Qwen90M step 700000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 695000,
+  "step": 700000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "7bc36582f93f2989aadd91232699a7af1bc7686d271b31d30c2381bb46514fed",
+    "model.safetensors": "d2961c6f05c442c4ab734fdc321c4bd99d4d2cdb08dfc0df55b424ae0cb668fd",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "b91f72a9cbd032ac4aee83653c42086a0ab52034712e3ebf828258d87653ea7c",
-    "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8",
+    "trainer.pt": "071781bcfaf5f9809ccf892899e3b350ca09fb10a20d4e9fa8aca6156c9779ea",
+    "training_config.json": "6b17d8d96cc52df4441b50c79edf3e942bfe52f3c28b64d97f8a592a94b8549e",
     "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29",
-    "shard_scheduler.json": "2790a0c930904e09253ffeb8a2fa8e78d0efbaea59ac655453a067775f8e70cb",
-    "coverage.json": "829a2aa07677fbb65f65c1542e4754afdcac1f9751729ae84cf87c3de9db2b47",
+    "shard_scheduler.json": "cc4225a4f4c10a3ebe7ae762beddc0b1028f2619039eed4fbdfc67099a29a929",
+    "coverage.json": "fbe12c0005986d94f81200adeea5b3161fe8c1a4817c2774a13be90699b44102",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 695000,
+  "step": 700000,
   "stats": {
-    "climbmix/blocks": 25020000,
-    "climbmix/targets": 41825829199,
-    "climbmix/input_tokens": 41828479893,
-    "ultra_qa/blocks": 9382500,
-    "ultra_qa/targets": 15306790749,
-    "ultra_qa/input_tokens": 15306820344,
-    "cortex/blocks": 9382500,
-    "cortex/targets": 356748789,
-    "cortex/input_tokens": 14914158761,
-    "ultra_style/blocks": 9382500,
-    "ultra_style/targets": 14840323358,
-    "ultra_style/input_tokens": 14841102239,
-    "openmath/blocks": 6255000,
-    "openmath/targets": 9665356984,
-    "openmath/input_tokens": 9668577218,
-    "rewrite6/blocks": 3127500,
-    "rewrite6/targets": 5127151612,
-    "rewrite6/input_tokens": 5132772712
+    "climbmix/blocks": 25200000,
+    "climbmix/targets": 42126492444,
+    "climbmix/input_tokens": 42129162281,
+    "ultra_qa/blocks": 9450000,
+    "ultra_qa/targets": 15416907813,
+    "ultra_qa/input_tokens": 15416937583,
+    "cortex/blocks": 9450000,
+    "cortex/targets": 359315602,
+    "cortex/input_tokens": 15021391599,
+    "ultra_style/blocks": 9450000,
+    "ultra_style/targets": 14946958997,
+    "ultra_style/input_tokens": 14947743302,
+    "openmath/blocks": 6300000,
+    "openmath/targets": 9734892069,
+    "openmath/input_tokens": 9738135363,
+    "rewrite6/blocks": 3150000,
+    "rewrite6/targets": 5163995804,
+    "rewrite6/input_tokens": 5169657315
   },
   "sources": {
     "climbmix": {
-      "page": 3426,
-      "row": 2946,
+      "page": 3451,
+      "row": 1860,
       "start": {
-        "file": 165,
+        "file": 167,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 67,
+          "row_group": 2,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 14035842
+      "offset": 2048,
+      "documents": 14137156
     },
     "rewrite6": {
-      "page": 49,
-      "row": 2054,
+      "page": 13,
+      "row": 1638,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 200704
+          "row": 53248
         }
       },
-      "epoch": 38,
+      "epoch": 39,
       "offset": 0,
-      "documents": 8008110
+      "documents": 8065642
     },
     "openmath": {
-      "page": 2141,
-      "row": 2196,
+      "page": 2156,
+      "row": 3724,
       "start": {
         "file": 20,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 36,
-          "row": 536
+          "row_group": 97,
+          "row": 976
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 8771732
+      "offset": 2048,
+      "documents": 8834700
     },
     "ultra_qa": {
-      "page": 925,
-      "row": 4086,
+      "page": 932,
+      "row": 2614,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 22133
+          "row": 50805
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3792886
+      "documents": 3820086
     },
     "ultra_style": {
-      "page": 1413,
-      "row": 1853,
+      "page": 1423,
+      "row": 2123,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 52009
+          "row_group": 4,
+          "row": 8311
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5789501
+      "documents": 5830731
     },
     "cortex": {
-      "page": 181966,
-      "row": 121,
+      "page": 183275,
+      "row": 1,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 23291769
+      "documents": 23459201
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 695000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "7bc36582f93f2989aadd91232699a7af1bc7686d271b31d30c2381bb46514fed", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b91f72a9cbd032ac4aee83653c42086a0ab52034712e3ebf828258d87653ea7c", "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8", "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29", "shard_scheduler.json": "2790a0c930904e09253ffeb8a2fa8e78d0efbaea59ac655453a067775f8e70cb", "coverage.json": "829a2aa07677fbb65f65c1542e4754afdcac1f9751729ae84cf87c3de9db2b47", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac", "folder": "step-000000695000"}
+{"step": 700000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "d2961c6f05c442c4ab734fdc321c4bd99d4d2cdb08dfc0df55b424ae0cb668fd", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "071781bcfaf5f9809ccf892899e3b350ca09fb10a20d4e9fa8aca6156c9779ea", "training_config.json": "6b17d8d96cc52df4441b50c79edf3e942bfe52f3c28b64d97f8a592a94b8549e", "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29", "shard_scheduler.json": "cc4225a4f4c10a3ebe7ae762beddc0b1028f2619039eed4fbdfc67099a29a929", "coverage.json": "fbe12c0005986d94f81200adeea5b3161fe8c1a4817c2774a13be90699b44102", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac", "folder": "step-000000700000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7bc36582f93f2989aadd91232699a7af1bc7686d271b31d30c2381bb46514fed
+oid sha256:d2961c6f05c442c4ab734fdc321c4bd99d4d2cdb08dfc0df55b424ae0cb668fd
 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": 77286.3711192274,
-      "required_pages_per_second_upper_estimate": 0.005346850483115876
+      "required_token_positions_per_second": 77507.70753695644,
+      "required_pages_per_second_upper_estimate": 0.005362163050065615
     },
     {
       "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": 9660.796389903426,
-      "required_pages_per_second_upper_estimate": 0.004044773708083281
+      "required_token_positions_per_second": 9688.463442119555,
+      "required_pages_per_second_upper_estimate": 0.004056357325092425
     },
     {
       "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": 19321.59277980685,
-      "required_pages_per_second_upper_estimate": 0.004024178231366769
+      "required_token_positions_per_second": 19376.92688423911,
+      "required_pages_per_second_upper_estimate": 0.004035702866061544
     },
     {
       "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": 28982.389169710274,
-      "required_pages_per_second_upper_estimate": 0.0017374499771782323
+      "required_token_positions_per_second": 29065.39032635867,
+      "required_pages_per_second_upper_estimate": 0.0017424257697838755
     },
     {
       "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": 28982.389169710274,
-      "required_pages_per_second_upper_estimate": 0.002253253277887524
+      "required_token_positions_per_second": 29065.39032635867,
+      "required_pages_per_second_upper_estimate": 0.002259706252733432
     },
     {
       "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": 28982.389169710274,
-      "required_pages_per_second_upper_estimate": 0.3717549694040645
+      "required_token_positions_per_second": 29065.39032635867,
+      "required_pages_per_second_upper_estimate": 0.3728196191218516
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,168 +155,30 @@
   "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": 2.558590814491007,
   "selected": {
-    "prefetch_batches": 63,
+    "prefetch_batches": 64,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.106444388218776,
-    "producer_p95_seconds": 0.20390691600005084,
-    "gpu_update_seconds": 0.953958620806015,
-    "producer_batches_per_second": 9.394576987418933,
-    "gpu_batches_per_second": 1.0482634971683404,
-    "producer_headroom": 8.962037705974092,
+    "producer_batch_seconds": 0.08243451003123425,
+    "producer_p95_seconds": 0.21277156500036654,
+    "gpu_update_seconds": 0.9512344300061482,
+    "producer_batches_per_second": 12.130841799400544,
+    "gpu_batches_per_second": 1.0512655644660978,
+    "producer_headroom": 11.539274384547534,
     "sustainable": true,
-    "buffer_seconds": 60.09939311077895,
+    "buffer_seconds": 60.87900352039348,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 83,
+    "upload_pages_per_commit": 80,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 375549.2298314112,
-    "upload_bytes_per_second": 3808720.5720877824,
+    "generation_bytes_per_second": 409692.50155643857,
+    "upload_bytes_per_second": 4323983.225698083,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 80.58412238318208,
+    "estimated_full_shards_per_hour": 87.91047368068568,
     "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": 0.9768797280003128,
-      "bytes": 23252,
-      "pages": 1,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0272072289999414,
-      "bytes": 7383107,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1484548910002559,
-      "bytes": 5982767,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.7043199660001846,
-      "bytes": 963598,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1770538490000035,
-      "bytes": 8003790,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2935991909998847,
-      "bytes": 11779732,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0504046130008646,
-      "bytes": 6464700,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9827144040000348,
-      "bytes": 5877076,
-      "pages": 26,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1320399030000772,
-      "bytes": 865700,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2121052059992508,
-      "bytes": 12463391,
-      "pages": 43,
-      "shards": 4
-    },
-    {
-      "seconds": 1.107387490000292,
-      "bytes": 5957600,
-      "pages": 27,
-      "shards": 2
-    },
-    {
-      "seconds": 25.62068473599993,
-      "bytes": 7109407,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.223384894999981,
-      "bytes": 7577595,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9591294869997,
-      "bytes": 6044241,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1359251849999055,
-      "bytes": 7093211,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1475876189997507,
-      "bytes": 6414179,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1066383229999701,
-      "bytes": 6242311,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1400584189996152,
-      "bytes": 8258283,
-      "pages": 32,
-      "shards": 4
-    },
-    {
-      "seconds": 1.5345939340004406,
-      "bytes": 10306218,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.8383978550000393,
-      "bytes": 2628306,
-      "pages": 33,
-      "shards": 2
-    },
-    {
-      "seconds": 1.650117919000877,
-      "bytes": 6106251,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0884742970001753,
-      "bytes": 6215302,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2488605079997797,
-      "bytes": 6396111,
-      "pages": 35,
-      "shards": 3
-    },
     {
       "seconds": 1.8616032419995463,
       "bytes": 11914182,
@@ -454,6 +316,252 @@
       "bytes": 11238485,
       "pages": 32,
       "shards": 4
+    },
+    {
+      "seconds": 0.9324971470014134,
+      "bytes": 1088704,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2869174759998714,
+      "bytes": 6108290,
+      "pages": 9,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1298799779997353,
+      "bytes": 7635963,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 1.3202205550005601,
+      "bytes": 6021371,
+      "pages": 27,
+      "shards": 2
+    },
+    {
+      "seconds": 1.0867800260002696,
+      "bytes": 6181304,
+      "pages": 41,
+      "shards": 2
+    },
+    {
+      "seconds": 25.14463806799904,
+      "bytes": 7664241,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.7939357480008766,
+      "bytes": 6243705,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.0855320710015803,
+      "bytes": 8387056,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.502407841000604,
+      "bytes": 9864729,
+      "pages": 33,
+      "shards": 3
+    },
+    {
+      "seconds": 1.077067022999472,
+      "bytes": 1093074,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.176304970000274,
+      "bytes": 7496411,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0410123799993016,
+      "bytes": 5644158,
+      "pages": 19,
+      "shards": 2
+    },
+    {
+      "seconds": 1.434721014000388,
+      "bytes": 6921105,
+      "pages": 46,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2365994310002861,
+      "bytes": 11662593,
+      "pages": 31,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2687110730003042,
+      "bytes": 6191110,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3943383940004423,
+      "bytes": 6935244,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2232918650006468,
+      "bytes": 6107930,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1886329499993735,
+      "bytes": 6263136,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9854559350005729,
+      "bytes": 2421861,
+      "pages": 30,
+      "shards": 2
+    },
+    {
+      "seconds": 1.444283180999264,
+      "bytes": 6519870,
+      "pages": 22,
+      "shards": 2
+    },
+    {
+      "seconds": 1.324441224000111,
+      "bytes": 10689415,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1898289029995794,
+      "bytes": 7719194,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9053547809999145,
+      "bytes": 813842,
+      "pages": 25,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1792037220002385,
+      "bytes": 7645453,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0877346289998968,
+      "bytes": 10017417,
+      "pages": 33,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1130377059998864,
+      "bytes": 986874,
+      "pages": 29,
+      "shards": 1
+    },
+    {
+      "seconds": 1.09502404000159,
+      "bytes": 13537871,
+      "pages": 40,
+      "shards": 4
+    },
+    {
+      "seconds": 0.918569364999712,
+      "bytes": 999510,
+      "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.372887202000129,
+      "bytes": 11719387,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0358149310013687,
+      "bytes": 6338461,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8834309400008351,
+      "bytes": 840516,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1130535650008824,
+      "bytes": 11588700,
+      "pages": 34,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0404952170010802,
+      "bytes": 6993608,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.010654464000254,
+      "bytes": 6096042,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1592712370002118,
+      "bytes": 7445713,
+      "pages": 36,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0245760489997338,
+      "bytes": 5205900,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 1.034200502999738,
+      "bytes": 7563521,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.312004507999518,
+      "bytes": 5834009,
+      "pages": 26,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3078732329995546,
+      "bytes": 7188440,
+      "pages": 42,
+      "shards": 2
+    },
+    {
+      "seconds": 1.37187312900096,
+      "bytes": 11914910,
+      "pages": 34,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1714240479996079,
+      "bytes": 6274938,
+      "pages": 37,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b91f72a9cbd032ac4aee83653c42086a0ab52034712e3ebf828258d87653ea7c
-size 721759359
+oid sha256:071781bcfaf5f9809ccf892899e3b350ca09fb10a20d4e9fa8aca6156c9779ea
+size 721768191

training_config.json

@@ -53,8 +53,8 @@
   "cpu_threads": 4,
   "page_rows": 4096,
   "cortex_page_rows": 128,
-  "prefetch_batches": 63,
-  "prefetch_low_water": 31,
+  "prefetch_batches": 64,
+  "prefetch_low_water": 32,
   "memory_pages": 12,
   "cache_gb": 20,
   "force_rebenchmark": false,