Qwen90M step 845000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 842507,
+  "step": 845000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "6f8382631e83ae42ed9759e16b99adad61f343fa6de9cb2d60b9dd6c9148bc70",
+    "model.safetensors": "28d525df26bddcdad1f7e0ee21d00c4e2f6f2750838939feccfbef122663248b",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "a2b248de903e43cc04b64f31262641a2a596e6720f6e8b83622de3cb8a1ad33a",
-    "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
-    "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
-    "shard_scheduler.json": "76298eb667981f18a15e74c27926fbcd7a9ae27df4f2ad6d655bc8499e520b76",
-    "coverage.json": "b6f294dc008ba78304b1fa2017cef88912e1ceaaf314ebcebdb825482582a1b3",
+    "trainer.pt": "87fc13206a6115414dc5698ba185a42d91483c871e4ad1d988dd31693f1ea24e",
+    "training_config.json": "0ab64376a29947d9a9f670c0ece06c1900ab6f9ec82482957af5fc2c70beec58",
+    "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367",
+    "shard_scheduler.json": "57f3f5db4104672d1623ef748776a5e35f680600431f1b4c0d4f28fc3f966af9",
+    "coverage.json": "e2b9a86150908d61f021a350e36374564851cf14a57027b0b71fa60173ea1a90",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 842507,
+  "step": 845000,
   "stats": {
-    "climbmix/blocks": 30330252,
-    "climbmix/targets": 50701917572,
-    "climbmix/input_tokens": 50705130186,
-    "ultra_qa/blocks": 11373844,
-    "ultra_qa/targets": 18555701427,
-    "ultra_qa/input_tokens": 18555737118,
-    "cortex/blocks": 11373845,
-    "cortex/targets": 432503890,
-    "cortex/input_tokens": 18079643538,
-    "ultra_style/blocks": 11373844,
-    "ultra_style/targets": 17989747078,
-    "ultra_style/input_tokens": 17990690711,
-    "openmath/blocks": 7582563,
-    "openmath/targets": 11716447202,
-    "openmath/input_tokens": 11720351418,
-    "rewrite6/blocks": 3791282,
-    "rewrite6/targets": 6215294727,
-    "rewrite6/input_tokens": 6222108705
+    "climbmix/blocks": 30420000,
+    "climbmix/targets": 50852257328,
+    "climbmix/input_tokens": 50855479577,
+    "ultra_qa/blocks": 11407500,
+    "ultra_qa/targets": 18610655674,
+    "ultra_qa/input_tokens": 18610691463,
+    "cortex/blocks": 11407500,
+    "cortex/targets": 433787145,
+    "cortex/input_tokens": 18133094420,
+    "ultra_style/blocks": 11407500,
+    "ultra_style/targets": 18043082009,
+    "ultra_style/input_tokens": 18044028468,
+    "openmath/blocks": 7605000,
+    "openmath/targets": 11751095221,
+    "openmath/input_tokens": 11755011037,
+    "rewrite6/blocks": 3802500,
+    "rewrite6/targets": 6233650055,
+    "rewrite6/input_tokens": 6240484559
   },
   "sources": {
     "climbmix": {
-      "page": 4153,
-      "row": 3572,
+      "page": 4166,
+      "row": 476,
       "start": {
         "file": 201,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
+          "row_group": 54,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 17014260
+      "offset": 0,
+      "documents": 17064412
     },
     "rewrite6": {
-      "page": 13,
-      "row": 316,
+      "page": 20,
+      "row": 812,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 53248
+          "row": 81920
         }
       },
       "epoch": 47,
       "offset": 2048,
-      "documents": 9707552
+      "documents": 9736720
     },
     "openmath": {
-      "page": 2596,
-      "row": 750,
+      "page": 2603,
+      "row": 3612,
       "start": {
         "file": 24,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 153,
-          "row": 617
+          "row_group": 182,
+          "row": 289
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 10633966
+      "offset": 2048,
+      "documents": 10665500
     },
     "ultra_qa": {
-      "page": 1122,
-      "row": 1847,
+      "page": 1125,
+      "row": 3194,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 51467
+          "row": 63755
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4597559
+      "documents": 4611194
     },
     "ultra_style": {
-      "page": 1713,
-      "row": 1089,
+      "page": 1718,
+      "row": 1338,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 81704
+          "row_group": 2,
+          "row": 17547
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 7017537
+      "offset": 4096,
+      "documents": 7038266
     },
     "cortex": {
-      "page": 220582,
-      "row": 48,
+      "page": 221233,
+      "row": 96,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 28234544
+      "documents": 28317920
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 842507, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "6f8382631e83ae42ed9759e16b99adad61f343fa6de9cb2d60b9dd6c9148bc70", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a2b248de903e43cc04b64f31262641a2a596e6720f6e8b83622de3cb8a1ad33a", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "76298eb667981f18a15e74c27926fbcd7a9ae27df4f2ad6d655bc8499e520b76", "coverage.json": "b6f294dc008ba78304b1fa2017cef88912e1ceaaf314ebcebdb825482582a1b3", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000842507"}
+{"step": 845000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "28d525df26bddcdad1f7e0ee21d00c4e2f6f2750838939feccfbef122663248b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "87fc13206a6115414dc5698ba185a42d91483c871e4ad1d988dd31693f1ea24e", "training_config.json": "0ab64376a29947d9a9f670c0ece06c1900ab6f9ec82482957af5fc2c70beec58", "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367", "shard_scheduler.json": "57f3f5db4104672d1623ef748776a5e35f680600431f1b4c0d4f28fc3f966af9", "coverage.json": "e2b9a86150908d61f021a350e36374564851cf14a57027b0b71fa60173ea1a90", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000845000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6f8382631e83ae42ed9759e16b99adad61f343fa6de9cb2d60b9dd6c9148bc70
+oid sha256:28d525df26bddcdad1f7e0ee21d00c4e2f6f2750838939feccfbef122663248b
 size 360622744

performance.json

@@ -49,11 +49,11 @@
   "selected": {
     "microbatch": 90,
     "mode": "compiled",
-    "microbatch_seconds": 0.9411944559999483,
+    "microbatch_seconds": 0.9390849369999614,
     "peak_gb": 89.37790203094482,
     "fits_budget": true,
-    "estimated_update_seconds": 0.9411944559999483,
-    "input_tokens_per_second": 195836.25766704488
+    "estimated_update_seconds": 0.9390849369999614,
+    "input_tokens_per_second": 196276.1756022156
   },
   "measurements": [
     {

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": 76375.6706640414,
-      "required_pages_per_second_upper_estimate": 0.005283846112509921
+      "required_token_positions_per_second": 76039.94462746628,
+      "required_pages_per_second_upper_estimate": 0.005260619806307925
     },
     {
       "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": 9546.958833005176,
-      "required_pages_per_second_upper_estimate": 0.003997112300208492
+      "required_token_positions_per_second": 9504.993078433285,
+      "required_pages_per_second_upper_estimate": 0.003979542115113849
     },
     {
       "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": 19093.91766601035,
-      "required_pages_per_second_upper_estimate": 0.003976759509359474
+      "required_token_positions_per_second": 19009.98615686657,
+      "required_pages_per_second_upper_estimate": 0.003959278789427566
     },
     {
       "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": 28640.876499015525,
-      "required_pages_per_second_upper_estimate": 0.0017169768140297367
+      "required_token_positions_per_second": 28514.979235299852,
+      "required_pages_per_second_upper_estimate": 0.001709429465303973
     },
     {
       "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": 28640.876499015525,
-      "required_pages_per_second_upper_estimate": 0.002226702169896493
+      "required_token_positions_per_second": 28514.979235299852,
+      "required_pages_per_second_upper_estimate": 0.002216914211406139
     },
     {
       "source": "cortex",
@@ -91,9 +91,9 @@
       "read_seconds": 0.04061180899998362,
       "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": 28640.876499015525,
-      "required_pages_per_second_upper_estimate": 0.36737441155212897
+      "estimated_pages_for_buffer": 22,
+      "required_token_positions_per_second": 28514.979235299852,
+      "required_pages_per_second_upper_estimate": 0.3657595366311342
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,412 +155,154 @@
   "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": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.15020428112444506,
-    "producer_p95_seconds": 0.3366278989997227,
-    "gpu_update_seconds": 0.9653335854072185,
-    "producer_batches_per_second": 6.657599853438895,
-    "gpu_batches_per_second": 1.0359113317062907,
-    "producer_headroom": 6.426804736726741,
+    "producer_batch_seconds": 0.05139413005469606,
+    "producer_p95_seconds": 0.15724994399988645,
+    "gpu_update_seconds": 0.969595656088482,
+    "producer_batches_per_second": 19.457474986652226,
+    "gpu_batches_per_second": 1.0313577559063893,
+    "producer_headroom": 18.86588322550829,
     "sustainable": true,
-    "buffer_seconds": 60.81601588065477,
+    "buffer_seconds": 60.11493067748588,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 273762.60107156524,
-    "upload_bytes_per_second": 4419757.457076919,
+    "generation_bytes_per_second": 403027.47604566225,
+    "upload_bytes_per_second": 3951706.6780117266,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 58.74308132276743,
+    "estimated_full_shards_per_hour": 86.48031435992623,
     "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": 1.4560003849983332,
-      "bytes": 6473170,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.6064382549957372,
-      "bytes": 6604109,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.684055683996121,
-      "bytes": 5176791,
-      "pages": 11,
-      "shards": 2
-    },
-    {
-      "seconds": 1.8713450549985282,
-      "bytes": 942860,
-      "pages": 27,
+      "seconds": 1.4923919180000098,
+      "bytes": 40304,
+      "pages": 1,
       "shards": 1
     },
     {
-      "seconds": 1.5147400999994716,
-      "bytes": 6664610,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4935083919990575,
-      "bytes": 2459570,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 2.1903198990039527,
-      "bytes": 16888519,
-      "pages": 41,
-      "shards": 4
-    },
-    {
-      "seconds": 2.158359011998982,
-      "bytes": 7618380,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.546953492994362,
-      "bytes": 982772,
-      "pages": 27,
+      "seconds": 1.3177459699999758,
+      "bytes": 94124,
+      "pages": 4,
       "shards": 1
     },
     {
-      "seconds": 1.5179005070021958,
-      "bytes": 6221682,
-      "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4849864080024417,
-      "bytes": 6331011,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.7072867230017437,
-      "bytes": 11821281,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.688430515001528,
-      "bytes": 5778338,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.6583789540018188,
-      "bytes": 2527741,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.606587074995332,
-      "bytes": 10744249,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0940176939984667,
-      "bytes": 7184499,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0425019949980197,
-      "bytes": 1083508,
-      "pages": 36,
+      "seconds": 1.576328019000016,
+      "bytes": 582303,
+      "pages": 17,
       "shards": 1
     },
     {
-      "seconds": 1.5655531079974025,
-      "bytes": 12067871,
-      "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 1.9511204809969058,
-      "bytes": 6917854,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1665768739985651,
-      "bytes": 5996687,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.6431592789958813,
-      "bytes": 7717315,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3486441710047075,
-      "bytes": 976930,
-      "pages": 30,
+      "seconds": 1.4797563120000632,
+      "bytes": 958250,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.5532393190005678,
-      "bytes": 10410458,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5896227589983027,
-      "bytes": 8024141,
-      "pages": 26,
+      "seconds": 1.1572930810000344,
+      "bytes": 7429293,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.570418814997538,
-      "bytes": 6148151,
-      "pages": 41,
-      "shards": 2
-    },
-    {
-      "seconds": 1.876006619000691,
-      "bytes": 11510011,
-      "pages": 30,
+      "seconds": 1.7505123790000425,
+      "bytes": 15799691,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.1747664850045112,
-      "bytes": 1216277,
-      "pages": 36,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0712760539972805,
-      "bytes": 6301436,
+      "seconds": 1.5947456709999415,
+      "bytes": 2710887,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2888189970035455,
-      "bytes": 7385771,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5881777600006899,
-      "bytes": 5494719,
-      "pages": 25,
-      "shards": 2
-    },
-    {
-      "seconds": 1.394116926996503,
-      "bytes": 11997464,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.7342578780007898,
-      "bytes": 2696654,
-      "pages": 39,
-      "shards": 2
-    },
-    {
-      "seconds": 1.611772108997684,
-      "bytes": 5903925,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.512940188003995,
-      "bytes": 7372289,
+      "seconds": 2.055465744000003,
+      "bytes": 6120157,
       "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9355933799961349,
-      "bytes": 5063118,
-      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.774848651002685,
-      "bytes": 5953016,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.5483520990001125,
+      "bytes": 929612,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.877187440994021,
-      "bytes": 13513556,
-      "pages": 41,
+      "seconds": 1.549194875000012,
+      "bytes": 12173751,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.8802343610004755,
-      "bytes": 5656226,
-      "pages": 32,
+      "seconds": 1.4142575109999598,
+      "bytes": 6051595,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.6476056760002393,
-      "bytes": 7839870,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3661578879982699,
-      "bytes": 1247022,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4103239969990682,
-      "bytes": 5948663,
+      "seconds": 1.509009006000042,
+      "bytes": 6599366,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2626624109980185,
-      "bytes": 7374458,
-      "pages": 32,
+      "seconds": 1.1468410420000055,
+      "bytes": 7711330,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.3394468149999739,
-      "bytes": 10758114,
-      "pages": 29,
+      "seconds": 1.7056902470001205,
+      "bytes": 10320081,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.5901181910012383,
-      "bytes": 260770,
-      "pages": 9,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5258419460005825,
-      "bytes": 5807963,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2447194750056951,
-      "bytes": 7832230,
-      "pages": 41,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9895193189950078,
-      "bytes": 741196,
-      "pages": 23,
-      "shards": 1
-    },
-    {
-      "seconds": 1.489227063000726,
-      "bytes": 11724055,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9323788800029433,
-      "bytes": 1111114,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5997961780012702,
-      "bytes": 11835967,
-      "pages": 35,
+      "seconds": 1.3121002539999154,
+      "bytes": 2619780,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.3631727270039846,
-      "bytes": 2146631,
-      "pages": 21,
+      "seconds": 1.2438676930000838,
+      "bytes": 6205549,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2327626959959161,
-      "bytes": 11052547,
-      "pages": 48,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4408944379974855,
-      "bytes": 7298137,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4694450449969736,
-      "bytes": 1110429,
-      "pages": 36,
+      "seconds": 1.3983366309998928,
+      "bytes": 1085324,
+      "pages": 32,
       "shards": 1
     },
     {
-      "seconds": 1.0526011850015493,
-      "bytes": 6120517,
-      "pages": 31,
-      "shards": 2
-    },
-    {
-      "seconds": 1.7904460849968018,
-      "bytes": 12133874,
-      "pages": 26,
-      "shards": 4
-    },
-    {
-      "seconds": 1.269238109998696,
-      "bytes": 6277214,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.02439347699692,
-      "bytes": 5979878,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.6373932590004188,
+      "bytes": 17208473,
+      "pages": 36,
+      "shards": 5
     },
     {
-      "seconds": 1.1111059189934167,
-      "bytes": 2867400,
-      "pages": 39,
+      "seconds": 1.523682868999913,
+      "bytes": 6058267,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2084345140028745,
-      "bytes": 10491799,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 0.8811795219953638,
-      "bytes": 776994,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.837297693004075,
-      "bytes": 13040260,
-      "pages": 38,
-      "shards": 4
-    },
-    {
-      "seconds": 1.5299789529963164,
-      "bytes": 6406874,
-      "pages": 38,
+      "seconds": 1.493248520999714,
+      "bytes": 2669063,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.7032136210036697,
-      "bytes": 6726184,
-      "pages": 31,
+      "seconds": 1.4799904070000593,
+      "bytes": 10661870,
+      "pages": 35,
       "shards": 3
     }
   ],
@@ -606,7 +348,7 @@
   },
   "cache_reused": true,
   "resume_probe_seconds": [
-    3.328129581999974,
-    0.10924991900014902
+    2.576500048000014,
+    0.012398547999964649
   ]
 }

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2b248de903e43cc04b64f31262641a2a596e6720f6e8b83622de3cb8a1ad33a
-size 721782015
+oid sha256:87fc13206a6115414dc5698ba185a42d91483c871e4ad1d988dd31693f1ea24e
+size 721767423

training_config.json

@@ -53,7 +53,7 @@
   "cpu_threads": 4,
   "page_rows": 4096,
   "cortex_page_rows": 128,
-  "prefetch_batches": 63,
+  "prefetch_batches": 62,
   "prefetch_low_water": 31,
   "memory_pages": 12,
   "cache_gb": 20,