Qwen90M step 100000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 95000,
+  "step": 100000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "ca36386789d96275ea2faa2f3de104514e2db7da233e3abb5b956d3bdb187f3a",
+    "model.safetensors": "7e19e918403cc37f5d1810c36858d079fd9a1abc5258ce8bcadd60f8273dbc27",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "cf0aa922cb78ea80dfb8a6f0763e2ef5f6017853fddd2994932d1a42ae3eb30f",
+    "trainer.pt": "daa6ed4f3041a7768410d18fc4c1955d5f77761c3177da20014d301a73cb18e5",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "7e827680f2cb27108bc66b58bb4d6ede7b8c07fab9033e664cac4bed5acb1a98",
-    "shard_scheduler.json": "f55f959a0031697ec017d1c28a38787189096f27d51840aa325240105f87f743",
-    "coverage.json": "1cece738300372590bb2ffc7f1ac492a978bb75e5d3e82a4fb6dc3ae8bb654b8",
+    "shard_scheduler.json": "bedd197b1b3543492a228dfee98da9e5c2b64ac839bdf45da0854830f0f34e05",
+    "coverage.json": "807feed36aee1f42950d924959cfc4503b766399e37994c3a9760018c4ca6080",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 95000,
+  "step": 100000,
   "stats": {
-    "climbmix/blocks": 3420000,
-    "climbmix/targets": 5714950454,
-    "climbmix/input_tokens": 5715313015,
-    "ultra_qa/blocks": 1282500,
-    "ultra_qa/targets": 2092302344,
-    "ultra_qa/input_tokens": 2092306420,
-    "cortex/blocks": 1282500,
-    "cortex/targets": 48791088,
-    "cortex/input_tokens": 2038487757,
-    "ultra_style/blocks": 1282500,
-    "ultra_style/targets": 2028161062,
-    "ultra_style/input_tokens": 2028267385,
-    "openmath/blocks": 855000,
-    "openmath/targets": 1321142454,
-    "openmath/input_tokens": 1321582983,
-    "rewrite6/blocks": 427500,
-    "rewrite6/targets": 700787253,
-    "rewrite6/input_tokens": 701555776
+    "climbmix/blocks": 3600000,
+    "climbmix/targets": 6015343871,
+    "climbmix/input_tokens": 6015725716,
+    "ultra_qa/blocks": 1350000,
+    "ultra_qa/targets": 2202272463,
+    "ultra_qa/input_tokens": 2202276743,
+    "cortex/blocks": 1350000,
+    "cortex/targets": 51355779,
+    "cortex/input_tokens": 2145762103,
+    "ultra_style/blocks": 1350000,
+    "ultra_style/targets": 2134287106,
+    "ultra_style/input_tokens": 2134399250,
+    "openmath/blocks": 900000,
+    "openmath/targets": 1390635021,
+    "openmath/input_tokens": 1391098627,
+    "rewrite6/blocks": 450000,
+    "rewrite6/targets": 737630929,
+    "rewrite6/input_tokens": 738439976
   },
   "sources": {
     "climbmix": {
-      "page": 468,
-      "row": 3082,
+      "page": 493,
+      "row": 2616,
       "start": {
-        "file": 22,
+        "file": 23,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 50,
+          "row_group": 66,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 1920010
+      "offset": 2048,
+      "documents": 2021944
     },
     "rewrite6": {
-      "page": 16,
-      "row": 2217,
+      "page": 30,
+      "row": 2677,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 65536
+          "row": 122880
         }
       },
       "epoch": 5,
-      "offset": 2048,
-      "documents": 1094773
+      "offset": 0,
+      "documents": 1152577
     },
     "openmath": {
-      "page": 292,
-      "row": 3373,
+      "page": 308,
+      "row": 982,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 322,
-          "row": 732
+          "row_group": 388,
+          "row": 268
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 1199405
+      "documents": 1262550
     },
     "ultra_qa": {
-      "page": 126,
-      "row": 2497,
+      "page": 133,
+      "row": 1106,
       "start": {
-        "file": 0,
+        "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 7,
-          "row": 64065
+          "row_group": 0,
+          "row": 25141
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 518593
+      "documents": 545874
     },
     "ultra_style": {
-      "page": 193,
-      "row": 966,
+      "page": 203,
+      "row": 2560,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 20912
+          "row": 61872
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 791494
+      "documents": 834048
     },
     "cortex": {
-      "page": 24868,
-      "row": 3,
+      "page": 26176,
+      "row": 120,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 3183107
+      "documents": 3350648
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 95000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "ca36386789d96275ea2faa2f3de104514e2db7da233e3abb5b956d3bdb187f3a", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "cf0aa922cb78ea80dfb8a6f0763e2ef5f6017853fddd2994932d1a42ae3eb30f", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "7e827680f2cb27108bc66b58bb4d6ede7b8c07fab9033e664cac4bed5acb1a98", "shard_scheduler.json": "f55f959a0031697ec017d1c28a38787189096f27d51840aa325240105f87f743", "coverage.json": "1cece738300372590bb2ffc7f1ac492a978bb75e5d3e82a4fb6dc3ae8bb654b8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000095000"}
+{"step": 100000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "7e19e918403cc37f5d1810c36858d079fd9a1abc5258ce8bcadd60f8273dbc27", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "daa6ed4f3041a7768410d18fc4c1955d5f77761c3177da20014d301a73cb18e5", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "7e827680f2cb27108bc66b58bb4d6ede7b8c07fab9033e664cac4bed5acb1a98", "shard_scheduler.json": "bedd197b1b3543492a228dfee98da9e5c2b64ac839bdf45da0854830f0f34e05", "coverage.json": "807feed36aee1f42950d924959cfc4503b766399e37994c3a9760018c4ca6080", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000100000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ca36386789d96275ea2faa2f3de104514e2db7da233e3abb5b956d3bdb187f3a
+oid sha256:7e19e918403cc37f5d1810c36858d079fd9a1abc5258ce8bcadd60f8273dbc27
 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": 75731.21103212437,
-      "required_pages_per_second_upper_estimate": 0.0065641894332649706
+      "required_token_positions_per_second": 75313.11788406817,
+      "required_pages_per_second_upper_estimate": 0.006527950178838855
     },
     {
       "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": 9466.401379015546,
-      "required_pages_per_second_upper_estimate": 0.0037762890454027228
+      "required_token_positions_per_second": 9414.139735508521,
+      "required_pages_per_second_upper_estimate": 0.003755441094426568
     },
     {
       "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": 18932.80275803109,
-      "required_pages_per_second_upper_estimate": 0.004379770712122348
+      "required_token_positions_per_second": 18828.279471017042,
+      "required_pages_per_second_upper_estimate": 0.0043555910892186704
     },
     {
       "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": 28399.20413704664,
-      "required_pages_per_second_upper_estimate": 0.0017293625328372911
+      "required_token_positions_per_second": 28242.419206525563,
+      "required_pages_per_second_upper_estimate": 0.0017198151531555152
     },
     {
       "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": 28399.20413704664,
-      "required_pages_per_second_upper_estimate": 0.002186249348498423
+      "required_token_positions_per_second": 28242.419206525563,
+      "required_pages_per_second_upper_estimate": 0.0021741796105384454
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "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": 28399.20413704664,
-      "required_pages_per_second_upper_estimate": 0.352623069360004
+      "required_token_positions_per_second": 28242.419206525563,
+      "required_pages_per_second_upper_estimate": 0.35067632524834
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,141 +158,27 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.19920030615637074,
-    "producer_p95_seconds": 0.24541962999865063,
-    "gpu_update_seconds": 0.9735484088419684,
-    "producer_batches_per_second": 5.0200726057870995,
-    "gpu_batches_per_second": 1.027170288521652,
-    "producer_headroom": 4.887283697635184,
+    "producer_batch_seconds": 0.11091515512487149,
+    "producer_p95_seconds": 0.30749590000050375,
+    "gpu_update_seconds": 0.978952964256397,
+    "producer_batches_per_second": 9.01590047702833,
+    "gpu_batches_per_second": 1.02149953727306,
+    "producer_headroom": 8.826142497427547,
     "sustainable": true,
-    "buffer_seconds": 60.36000134820204,
+    "buffer_seconds": 60.695083783896614,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 442214.1127743015,
-    "upload_bytes_per_second": 5984337.118994301,
+    "generation_bytes_per_second": 359571.1211645759,
+    "upload_bytes_per_second": 6007747.908546246,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 94.88885438367637,
+    "estimated_full_shards_per_hour": 77.15559221461257,
     "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": 0.9955667570011428,
-      "bytes": 43764,
-      "pages": 1,
-      "shards": 1
-    },
-    {
-      "seconds": 0.9273372809984721,
-      "bytes": 2566992,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1000417619998188,
-      "bytes": 10096800,
-      "pages": 18,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2823710859993298,
-      "bytes": 6954773,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0888818870007526,
-      "bytes": 7323944,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.025877304999085,
-      "bytes": 6206617,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8778860090005765,
-      "bytes": 2613270,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1979163849991892,
-      "bytes": 9677162,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9057083049992798,
-      "bytes": 1081901,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.247454252999887,
-      "bytes": 13520201,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9827529990016046,
-      "bytes": 788256,
-      "pages": 22,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1497582369993324,
-      "bytes": 10983620,
-      "pages": 42,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0362804550004512,
-      "bytes": 7507968,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9569115700014663,
-      "bytes": 1242347,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.07777969900053,
-      "bytes": 7140650,
-      "pages": 25,
-      "shards": 3
-    },
-    {
-      "seconds": 1.360217057999762,
-      "bytes": 16210349,
-      "pages": 37,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9705395430009958,
-      "bytes": 1195413,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1174536390008143,
-      "bytes": 7647354,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2463584760007507,
-      "bytes": 4721254,
-      "pages": 27,
-      "shards": 2
-    },
     {
       "seconds": 1.214833362000718,
       "bytes": 7500577,
@@ -430,6 +316,252 @@
       "bytes": 7549095,
       "pages": 36,
       "shards": 3
+    },
+    {
+      "seconds": 1.0992059410018555,
+      "bytes": 4757047,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9332335270009935,
+      "bytes": 581892,
+      "pages": 17,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1391355480009224,
+      "bytes": 7817543,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0901677820002078,
+      "bytes": 6115749,
+      "pages": 34,
+      "shards": 2
+    },
+    {
+      "seconds": 1.5552335219981615,
+      "bytes": 12252293,
+      "pages": 37,
+      "shards": 4
+    },
+    {
+      "seconds": 0.9495614830011618,
+      "bytes": 5780919,
+      "pages": 27,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0740993799990974,
+      "bytes": 5965207,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9706554090007558,
+      "bytes": 2646898,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.0982734540011734,
+      "bytes": 10287883,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8711374440026702,
+      "bytes": 2523849,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.5673655419996066,
+      "bytes": 11589450,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.247284810997371,
+      "bytes": 6017710,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9974236470006872,
+      "bytes": 6556698,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0323645409989695,
+      "bytes": 5955575,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8832039090011676,
+      "bytes": 6189230,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9038558969987207,
+      "bytes": 2334921,
+      "pages": 27,
+      "shards": 2
+    },
+    {
+      "seconds": 1.410665313997015,
+      "bytes": 15644643,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 0.8785766789987974,
+      "bytes": 2610651,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.0569672790006734,
+      "bytes": 5933537,
+      "pages": 27,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8821923329996935,
+      "bytes": 961115,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2227748900004372,
+      "bytes": 12637313,
+      "pages": 46,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1576985259998764,
+      "bytes": 6035027,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.131166884999402,
+      "bytes": 11754261,
+      "pages": 32,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3580378289989312,
+      "bytes": 2581856,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.6505394759988121,
+      "bytes": 9887447,
+      "pages": 28,
+      "shards": 3
+    },
+    {
+      "seconds": 0.7642709539977659,
+      "bytes": 2553358,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2285513930000889,
+      "bytes": 6249359,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8377171489992179,
+      "bytes": 5823773,
+      "pages": 30,
+      "shards": 2
+    },
+    {
+      "seconds": 1.6930653950003034,
+      "bytes": 12802437,
+      "pages": 34,
+      "shards": 4
+    },
+    {
+      "seconds": 1.2719050730011077,
+      "bytes": 5991225,
+      "pages": 33,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9874920929978543,
+      "bytes": 1116876,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0614115569987916,
+      "bytes": 7614945,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0503759480016015,
+      "bytes": 9758074,
+      "pages": 29,
+      "shards": 3
+    },
+    {
+      "seconds": 1.035596928002633,
+      "bytes": 2562080,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3296433830000751,
+      "bytes": 11732086,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8983965400002489,
+      "bytes": 906815,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.218643701999099,
+      "bytes": 11866314,
+      "pages": 38,
+      "shards": 5
+    },
+    {
+      "seconds": 1.0883352769997146,
+      "bytes": 6129558,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.048354777998611,
+      "bytes": 2655724,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8924994240005617,
+      "bytes": 5699054,
+      "pages": 19,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2686124620013288,
+      "bytes": 11642195,
+      "pages": 44,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cf0aa922cb78ea80dfb8a6f0763e2ef5f6017853fddd2994932d1a42ae3eb30f
-size 721772287
+oid sha256:daa6ed4f3041a7768410d18fc4c1955d5f77761c3177da20014d301a73cb18e5
+size 721768255