Qwen90M step 70000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 65000,
+  "step": 70000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "e33da1dad32ae2433103ec21715c2d542c98dc7e7caa2f82997e62b6fe8342d2",
+    "model.safetensors": "1c5054d917de2693f770bc99e564913ba6cc2edd42ed1e61bc5076094a4f4cc1",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "e3e3dd01665bea5cf593b73f4417e92f7e07fdb4e1949ec1cfe309333ab911ee",
-    "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
+    "trainer.pt": "7ca837f5b08d05bbf41cfc5f430f7d7929aa5dfd920fa6c66696541123b2d2a1",
+    "training_config.json": "a9be8fd9bf1c8d33aae0c56259fc899809df0d3ff60285ef0a4c2ea73471632c",
     "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f",
-    "shard_scheduler.json": "fb85ef21cda081906908c01fe4de71830aa788514ed6534050e7da541b10e3e2",
-    "coverage.json": "f8f337b02bacae59e68d5dbc5eb4b6339c704cf53f376b91e8950e701871a03e",
+    "shard_scheduler.json": "eb1e6871a1097af2c4c674b06e837fee7074d58979d436eda419016044a46189",
+    "coverage.json": "1b3adc06f707cf80a7d00bfb046d62b635ee1b8cd68ef5217768a4182ba6132e",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 65000,
+  "step": 70000,
   "stats": {
-    "climbmix/blocks": 2340000,
-    "climbmix/targets": 3910377700,
-    "climbmix/input_tokens": 3910625311,
-    "ultra_qa/blocks": 877500,
-    "ultra_qa/targets": 1431685811,
-    "ultra_qa/input_tokens": 1431688622,
-    "cortex/blocks": 877500,
-    "cortex/targets": 33383362,
-    "cortex/input_tokens": 1394824403,
-    "ultra_style/blocks": 877500,
-    "ultra_style/targets": 1387878431,
-    "ultra_style/input_tokens": 1387950984,
-    "openmath/blocks": 585000,
-    "openmath/targets": 903936284,
-    "openmath/input_tokens": 904237672,
-    "rewrite6/blocks": 292500,
-    "rewrite6/targets": 479420381,
-    "rewrite6/input_tokens": 479946209
+    "climbmix/blocks": 2520000,
+    "climbmix/targets": 4211257891,
+    "climbmix/input_tokens": 4211524900,
+    "ultra_qa/blocks": 945000,
+    "ultra_qa/targets": 1541761818,
+    "ultra_qa/input_tokens": 1541764859,
+    "cortex/blocks": 945000,
+    "cortex/targets": 35955650,
+    "cortex/input_tokens": 1502146727,
+    "ultra_style/blocks": 945000,
+    "ultra_style/targets": 1494609054,
+    "ultra_style/input_tokens": 1494687200,
+    "openmath/blocks": 630000,
+    "openmath/targets": 973552847,
+    "openmath/input_tokens": 973877515,
+    "rewrite6/blocks": 315000,
+    "rewrite6/targets": 516437914,
+    "rewrite6/input_tokens": 517004193
   },
   "sources": {
     "climbmix": {
-      "page": 320,
-      "row": 1789,
+      "page": 345,
+      "row": 1005,
       "start": {
-        "file": 15,
+        "file": 16,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 39,
+          "row_group": 56,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 1312509
+      "offset": 10240,
+      "documents": 1414125
     },
     "rewrite6": {
-      "page": 32,
-      "row": 1861,
+      "page": 46,
+      "row": 2052,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 131072
+          "row": 188416
         }
       },
       "epoch": 3,
       "offset": 0,
-      "documents": 749145
+      "documents": 806680
     },
     "openmath": {
-      "page": 200,
-      "row": 1224,
+      "page": 215,
+      "row": 3124,
       "start": {
-        "file": 1,
+        "file": 2,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 382,
-          "row": 550
+          "row_group": 7,
+          "row": 340
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 820424
+      "documents": 883764
     },
     "ultra_qa": {
-      "page": 86,
-      "row": 2631,
+      "page": 93,
+      "row": 1286,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 29519
+          "row": 58191
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 354887
+      "offset": 4096,
+      "documents": 382214
     },
     "ultra_style": {
-      "page": 132,
-      "row": 470,
+      "page": 142,
+      "row": 1102,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 30335
+          "row": 71295
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 541142
+      "offset": 0,
+      "documents": 582734
     },
     "cortex": {
-      "page": 17014,
-      "row": 1,
+      "page": 18324,
+      "row": 41,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 2177793
+      "documents": 2345513
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 65000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e33da1dad32ae2433103ec21715c2d542c98dc7e7caa2f82997e62b6fe8342d2", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "e3e3dd01665bea5cf593b73f4417e92f7e07fdb4e1949ec1cfe309333ab911ee", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "fb85ef21cda081906908c01fe4de71830aa788514ed6534050e7da541b10e3e2", "coverage.json": "f8f337b02bacae59e68d5dbc5eb4b6339c704cf53f376b91e8950e701871a03e", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000065000"}
+{"step": 70000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "1c5054d917de2693f770bc99e564913ba6cc2edd42ed1e61bc5076094a4f4cc1", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "7ca837f5b08d05bbf41cfc5f430f7d7929aa5dfd920fa6c66696541123b2d2a1", "training_config.json": "a9be8fd9bf1c8d33aae0c56259fc899809df0d3ff60285ef0a4c2ea73471632c", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "eb1e6871a1097af2c4c674b06e837fee7074d58979d436eda419016044a46189", "coverage.json": "1b3adc06f707cf80a7d00bfb046d62b635ee1b8cd68ef5217768a4182ba6132e", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000070000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e33da1dad32ae2433103ec21715c2d542c98dc7e7caa2f82997e62b6fe8342d2
+oid sha256:1c5054d917de2693f770bc99e564913ba6cc2edd42ed1e61bc5076094a4f4cc1
 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": 75748.15161257412,
-      "required_pages_per_second_upper_estimate": 0.006565657799842846
+      "required_token_positions_per_second": 73286.81074761767,
+      "required_pages_per_second_upper_estimate": 0.006352315011879811
     },
     {
       "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": 9468.518951571765,
-      "required_pages_per_second_upper_estimate": 0.0037771337767559298
+      "required_token_positions_per_second": 9160.851343452208,
+      "required_pages_per_second_upper_estimate": 0.0036544005678363686
     },
     {
       "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": 18937.03790314353,
-      "required_pages_per_second_upper_estimate": 0.0043807504384081025
+      "required_token_positions_per_second": 18321.702686904417,
+      "required_pages_per_second_upper_estimate": 0.004238403465661115
     },
     {
       "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": 28405.556854715294,
-      "required_pages_per_second_upper_estimate": 0.0017297493800131785
+      "required_token_positions_per_second": 27482.554030356623,
+      "required_pages_per_second_upper_estimate": 0.0016735433506312972
     },
     {
       "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": 28405.556854715294,
-      "required_pages_per_second_upper_estimate": 0.0021867383982900044
+      "required_token_positions_per_second": 27482.554030356623,
+      "required_pages_per_second_upper_estimate": 0.0021156830858355268
     },
     {
       "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": 28405.556854715294,
-      "required_pages_per_second_upper_estimate": 0.3527019488522703
+      "estimated_pages_for_buffer": 21,
+      "required_token_positions_per_second": 27482.554030356623,
+      "required_pages_per_second_upper_estimate": 0.3412413428654733
     }
   ],
   "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": 62,
+    "prefetch_batches": 60,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.11656077062494319,
-    "producer_p95_seconds": 0.29636224299974856,
-    "gpu_update_seconds": 0.9733306810850448,
-    "producer_batches_per_second": 8.579215757055119,
-    "gpu_batches_per_second": 1.0274000598493669,
-    "producer_headroom": 8.350413915990007,
+    "producer_batch_seconds": 0.43211702229669413,
+    "producer_p95_seconds": 0.33367446499687503,
+    "gpu_update_seconds": 1.0060200361822496,
+    "producer_batches_per_second": 2.314187936140581,
+    "gpu_batches_per_second": 0.994015987787783,
+    "producer_headroom": 2.328119431248673,
     "sustainable": true,
-    "buffer_seconds": 60.34650222727278,
+    "buffer_seconds": 60.36120217093497,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 86,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 512873.7689111375,
-    "upload_bytes_per_second": 4674834.667993678,
+    "generation_bytes_per_second": 266264.8417269639,
+    "upload_bytes_per_second": 3764138.0954640657,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 110.05077171803086,
+    "estimated_full_shards_per_hour": 57.13423670632064,
     "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": 1.406286874000216,
-      "bytes": 18261180,
-      "pages": 44,
-      "shards": 5
-    },
-    {
-      "seconds": 1.0190579750014876,
-      "bytes": 5885831,
+      "seconds": 1.0164730639990012,
+      "bytes": 6038957,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.8101317319997179,
-      "bytes": 713378,
-      "pages": 25,
-      "shards": 2
+      "seconds": 0.9485057740021148,
+      "bytes": 880447,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.03739196099923,
-      "bytes": 7802740,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0207911869983946,
+      "bytes": 13578331,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.1620692619981128,
-      "bytes": 6079962,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.2704945900004532,
+      "bytes": 9798343,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 0.8016836730021168,
-      "bytes": 4650466,
-      "pages": 32,
+      "seconds": 0.7868170459987596,
+      "bytes": 2512813,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2261960830001044,
-      "bytes": 13273679,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.0122808699998131,
+      "bytes": 6141256,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.7825120570014406,
-      "bytes": 1093241,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.8550289739978325,
+      "bytes": 5835711,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.4853517330011528,
-      "bytes": 11092996,
-      "pages": 24,
-      "shards": 4
+      "seconds": 1.051623090999783,
+      "bytes": 7071478,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 0.8748619340003643,
-      "bytes": 6384371,
-      "pages": 42,
-      "shards": 2
+      "seconds": 1.1867464049973933,
+      "bytes": 12128568,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9224505880010838,
-      "bytes": 1117951,
+      "seconds": 0.8681122560010408,
+      "bytes": 1227302,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.0108905939996475,
-      "bytes": 7845615,
-      "pages": 36,
+      "seconds": 1.205836907000048,
+      "bytes": 7335325,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.27624071000173,
-      "bytes": 16075425,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.085818534000282,
+      "bytes": 10112362,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 0.8457192399982887,
-      "bytes": 2628229,
+      "seconds": 0.927039637001144,
+      "bytes": 2682847,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9226963619985327,
-      "bytes": 5611919,
-      "pages": 27,
+      "seconds": 0.8612236650005798,
+      "bytes": 6230072,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.7924021649996575,
-      "bytes": 1122256,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0601522930010105,
+      "bytes": 12072218,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.1671633029982331,
-      "bytes": 11363880,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.126622554998903,
+      "bytes": 6643361,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.9065282010014926,
-      "bytes": 6177542,
-      "pages": 28,
+      "seconds": 0.8856843470020976,
+      "bytes": 5828359,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.0971184539994283,
-      "bytes": 8482200,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0851010650003445,
+      "bytes": 1256653,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.1112909670009685,
-      "bytes": 6011206,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.3800626919983188,
+      "bytes": 7827681,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.9119941530007054,
-      "bytes": 5747516,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.5110167620005086,
+      "bytes": 10255842,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.016742787000112,
-      "bytes": 7467282,
+      "seconds": 8.627843940001185,
+      "bytes": 8388512,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.8858778819994768,
-      "bytes": 5985432,
-      "pages": 27,
+      "seconds": 1.2330220080002618,
+      "bytes": 6137294,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.7830291239988583,
-      "bytes": 1207474,
-      "pages": 35,
+      "seconds": 1.6848288789988146,
+      "bytes": 823777,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.2103096939972602,
-      "bytes": 12749523,
-      "pages": 22,
-      "shards": 4
+      "seconds": 1.0912059820002469,
+      "bytes": 11909059,
+      "pages": 36,
+      "shards": 5
     },
     {
-      "seconds": 0.8322358400000667,
-      "bytes": 4694947,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.1786047089990461,
+      "bytes": 318780,
+      "pages": 9,
+      "shards": 1
     },
     {
-      "seconds": 0.8593272979996982,
-      "bytes": 6225685,
+      "seconds": 0.9711113170014869,
+      "bytes": 6125935,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9578338470018934,
-      "bytes": 2306768,
+      "seconds": 1.0254750570020406,
+      "bytes": 8509608,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 15.209589787002187,
+      "bytes": 6010603,
       "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.8912073430001328,
-      "bytes": 6007286,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.505151831999683,
+      "bytes": 12552564,
+      "pages": 42,
+      "shards": 4
+    },
+    {
+      "seconds": 13.275076857000386,
+      "bytes": 1045844,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.088943154001754,
-      "bytes": 10590685,
+      "seconds": 2.272034231998987,
+      "bytes": 5974861,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.9887308050019783,
-      "bytes": 8510426,
-      "pages": 40,
+      "seconds": 1.2107169869996142,
+      "bytes": 7639876,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0075569920009002,
-      "bytes": 5909692,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.2838386629991874,
+      "bytes": 10633493,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 0.7540812590013957,
-      "bytes": 2508470,
-      "pages": 31,
+      "seconds": 1.5226552149979398,
+      "bytes": 6060968,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.3975003769992327,
-      "bytes": 10375015,
-      "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0634488309988228,
-      "bytes": 6386527,
+      "seconds": 1.020731942000566,
+      "bytes": 2578449,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 14.660640815000079,
-      "bytes": 2712693,
-      "pages": 36,
+      "seconds": 1.7385529189996305,
+      "bytes": 5799908,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 3.9525413480005227,
-      "bytes": 15333188,
+      "seconds": 1.6018576979986392,
+      "bytes": 11859045,
       "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 2.024789704999421,
-      "bytes": 1170220,
-      "pages": 34,
+      "seconds": 0.8893314249980904,
+      "bytes": 1064415,
+      "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.2876270810011192,
-      "bytes": 7494644,
-      "pages": 37,
+      "seconds": 1.0851846760015178,
+      "bytes": 11839223,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.8472818170012033,
-      "bytes": 6269698,
-      "pages": 35,
+      "seconds": 1.710380387998157,
+      "bytes": 2330946,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9710190369987686,
-      "bytes": 6743419,
-      "pages": 30,
+      "seconds": 1.155976900998212,
+      "bytes": 10306222,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0164730639990012,
-      "bytes": 6038957,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.1758935670004576,
+      "bytes": 7725078,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.9485057740021148,
-      "bytes": 880447,
+      "seconds": 1.0130331639993528,
+      "bytes": 839053,
       "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.0207911869983946,
-      "bytes": 13578331,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.1233831270001247,
+      "bytes": 6290212,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.2704945900004532,
-      "bytes": 9798343,
-      "pages": 31,
+      "seconds": 0.8791882089972205,
+      "bytes": 6587099,
+      "pages": 27,
       "shards": 3
     },
     {
-      "seconds": 0.7868170459987596,
-      "bytes": 2512813,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0528832150012022,
+      "bytes": 11948549,
+      "pages": 44,
+      "shards": 3
     },
     {
-      "seconds": 1.0122808699998131,
-      "bytes": 6141256,
+      "seconds": 1.0517111139997724,
+      "bytes": 6085549,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8550289739978325,
-      "bytes": 5835711,
+      "seconds": 0.9597967700028676,
+      "bytes": 2343290,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.051623090999783,
-      "bytes": 7071478,
-      "pages": 33,
+      "seconds": 1.0771120079989487,
+      "bytes": 10854426,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1867464049973933,
-      "bytes": 12128568,
-      "pages": 40,
+      "seconds": 1.2815551900021092,
+      "bytes": 7356201,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.8681122560010408,
-      "bytes": 1227302,
-      "pages": 34,
+      "seconds": 1.2021151590015506,
+      "bytes": 960345,
+      "pages": 28,
       "shards": 1
     },
     {
-      "seconds": 1.205836907000048,
-      "bytes": 7335325,
-      "pages": 30,
+      "seconds": 3.8789173690020107,
+      "bytes": 11941335,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.085818534000282,
-      "bytes": 10112362,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.0000528349992237,
+      "bytes": 11258395,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 0.927039637001144,
-      "bytes": 2682847,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.9514743170002475,
+      "bytes": 822769,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 0.8612236650005798,
-      "bytes": 6230072,
-      "pages": 34,
-      "shards": 2
+      "seconds": 4.258936992002418,
+      "bytes": 6104340,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.0601522930010105,
-      "bytes": 12072218,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.0792534860011074,
+      "bytes": 2442489,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.126622554998903,
-      "bytes": 6643361,
+      "seconds": 1.0146635470009642,
+      "bytes": 11041931,
       "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.8856843470020976,
-      "bytes": 5828359,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.0869366869992518,
+      "bytes": 13432151,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.0851010650003445,
-      "bytes": 1256653,
-      "pages": 34,
+      "seconds": 0.7862899389983795,
+      "bytes": 1239440,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.3800626919983188,
-      "bytes": 7827681,
-      "pages": 39,
-      "shards": 3
+      "seconds": 0.8729745059972629,
+      "bytes": 5859266,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.5110167620005086,
-      "bytes": 10255842,
-      "pages": 33,
+      "seconds": 1.0330235869987519,
+      "bytes": 6506267,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 8.627843940001185,
-      "bytes": 8388512,
-      "pages": 37,
+      "seconds": 1.471190406999085,
+      "bytes": 6116938,
+      "pages": 27,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1064680679992307,
+      "bytes": 7358899,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.2330220080002618,
-      "bytes": 6137294,
+      "seconds": 1.2355712200005655,
+      "bytes": 6983067,
       "pages": 36,
       "shards": 2
-    },
-    {
-      "seconds": 1.6848288789988146,
-      "bytes": 823777,
-      "pages": 25,
-      "shards": 1
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e3e3dd01665bea5cf593b73f4417e92f7e07fdb4e1949ec1cfe309333ab911ee
-size 721787647
+oid sha256:7ca837f5b08d05bbf41cfc5f430f7d7929aa5dfd920fa6c66696541123b2d2a1
+size 721775935

training_config.json

@@ -46,8 +46,8 @@
   "cpu_threads": 4,
   "page_rows": 4096,
   "cortex_page_rows": 128,
-  "prefetch_batches": 62,
-  "prefetch_low_water": 31,
+  "prefetch_batches": 60,
+  "prefetch_low_water": 30,
   "memory_pages": 12,
   "cache_gb": 20,
   "force_rebenchmark": false,