Qwen90M step 710000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 705000,
+  "step": 710000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "c24a63cd2e1cb9edf30d6409a48273ab6c01c49122624fe51ba44135f928c526",
+    "model.safetensors": "528650dcc6ab39aee7440160db0dc09ac752a15db67f68bdafee5cad4fd8ad5e",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "7d0f9cff5ba25bc9ee4e96e42582ae75644b70f80f581eeb59cc1552ebe7c806",
+    "trainer.pt": "d5cbf282cff5336f8a6c08c67961d930cbf65ac0ca4924bf16b44d474c039bd5",
     "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8",
     "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29",
-    "shard_scheduler.json": "92f5af91d86da30a207fc2ede8db0f46e2f61efa3eee22a9c1c2a58df9954b09",
-    "coverage.json": "8a4c9c5871c780362c346a7c8a31c7e57337e0f11d467c16d825aefec83f9598",
+    "shard_scheduler.json": "0f9d138b124eabf61b099cfa198c8250d3e54864bf239b2eea2d45f1ef18a562",
+    "coverage.json": "482dd605a5ff6df5b7d33f82302b75d0d8a33c9cbccd52de27daa8fc4997f99e",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 705000,
+  "step": 710000,
   "stats": {
-    "climbmix/blocks": 25380000,
-    "climbmix/targets": 42427427006,
-    "climbmix/input_tokens": 42430116101,
-    "ultra_qa/blocks": 9517500,
-    "ultra_qa/targets": 15526937657,
-    "ultra_qa/input_tokens": 15526967636,
-    "cortex/blocks": 9517500,
-    "cortex/targets": 361880143,
-    "cortex/input_tokens": 15128656803,
-    "ultra_style/blocks": 9517500,
-    "ultra_style/targets": 15053668730,
-    "ultra_style/input_tokens": 15054458776,
-    "openmath/blocks": 6345000,
-    "openmath/targets": 9804346129,
-    "openmath/input_tokens": 9807612496,
-    "rewrite6/blocks": 3172500,
-    "rewrite6/targets": 5200843675,
-    "rewrite6/input_tokens": 5206545918
+    "climbmix/blocks": 25560000,
+    "climbmix/targets": 42728176177,
+    "climbmix/input_tokens": 42730884305,
+    "ultra_qa/blocks": 9585000,
+    "ultra_qa/targets": 15637043461,
+    "ultra_qa/input_tokens": 15637073679,
+    "cortex/blocks": 9585000,
+    "cortex/targets": 364460015,
+    "cortex/input_tokens": 15235888868,
+    "ultra_style/blocks": 9585000,
+    "ultra_style/targets": 15160318529,
+    "ultra_style/input_tokens": 15161114129,
+    "openmath/blocks": 6390000,
+    "openmath/targets": 9873849425,
+    "openmath/input_tokens": 9877138987,
+    "rewrite6/blocks": 3195000,
+    "rewrite6/targets": 5237762634,
+    "rewrite6/input_tokens": 5243505154
   },
   "sources": {
     "climbmix": {
-      "page": 3476,
-      "row": 737,
+      "page": 3500,
+      "row": 3332,
       "start": {
-        "file": 168,
+        "file": 169,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 19,
+          "row_group": 32,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 14238433
+      "offset": 4096,
+      "documents": 14339332
     },
     "rewrite6": {
-      "page": 27,
-      "row": 2327,
+      "page": 41,
+      "row": 2421,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 110592
+          "row": 167936
         }
       },
       "epoch": 39,
-      "offset": 0,
-      "documents": 8123675
+      "offset": 2048,
+      "documents": 8181113
     },
     "openmath": {
-      "page": 2172,
-      "row": 1176,
+      "page": 2187,
+      "row": 2884,
       "start": {
         "file": 20,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 163,
-          "row": 512
+          "row_group": 224,
+          "row": 952
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 8897688
+      "documents": 8960836
     },
     "ultra_qa": {
-      "page": 939,
-      "row": 1127,
+      "page": 945,
+      "row": 3804,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 15078
+          "row": 39654
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 3847271
+      "offset": 4096,
+      "documents": 3874524
     },
     "ultra_style": {
-      "page": 1433,
-      "row": 3105,
+      "page": 1443,
+      "row": 3759,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 49271
+          "row_group": 5,
+          "row": 4565
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5872673
+      "documents": 5914287
     },
     "cortex": {
-      "page": 184583,
-      "row": 79,
+      "page": 185891,
+      "row": 74,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 23626703
+      "documents": 23794122
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 705000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "c24a63cd2e1cb9edf30d6409a48273ab6c01c49122624fe51ba44135f928c526", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "7d0f9cff5ba25bc9ee4e96e42582ae75644b70f80f581eeb59cc1552ebe7c806", "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8", "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29", "shard_scheduler.json": "92f5af91d86da30a207fc2ede8db0f46e2f61efa3eee22a9c1c2a58df9954b09", "coverage.json": "8a4c9c5871c780362c346a7c8a31c7e57337e0f11d467c16d825aefec83f9598", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac", "folder": "step-000000705000"}
+{"step": 710000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "528650dcc6ab39aee7440160db0dc09ac752a15db67f68bdafee5cad4fd8ad5e", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "d5cbf282cff5336f8a6c08c67961d930cbf65ac0ca4924bf16b44d474c039bd5", "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8", "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29", "shard_scheduler.json": "0f9d138b124eabf61b099cfa198c8250d3e54864bf239b2eea2d45f1ef18a562", "coverage.json": "482dd605a5ff6df5b7d33f82302b75d0d8a33c9cbccd52de27daa8fc4997f99e", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac", "folder": "step-000000710000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c24a63cd2e1cb9edf30d6409a48273ab6c01c49122624fe51ba44135f928c526
+oid sha256:528650dcc6ab39aee7440160db0dc09ac752a15db67f68bdafee5cad4fd8ad5e
 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": 77341.28166839482,
-      "required_pages_per_second_upper_estimate": 0.005350649322317305
+      "required_token_positions_per_second": 77215.80425388893,
+      "required_pages_per_second_upper_estimate": 0.005341968503634073
     },
     {
       "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": 9667.660208549352,
-      "required_pages_per_second_upper_estimate": 0.004047647445617498
+      "required_token_positions_per_second": 9651.975531736116,
+      "required_pages_per_second_upper_estimate": 0.004041080598969093
     },
     {
       "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": 19335.320417098705,
-      "required_pages_per_second_upper_estimate": 0.0040270373361925215
+      "required_token_positions_per_second": 19303.951063472232,
+      "required_pages_per_second_upper_estimate": 0.004020503927097614
     },
     {
       "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": 29002.98062564806,
-      "required_pages_per_second_upper_estimate": 0.0017386844035203698
+      "required_token_positions_per_second": 28955.92659520835,
+      "required_pages_per_second_upper_estimate": 0.0017358635862428536
     },
     {
       "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": 29002.98062564806,
-      "required_pages_per_second_upper_estimate": 0.002254854173014444
+      "required_token_positions_per_second": 28955.92659520835,
+      "required_pages_per_second_upper_estimate": 0.0022511959291165637
     },
     {
       "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": 29002.98062564806,
-      "required_pages_per_second_upper_estimate": 0.3720190944914516
+      "required_token_positions_per_second": 28955.92659520835,
+      "required_pages_per_second_upper_estimate": 0.37141553591165266
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.09527713949219674,
-    "producer_p95_seconds": 0.2191641060016991,
-    "gpu_update_seconds": 0.9532813319038729,
-    "producer_batches_per_second": 10.495697135007928,
-    "gpu_batches_per_second": 1.0490082691568308,
-    "producer_headroom": 10.00535214412002,
+    "producer_batch_seconds": 0.10586103178115991,
+    "producer_p95_seconds": 0.24477108100109035,
+    "gpu_update_seconds": 0.9548304354582531,
+    "producer_batches_per_second": 9.446346622308003,
+    "gpu_batches_per_second": 1.047306372801228,
+    "producer_headroom": 9.019659258867948,
     "sustainable": true,
-    "buffer_seconds": 60.05672390994399,
+    "buffer_seconds": 60.15431743386994,
     "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": 402564.9410295652,
-    "upload_bytes_per_second": 5735685.118812759,
+    "generation_bytes_per_second": 273921.71331875265,
+    "upload_bytes_per_second": 5427861.974460425,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 86.38106511273591,
+    "estimated_full_shards_per_hour": 58.7772231070703,
     "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": 1.1886329499993735,
-      "bytes": 6263136,
+      "seconds": 1.1621864580010879,
+      "bytes": 7532043,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2297442769995541,
+      "bytes": 6251173,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9854559350005729,
-      "bytes": 2421861,
-      "pages": 30,
+      "seconds": 0.9681137329971534,
+      "bytes": 5823551,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.444283180999264,
-      "bytes": 6519870,
-      "pages": 22,
-      "shards": 2
+      "seconds": 1.3083030150010018,
+      "bytes": 7346473,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.324441224000111,
-      "bytes": 10689415,
-      "pages": 37,
+      "seconds": 0.8391288720013108,
+      "bytes": 1133832,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.679156835001777,
+      "bytes": 13329088,
+      "pages": 34,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1372946299998148,
+      "bytes": 9925526,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1898289029995794,
-      "bytes": 7719194,
+      "seconds": 1.0260383269996964,
+      "bytes": 1162438,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.0241040929977316,
+      "bytes": 7727150,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9053547809999145,
-      "bytes": 813842,
-      "pages": 25,
+      "seconds": 1.1232291689993872,
+      "bytes": 903046,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.1792037220002385,
-      "bytes": 7645453,
-      "pages": 39,
+      "seconds": 1.245589108999411,
+      "bytes": 10344850,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0877346289998968,
-      "bytes": 10017417,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.3122096120023343,
+      "bytes": 13394029,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.1130377059998864,
-      "bytes": 986874,
+      "seconds": 1.2143651210026292,
+      "bytes": 938191,
       "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.09502404000159,
-      "bytes": 13537871,
+      "seconds": 1.3705294199971831,
+      "bytes": 12626897,
       "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 0.918569364999712,
-      "bytes": 999510,
-      "pages": 30,
+      "seconds": 1.047090729996853,
+      "bytes": 5981571,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8534065610001562,
+      "bytes": 1138564,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.372887202000129,
-      "bytes": 11719387,
-      "pages": 39,
+      "seconds": 1.0472652070020558,
+      "bytes": 7166131,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1100580310012447,
+      "bytes": 6769711,
+      "pages": 33,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1388595060016087,
+      "bytes": 11609898,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.0358149310013687,
-      "bytes": 6338461,
-      "pages": 36,
+      "seconds": 1.240830022998125,
+      "bytes": 6131822,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8834309400008351,
-      "bytes": 840516,
-      "pages": 27,
+      "seconds": 1.2179747789996327,
+      "bytes": 1206466,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1130535650008824,
-      "bytes": 11588700,
+      "seconds": 1.7537132789984753,
+      "bytes": 11388396,
       "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.0404952170010802,
-      "bytes": 6993608,
-      "pages": 35,
+      "seconds": 1.1776387320023787,
+      "bytes": 5920541,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.010654464000254,
-      "bytes": 6096042,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1592712370002118,
-      "bytes": 7445713,
-      "pages": 36,
+      "seconds": 1.2547966029997042,
+      "bytes": 8617085,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.0245760489997338,
-      "bytes": 5205900,
-      "pages": 31,
+      "seconds": 0.9656906599993818,
+      "bytes": 5770022,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.034200502999738,
-      "bytes": 7563521,
-      "pages": 38,
+      "seconds": 1.14655600399783,
+      "bytes": 10568739,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.312004507999518,
-      "bytes": 5834009,
-      "pages": 26,
+      "seconds": 0.9451345529996615,
+      "bytes": 2348924,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.3078732329995546,
-      "bytes": 7188440,
-      "pages": 42,
+      "seconds": 0.9220264059986221,
+      "bytes": 6175250,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.37187312900096,
-      "bytes": 11914910,
-      "pages": 34,
-      "shards": 4
+      "seconds": 0.8423896300009801,
+      "bytes": 1034182,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 1.1714240479996079,
-      "bytes": 6274938,
+      "seconds": 1.459789966000244,
+      "bytes": 17117955,
       "pages": 37,
-      "shards": 2
+      "shards": 5
     },
     {
-      "seconds": 1.1267517289998068,
-      "bytes": 2389323,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.253374073001396,
-      "bytes": 6015964,
+      "seconds": 1.0440574520034716,
+      "bytes": 6135369,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0633229350005422,
-      "bytes": 4861789,
-      "pages": 27,
+      "seconds": 0.9132081499992637,
+      "bytes": 2381673,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.667441460000191,
-      "bytes": 13590477,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1102066319999722,
-      "bytes": 6253510,
+      "seconds": 1.0147103390008851,
+      "bytes": 6222923,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0856502950009599,
-      "bytes": 2667766,
-      "pages": 38,
+      "seconds": 1.2373420869989786,
+      "bytes": 5156688,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.1392225029994734,
-      "bytes": 10365382,
+      "seconds": 1.0714794720006466,
+      "bytes": 7551415,
       "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.9051334899995709,
-      "bytes": 1149324,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.495498950000183,
-      "bytes": 7289548,
-      "pages": 26,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2270069759997568,
-      "bytes": 11800938,
+      "seconds": 1.4946176230005221,
+      "bytes": 11950661,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0093337090002024,
-      "bytes": 5575178,
+      "seconds": 1.2175075589984772,
+      "bytes": 2434158,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.1225266829987959,
-      "bytes": 7591094,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4399703239996597,
-      "bytes": 6008739,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.370172322000144,
+      "bytes": 11210739,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.3038473750020785,
-      "bytes": 2307657,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.054491422000865,
+      "bytes": 881075,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.5242876569973305,
-      "bytes": 9910776,
-      "pages": 34,
+      "seconds": 1.0468707559994073,
+      "bytes": 7596077,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9998362250007631,
-      "bytes": 6697585,
-      "pages": 27,
+      "seconds": 1.3744402519987489,
+      "bytes": 5869308,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.1621864580010879,
-      "bytes": 7532043,
+      "seconds": 1.611063919001026,
+      "bytes": 11001266,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 2.3035276259979582,
+      "bytes": 7703540,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.2297442769995541,
-      "bytes": 6251173,
-      "pages": 35,
+      "seconds": 1.1081734670005972,
+      "bytes": 5790944,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9681137329971534,
-      "bytes": 5823551,
-      "pages": 34,
+      "seconds": 1.2694779230005224,
+      "bytes": 2738639,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3083030150010018,
-      "bytes": 7346473,
-      "pages": 31,
+      "seconds": 1.369231013002718,
+      "bytes": 10621268,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.8391288720013108,
-      "bytes": 1133832,
-      "pages": 34,
+      "seconds": 0.9135342390000005,
+      "bytes": 1211398,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.679156835001777,
-      "bytes": 13329088,
+      "seconds": 1.2419597409971175,
+      "bytes": 13297063,
       "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.1372946299998148,
-      "bytes": 9925526,
-      "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0260383269996964,
-      "bytes": 1162438,
-      "pages": 35,
+      "seconds": 1.2846820170016144,
+      "bytes": 6246650,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0241040929977316,
-      "bytes": 7727150,
-      "pages": 39,
+      "seconds": 1.021862665998924,
+      "bytes": 6426718,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.1232291689993872,
-      "bytes": 903046,
-      "pages": 26,
-      "shards": 1
+      "seconds": 0.9455101860003197,
+      "bytes": 6297280,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.245589108999411,
-      "bytes": 10344850,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.347191354001552,
+      "bytes": 5982017,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.3122096120023343,
-      "bytes": 13394029,
-      "pages": 40,
-      "shards": 4
+      "seconds": 3.3193765439973504,
+      "bytes": 2461242,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.2143651210026292,
-      "bytes": 938191,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.3424353250011336,
+      "bytes": 15776988,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.3705294199971831,
-      "bytes": 12626897,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.047338389998913,
+      "bytes": 2628784,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.047090729996853,
-      "bytes": 5981571,
-      "pages": 35,
+      "seconds": 1.2893916930006526,
+      "bytes": 6001693,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 0.8534065610001562,
-      "bytes": 1138564,
-      "pages": 34,
+      "seconds": 1.5835260300009395,
+      "bytes": 761598,
+      "pages": 23,
       "shards": 1
     },
     {
-      "seconds": 1.0472652070020558,
-      "bytes": 7166131,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.654032960999757,
+      "bytes": 12595480,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.1100580310012447,
-      "bytes": 6769711,
-      "pages": 33,
+      "seconds": 1.5982389480013808,
+      "bytes": 6084209,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1388595060016087,
-      "bytes": 11609898,
-      "pages": 38,
-      "shards": 4
-    },
-    {
-      "seconds": 1.240830022998125,
-      "bytes": 6131822,
-      "pages": 37,
+      "seconds": 0.98295405799945,
+      "bytes": 6764576,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2179747789996327,
-      "bytes": 1206466,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.2006459389995143,
+      "bytes": 7651087,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.7537132789984753,
-      "bytes": 11388396,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.3339825349976309,
+      "bytes": 10020438,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.1776387320023787,
-      "bytes": 5920541,
-      "pages": 31,
-      "shards": 2
+      "seconds": 0.850007712997467,
+      "bytes": 1050252,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.2547966029997042,
-      "bytes": 8617085,
-      "pages": 41,
+      "seconds": 1.2932650000002468,
+      "bytes": 7701652,
+      "pages": 38,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7d0f9cff5ba25bc9ee4e96e42582ae75644b70f80f581eeb59cc1552ebe7c806
-size 721782079
+oid sha256:d5cbf282cff5336f8a6c08c67961d930cbf65ac0ca4924bf16b44d474c039bd5
+size 721757887