": 27,
+      "row": 2421,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 53248
+          "row": 110592
         }
       },
       "epoch": 16,
       "offset": 0,
-      "documents": 3341441
+      "documents": 3399477
     },
     "openmath": {
-      "page": 893,
-      "row": 3429,
+      "page": 909,
+      "row": 1156,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 164,
-          "row": 528
+          "row_group": 230,
+          "row": 64
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 3661157
+      "documents": 3724420
     },
     "ultra_qa": {
-      "page": 386,
-      "row": 1389,
+      "page": 393,
+      "row": 0,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 0,
-          "row": 22181
+          "row": 50853
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1582445
+      "documents": 1609728
     },
     "ultra_style": {
-      "page": 590,
-      "row": 113,
+      "page": 600,
+      "row": 410,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 20398
+          "row": 61358
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 2416753
+      "offset": 4096,
+      "documents": 2458010
     },
     "cortex": {
-      "page": 75929,
-      "row": 52,
+      "page": 77238,
+      "row": 62,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 9718964
+      "documents": 9886526
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 290000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "0a0aa696d2def99e7d22de3de4aeff668d43be50c970daf9230586d60febe595", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "00b52798ce4bbc979b3cc619c785cf5f2e3b8c46472773cb746b19a780d676ff", "training_config.json": "65390fe09f12d0188f44175607f7a9d1665082a136656274cb5d054eb998e80b", "performance.json": "f89f82c3f4c44a52fcd532346ea71580902d757e59e1961e9b2bed0030259dba", "shard_scheduler.json": "6b224639b01f5768d2f11ee38fffbf310fdb30edc5736dacb2cee65fc42a8162", "coverage.json": "244fa16ffdbcd8c9dc4f95fdb90318ed6782c1c785fc2b0b5b817be4e931ecb6", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000290000"}
+{"step": 295000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "59f58f5c4bdb2a0576a3b78dd8f8666a1dd1d6108b3210d2d71cfcc0ff980702", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "9182b42958ff73eae94225e7d961ee0aec9363f3f5f387fc8e3072607f9dc6fc", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "f89f82c3f4c44a52fcd532346ea71580902d757e59e1961e9b2bed0030259dba", "shard_scheduler.json": "aeba209266c0f57fa8c60a462539562d488605c9131398cabd5bb4cec88efffe", "coverage.json": "517abe535f2de9e8ba9121e0877123b28aeea27b9303155337162e9d0163f11e", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000295000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0a0aa696d2def99e7d22de3de4aeff668d43be50c970daf9230586d60febe595
+oid sha256:59f58f5c4bdb2a0576a3b78dd8f8666a1dd1d6108b3210d2d71cfcc0ff980702
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 1062967.6168812835,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 76378.33191482387,
-      "required_pages_per_second_upper_estimate": 0.006620280231264481
+      "required_token_positions_per_second": 73779.41610322442,
+      "required_pages_per_second_upper_estimate": 0.006395012795606944
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 1206672.263377472,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9547.291489352983,
-      "required_pages_per_second_upper_estimate": 0.0038085573198312524
+      "required_token_positions_per_second": 9222.427012903052,
+      "required_pages_per_second_upper_estimate": 0.0036789640230186103
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 410042.7581134458,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 19094.582978705967,
-      "required_pages_per_second_upper_estimate": 0.004417195718940841
+      "required_token_positions_per_second": 18444.854025806104,
+      "required_pages_per_second_upper_estimate": 0.004266892360526481
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 2163486.5624068812,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28641.87446805895,
-      "required_pages_per_second_upper_estimate": 0.0017441398828031117
+      "required_token_positions_per_second": 27667.281038709156,
+      "required_pages_per_second_upper_estimate": 0.0016847922562522565
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 1546694.2737633225,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28641.87446805895,
-      "required_pages_per_second_upper_estimate": 0.0022049307823342216
+      "required_token_positions_per_second": 27667.281038709156,
+      "required_pages_per_second_upper_estimate": 0.002129903882295592
     },
     {
       "source": "cortex",
@@ -91,9 +91,9 @@
       "read_seconds": 0.042780299000014566,
       "tokens_per_second": 1015177.8395898147,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
-      "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28641.87446805895,
-      "required_pages_per_second_upper_estimate": 0.35563622270582407
+      "estimated_pages_for_buffer": 21,
+      "required_token_positions_per_second": 27667.281038709156,
+      "required_pages_per_second_upper_estimate": 0.3435350340676851
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,413 +155,413 @@
   "shard_benchmark_scope": "Local packing/read measurements; upload capacity measured only on real data commits. Targets larger than available pages are sample-limited.",
   "estimated_fresh_generation_batch_seconds": 0.1434574083060457,
   "selected": {
-    "prefetch_batches": 63,
+    "prefetch_batches": 61,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 690670,
-    "producer_batch_seconds": 0.0629032884064884,
-    "producer_p95_seconds": 0.20650791799926083,
-    "gpu_update_seconds": 0.9652999502819793,
-    "producer_batches_per_second": 15.897420076640241,
-    "gpu_batches_per_second": 1.0359474272301414,
-    "producer_headroom": 15.345778809592565,
+    "producer_batch_seconds": 0.4320460435155269,
+    "producer_p95_seconds": 0.3225658029987244,
+    "gpu_update_seconds": 0.9993031104616974,
+    "producer_batches_per_second": 2.314568123024744,
+    "gpu_batches_per_second": 1.0006973755320152,
+    "producer_headroom": 2.3129551247141196,
     "sustainable": true,
-    "buffer_seconds": 60.8138968677647,
+    "buffer_seconds": 60.95748973816354,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 543062.8026689846,
-    "upload_bytes_per_second": 5443648.407667226,
+    "generation_bytes_per_second": 296585.41787994764,
+    "upload_bytes_per_second": 3972390.366839625,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 116.52863559772638,
+    "estimated_full_shards_per_hour": 63.640326521862235,
     "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": 1.2733040839993919,
-      "bytes": 13322526,
-      "pages": 35,
+      "seconds": 1.2214710799999011,
+      "bytes": 12703928,
+      "pages": 49,
       "shards": 4
     },
     {
-      "seconds": 0.8405508110008668,
-      "bytes": 1090490,
-      "pages": 33,
+      "seconds": 1.1181347120000282,
+      "bytes": 6085437,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1846620260002965,
+      "bytes": 927364,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.1666756240010727,
-      "bytes": 10146533,
-      "pages": 27,
+      "seconds": 1.263599650001197,
+      "bytes": 13782793,
+      "pages": 42,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1351800550000917,
+      "bytes": 9521447,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.1112560749988916,
-      "bytes": 2663601,
-      "pages": 36,
+      "seconds": 1.0451152109999384,
+      "bytes": 1099454,
+      "pages": 31,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1156518739990133,
+      "bytes": 7370564,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.182102421998934,
-      "bytes": 5962190,
-      "pages": 36,
+      "seconds": 1.0604308100009803,
+      "bytes": 5273416,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.123748488000274,
+      "bytes": 5660713,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.9607173089989374,
-      "bytes": 7750938,
+      "seconds": 1.1639190339992638,
+      "bytes": 7647995,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1050905230003991,
-      "bytes": 6187776,
-      "pages": 15,
+      "seconds": 1.0442986829984875,
+      "bytes": 6780113,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.2981188600006135,
-      "bytes": 10611352,
-      "pages": 45,
-      "shards": 3
+      "seconds": 1.2669327539988444,
+      "bytes": 12406942,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.1970796210007393,
-      "bytes": 2312155,
-      "pages": 27,
+      "seconds": 1.4563389440008905,
+      "bytes": 6496961,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.6818293220003397,
-      "bytes": 5841398,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.1376306590027525,
+      "bytes": 968236,
+      "pages": 28,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3645541389996652,
+      "bytes": 7442538,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.1767334710002615,
-      "bytes": 1036234,
+      "seconds": 0.78800499800127,
+      "bytes": 976266,
       "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.7388112589997036,
-      "bytes": 12464285,
-      "pages": 43,
-      "shards": 4
+      "seconds": 1.7201119790006487,
+      "bytes": 17653647,
+      "pages": 35,
+      "shards": 5
     },
     {
-      "seconds": 0.8487006530012877,
-      "bytes": 592565,
-      "pages": 19,
-      "shards": 1
+      "seconds": 1.0727579819977109,
+      "bytes": 6128146,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.3723258429999987,
-      "bytes": 11921886,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.0307726470018679,
+      "bytes": 1163007,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.34920726099881,
-      "bytes": 7816018,
-      "pages": 35,
+      "seconds": 1.1981934089999413,
+      "bytes": 7536231,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.4124847809998755,
-      "bytes": 4874448,
-      "pages": 37,
+      "seconds": 1.27145491400006,
+      "bytes": 5090460,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.2419382850002876,
-      "bytes": 7186580,
-      "pages": 29,
+      "seconds": 1.1308283350008423,
+      "bytes": 7872807,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1265525479993812,
-      "bytes": 6298017,
-      "pages": 35,
-      "shards": 2
+      "seconds": 2.8297132219995547,
+      "bytes": 11437747,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 0.9624441889991431,
-      "bytes": 1027701,
-      "pages": 30,
+      "seconds": 1.1368421070001204,
+      "bytes": 1161800,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.7364580140001635,
-      "bytes": 17104134,
-      "pages": 27,
-      "shards": 5
+      "seconds": 1.1932739869989746,
+      "bytes": 11243706,
+      "pages": 26,
+      "shards": 4
     },
     {
-      "seconds": 0.9077872060006484,
-      "bytes": 916446,
-      "pages": 29,
+      "seconds": 1.4697227199976624,
+      "bytes": 921486,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.0004914799992548,
-      "bytes": 6063816,
+      "seconds": 1.586299536000297,
+      "bytes": 6350565,
       "pages": 36,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.9483974380000291,
-      "bytes": 2643403,
-      "pages": 36,
+      "seconds": 0.9621634939976502,
+      "bytes": 2554887,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2182073419990047,
-      "bytes": 10399159,
-      "pages": 23,
-      "shards": 3
+      "seconds": 1.2608247060015856,
+      "bytes": 16000072,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 0.9434353479991842,
-      "bytes": 5975783,
-      "pages": 35,
+      "seconds": 1.1701329820025421,
+      "bytes": 6128160,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9842061780000222,
-      "bytes": 8165489,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9350508080005966,
-      "bytes": 6046815,
+      "seconds": 0.9900568350021786,
+      "bytes": 2575859,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3128581109995139,
-      "bytes": 2374553,
+      "seconds": 1.209377144001337,
+      "bytes": 5838608,
       "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.2559754029989563,
-      "bytes": 9926404,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.249938580000162,
+      "bytes": 4790943,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 0.9813094319997617,
-      "bytes": 1049873,
-      "pages": 31,
-      "shards": 1
+      "seconds": 1.2871860160012147,
+      "bytes": 7434728,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.218386226999428,
-      "bytes": 7220161,
-      "pages": 32,
+      "seconds": 31.810438357002567,
+      "bytes": 11940488,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.9670737249998638,
-      "bytes": 5661886,
-      "pages": 23,
-      "shards": 2
+      "seconds": 1.3719915590008895,
+      "bytes": 12062246,
+      "pages": 44,
+      "shards": 4
     },
     {
-      "seconds": 1.3326636759993562,
-      "bytes": 12247274,
-      "pages": 47,
-      "shards": 3
+      "seconds": 0.804579094001383,
+      "bytes": 1179275,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.6376108280001063,
-      "bytes": 7445606,
-      "pages": 29,
+      "seconds": 1.2755006959996535,
+      "bytes": 7429804,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.7973876159994688,
-      "bytes": 1188820,
+      "seconds": 1.0133618440013379,
+      "bytes": 1167322,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1932972649992735,
-      "bytes": 7607279,
-      "pages": 37,
-      "shards": 4
-    },
-    {
-      "seconds": 1.4406277779999073,
-      "bytes": 9581963,
-      "pages": 33,
+      "seconds": 1.291220240000257,
+      "bytes": 10903036,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.2597373989992775,
-      "bytes": 7034037,
-      "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1449642560000939,
-      "bytes": 7450954,
-      "pages": 28,
-      "shards": 3
+      "seconds": 1.2618899920016702,
+      "bytes": 13332712,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 0.8681581979999464,
-      "bytes": 818733,
-      "pages": 23,
+      "seconds": 0.7849141779988713,
+      "bytes": 1153413,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.2214710799999011,
-      "bytes": 12703928,
-      "pages": 49,
-      "shards": 4
+      "seconds": 1.3497758060002525,
+      "bytes": 7313270,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.1181347120000282,
-      "bytes": 6085437,
+      "seconds": 1.170668664999539,
+      "bytes": 6087923,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1846620260002965,
-      "bytes": 927364,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.263599650001197,
-      "bytes": 13782793,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.1090394350030692,
+      "bytes": 4744304,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.1351800550000917,
-      "bytes": 9521447,
-      "pages": 29,
+      "seconds": 1.304362898998079,
+      "bytes": 7978674,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0451152109999384,
-      "bytes": 1099454,
-      "pages": 31,
-      "shards": 1
+      "seconds": 1.0655079770003795,
+      "bytes": 7055873,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1156518739990133,
-      "bytes": 7370564,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.268422623998049,
+      "bytes": 5726191,
+      "pages": 20,
+      "shards": 2
     },
     {
-      "seconds": 1.0604308100009803,
-      "bytes": 5273416,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1066016949989717,
+      "bytes": 7887477,
+      "pages": 44,
+      "shards": 3
     },
     {
-      "seconds": 1.123748488000274,
-      "bytes": 5660713,
-      "pages": 27,
+      "seconds": 1.019014984001842,
+      "bytes": 6005453,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1639190339992638,
-      "bytes": 7647995,
-      "pages": 37,
+      "seconds": 1.00243121099993,
+      "bytes": 7452907,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.0442986829984875,
-      "bytes": 6780113,
-      "pages": 31,
-      "shards": 2
+      "seconds": 0.9669066969981941,
+      "bytes": 1217353,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.2669327539988444,
-      "bytes": 12406942,
-      "pages": 41,
+      "seconds": 1.4986801279992505,
+      "bytes": 15911648,
+      "pages": 32,
       "shards": 4
     },
     {
-      "seconds": 1.4563389440008905,
-      "bytes": 6496961,
-      "pages": 33,
+      "seconds": 1.1255449119998957,
+      "bytes": 2547314,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.1376306590027525,
-      "bytes": 968236,
-      "pages": 28,
-      "shards": 1
+      "seconds": 1.1576964999985648,
+      "bytes": 6153703,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.3645541389996652,
-      "bytes": 7442538,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.0455056530008733,
+      "bytes": 6073870,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 0.78800499800127,
-      "bytes": 976266,
+      "seconds": 1.0066101270022045,
+      "bytes": 7063346,
       "pages": 30,
-      "shards": 1
-    },
-    {
-      "seconds": 1.7201119790006487,
-      "bytes": 17653647,
-      "pages": 35,
-      "shards": 5
+      "shards": 3
     },
     {
-      "seconds": 1.0727579819977109,
-      "bytes": 6128146,
+      "seconds": 1.161024010998517,
+      "bytes": 6075989,
       "pages": 36,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.0307726470018679,
-      "bytes": 1163007,
-      "pages": 34,
+      "seconds": 0.8424355740025931,
+      "bytes": 669079,
+      "pages": 20,
       "shards": 1
     },
     {
-      "seconds": 1.1981934089999413,
-      "bytes": 7536231,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.4896883629990043,
+      "bytes": 13505965,
+      "pages": 48,
+      "shards": 4
     },
     {
-      "seconds": 1.27145491400006,
-      "bytes": 5090460,
-      "pages": 31,
+      "seconds": 1.098302499998681,
+      "bytes": 6017424,
+      "pages": 23,
       "shards": 2
     },
     {
-      "seconds": 1.1308283350008423,
-      "bytes": 7872807,
-      "pages": 35,
+      "seconds": 1.0887724580024951,
+      "bytes": 6603634,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 2.8297132219995547,
-      "bytes": 11437747,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.2179049539990956,
+      "bytes": 6068321,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.1368421070001204,
-      "bytes": 1161800,
-      "pages": 34,
-      "shards": 1
+      "seconds": 2.3319496820004133,
+      "bytes": 5855101,
+      "pages": 27,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:00b52798ce4bbc979b3cc619c785cf5f2e3b8c46472773cb746b19a780d676ff
-size 721780479
+oid sha256:9182b42958ff73eae94225e7d961ee0aec9363f3f5f387fc8e3072607f9dc6fc
+size 721764223

training_config.json

@@ -46,8 +46,8 @@
   "cpu_threads": 4,
   "page_rows": 4096,
   "cortex_page_rows": 128,
-  "prefetch_batches": 63,
-  "prefetch_low_water": 31,
+  "prefetch_batches": 61,
+  "prefetch_low_water": 30,
   "memory_pages": 12,
   "cache_gb": 20,
   "force_rebenchmark": false,