Qwen90M step 640000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 635000,
+  "step": 640000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "aee83f47a1033d9e4d5e4cf38362bb2e073804129f1b99ebd3e435b22ca3f85c",
+    "model.safetensors": "ebc22606ea875e024664b2b5ba82033f919f8d1a96f22dfebda33a7cb5296c83",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "4a95f31929a892cce3d118cf8d73e75858c30e57cad5a3ec036a4b699b0452b6",
+    "trainer.pt": "2e88f84323ca4bdc4a8a0fa1302e353f3a442dcb7967001b5b87a164915124f0",
     "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb",
     "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e",
-    "shard_scheduler.json": "b4388fd71b8b8bba9cd013a154900d67eb794933f8eaad21782341991d4d34b5",
-    "coverage.json": "73ad8690a37dbe8bda15b94f41cc2fef8466db1737bf7d9eba4081b860f31f0c",
+    "shard_scheduler.json": "bf475a77a1c1a0c722213e8028f3d9758aacd623f550f0da5d55fe94219c3d81",
+    "coverage.json": "33bc21ae73a3caaa60d3061ffb19116a58b5eaf2262d82489a7714a0458299ba",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 635000,
+  "step": 640000,
   "stats": {
-    "climbmix/blocks": 22860000,
-    "climbmix/targets": 38215482464,
-    "climbmix/input_tokens": 38217903896,
-    "ultra_qa/blocks": 8572500,
-    "ultra_qa/targets": 13985741427,
-    "ultra_qa/input_tokens": 13985768433,
-    "cortex/blocks": 8572500,
-    "cortex/targets": 325969268,
-    "cortex/input_tokens": 13626603924,
-    "ultra_style/blocks": 8572500,
-    "ultra_style/targets": 13558916090,
-    "ultra_style/input_tokens": 13559627493,
-    "openmath/blocks": 5715000,
-    "openmath/targets": 8830719600,
-    "openmath/input_tokens": 8833661278,
-    "rewrite6/blocks": 2857500,
-    "rewrite6/targets": 4684417110,
-    "rewrite6/input_tokens": 4689553106
+    "climbmix/blocks": 23040000,
+    "climbmix/targets": 38516782526,
+    "climbmix/input_tokens": 38519223105,
+    "ultra_qa/blocks": 8640000,
+    "ultra_qa/targets": 14095882783,
+    "ultra_qa/input_tokens": 14095910016,
+    "cortex/blocks": 8640000,
+    "cortex/targets": 328530053,
+    "cortex/input_tokens": 13733804797,
+    "ultra_style/blocks": 8640000,
+    "ultra_style/targets": 13665830277,
+    "ultra_style/input_tokens": 13666547133,
+    "openmath/blocks": 5760000,
+    "openmath/targets": 8900224915,
+    "openmath/input_tokens": 8903189812,
+    "rewrite6/blocks": 2880000,
+    "rewrite6/targets": 4721409048,
+    "rewrite6/input_tokens": 4726585429
   },
   "sources": {
     "climbmix": {
-      "page": 3130,
-      "row": 2106,
+      "page": 3155,
+      "row": 394,
       "start": {
-        "file": 151,
+        "file": 152,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 42,
+          "row_group": 60,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 12822586
+      "documents": 12923274
     },
     "rewrite6": {
-      "page": 31,
-      "row": 886,
+      "page": 45,
+      "row": 1043,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 126976
+          "row": 184320
         }
       },
       "epoch": 35,
       "offset": 2048,
-      "documents": 7317002
+      "documents": 7374503
     },
     "openmath": {
-      "page": 1956,
-      "row": 1912,
+      "page": 1971,
+      "row": 3641,
       "start": {
         "file": 18,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 152,
-          "row": 76
+          "row_group": 213,
+          "row": 516
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 8013688
+      "documents": 8076857
     },
     "ultra_qa": {
-      "page": 846,
-      "row": 64,
+      "page": 852,
+      "row": 2758,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 24839
+          "row": 49415
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3465280
+      "documents": 3492550
     },
     "ultra_style": {
-      "page": 1291,
-      "row": 1369,
+      "page": 1301,
+      "row": 1747,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 5,
-          "row": 66925
+          "row_group": 6,
+          "row": 23328
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 5289305
+      "offset": 0,
+      "documents": 5330643
     },
     "cortex": {
-      "page": 166257,
-      "row": 71,
+      "page": 167566,
+      "row": 31,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 21280967
+      "documents": 21448479
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 635000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "aee83f47a1033d9e4d5e4cf38362bb2e073804129f1b99ebd3e435b22ca3f85c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "4a95f31929a892cce3d118cf8d73e75858c30e57cad5a3ec036a4b699b0452b6", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e", "shard_scheduler.json": "b4388fd71b8b8bba9cd013a154900d67eb794933f8eaad21782341991d4d34b5", "coverage.json": "73ad8690a37dbe8bda15b94f41cc2fef8466db1737bf7d9eba4081b860f31f0c", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000635000"}
+{"step": 640000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "ebc22606ea875e024664b2b5ba82033f919f8d1a96f22dfebda33a7cb5296c83", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "2e88f84323ca4bdc4a8a0fa1302e353f3a442dcb7967001b5b87a164915124f0", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e", "shard_scheduler.json": "bf475a77a1c1a0c722213e8028f3d9758aacd623f550f0da5d55fe94219c3d81", "coverage.json": "33bc21ae73a3caaa60d3061ffb19116a58b5eaf2262d82489a7714a0458299ba", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000640000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aee83f47a1033d9e4d5e4cf38362bb2e073804129f1b99ebd3e435b22ca3f85c
+oid sha256:ebc22606ea875e024664b2b5ba82033f919f8d1a96f22dfebda33a7cb5296c83
 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": 76226.1169229518,
-      "required_pages_per_second_upper_estimate": 0.005273499637688854
+      "required_token_positions_per_second": 77349.77256644783,
+      "required_pages_per_second_upper_estimate": 0.00535123674234621
     },
     {
       "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": 9528.264615368975,
-      "required_pages_per_second_upper_estimate": 0.003989285421663871
+      "required_token_positions_per_second": 9668.72157080598,
+      "required_pages_per_second_upper_estimate": 0.004048091815830584
     },
     {
       "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": 19056.52923073795,
-      "required_pages_per_second_upper_estimate": 0.003968972484291576
+      "required_token_positions_per_second": 19337.44314161196,
+      "required_pages_per_second_upper_estimate": 0.004027479443728623
     },
     {
       "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": 28584.793846106924,
-      "required_pages_per_second_upper_estimate": 0.0017136147446211187
+      "required_token_positions_per_second": 29006.16471241794,
+      "required_pages_per_second_upper_estimate": 0.001738875284660405
     },
     {
       "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": 28584.793846106924,
-      "required_pages_per_second_upper_estimate": 0.002222341990314369
+      "required_token_positions_per_second": 29006.16471241794,
+      "required_pages_per_second_upper_estimate": 0.0022551017217554846
     },
     {
       "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": 28584.793846106924,
-      "required_pages_per_second_upper_estimate": 0.3666550434974785
+      "required_token_positions_per_second": 29006.16471241794,
+      "required_pages_per_second_upper_estimate": 0.3720599365377296
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.27300006673436883,
-    "producer_p95_seconds": 0.3025267820012232,
-    "gpu_update_seconds": 0.9672275458360701,
-    "producer_batches_per_second": 3.663002767589093,
-    "gpu_batches_per_second": 1.033882879271807,
-    "producer_headroom": 3.5429571772859307,
+    "producer_batch_seconds": 0.10940042345293932,
+    "producer_p95_seconds": 0.3257678099980694,
+    "gpu_update_seconds": 0.9531766875806063,
+    "producer_batches_per_second": 9.140732443601273,
+    "gpu_batches_per_second": 1.049123434332246,
+    "producer_headroom": 8.712733072652442,
     "sustainable": true,
-    "buffer_seconds": 60.93533538767242,
+    "buffer_seconds": 60.05013131757819,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 78,
+    "upload_pages_per_commit": 81,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 339322.7889448642,
-    "upload_bytes_per_second": 5346631.938425806,
+    "generation_bytes_per_second": 425795.8908965641,
+    "upload_bytes_per_second": 5026975.921434505,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 72.81077147731251,
+    "estimated_full_shards_per_hour": 91.36588616535848,
     "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": 0.9240188730000227,
-      "bytes": 2599575,
+      "seconds": 1.1706882670005143,
+      "bytes": 13271387,
+      "pages": 35,
+      "shards": 4
+    },
+    {
+      "seconds": 0.8168831449984282,
+      "bytes": 1231448,
       "pages": 36,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.3233367040011217,
-      "bytes": 5989613,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1650056679973204,
+      "bytes": 7481151,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.4566520219996164,
-      "bytes": 7624893,
-      "pages": 38,
+      "seconds": 0.9575125699993805,
+      "bytes": 1008516,
+      "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.5694887619974907,
+      "bytes": 10308114,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.1967464800000016,
-      "bytes": 10837616,
-      "pages": 33,
+      "seconds": 1.0191816589976952,
+      "bytes": 7135611,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1927219149984012,
-      "bytes": 6126779,
-      "pages": 36,
+      "seconds": 1.1397755830003007,
+      "bytes": 6913587,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.3350182510002924,
-      "bytes": 7769901,
-      "pages": 39,
+      "seconds": 1.4925930749996041,
+      "bytes": 12021057,
+      "pages": 40,
+      "shards": 4
+    },
+    {
+      "seconds": 1.3497725520028325,
+      "bytes": 6087212,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 0.784925423002278,
+      "bytes": 818137,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.161659340999904,
+      "bytes": 8019004,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 2.9742520910003805,
-      "bytes": 5474226,
+      "seconds": 1.0203383750013018,
+      "bytes": 5011407,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.198957466000138,
-      "bytes": 7513050,
+      "seconds": 1.1040821489987138,
+      "bytes": 11902249,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1162976039995556,
+      "bytes": 7875044,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 3.0025762599980226,
-      "bytes": 963365,
-      "pages": 26,
+      "seconds": 1.1005420340006822,
+      "bytes": 1290047,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1665040549996775,
-      "bytes": 11979568,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.504874573998677,
+      "bytes": 12022604,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.0769242890019086,
-      "bytes": 7547252,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.3143507820022933,
+      "bytes": 918492,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.9740304760016443,
-      "bytes": 5265802,
-      "pages": 40,
+      "seconds": 1.5125747689999116,
+      "bytes": 6481526,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0207744180006557,
-      "bytes": 5943379,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.7741823629985447,
+      "bytes": 13400964,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.2191226570030267,
-      "bytes": 2638070,
-      "pages": 37,
+      "seconds": 1.2863249110014294,
+      "bytes": 4967385,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.1678417209986947,
-      "bytes": 10292515,
-      "pages": 32,
+      "seconds": 1.6688778630013985,
+      "bytes": 7600081,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9083037529999274,
-      "bytes": 719591,
-      "pages": 20,
+      "seconds": 1.2229772380014765,
+      "bytes": 1028666,
+      "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.395833352999034,
-      "bytes": 13910666,
-      "pages": 50,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8840833290014416,
-      "bytes": 6069784,
-      "pages": 36,
+      "seconds": 1.4301322219980648,
+      "bytes": 6037828,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.1806990879995283,
-      "bytes": 12168625,
-      "pages": 35,
+      "seconds": 1.568642534999526,
+      "bytes": 11958930,
+      "pages": 32,
       "shards": 4
     },
     {
-      "seconds": 0.9337092679998023,
-      "bytes": 919576,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2493900669978757,
-      "bytes": 6142029,
-      "pages": 37,
+      "seconds": 1.4158421129977796,
+      "bytes": 6969172,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.9163747099992179,
-      "bytes": 2165474,
-      "pages": 22,
+      "seconds": 1.4889061510002648,
+      "bytes": 5773625,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0913281839966658,
-      "bytes": 12193503,
-      "pages": 46,
+      "seconds": 1.7626394689978042,
+      "bytes": 7617654,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1941587969995453,
-      "bytes": 6022783,
+      "seconds": 1.338195986001665,
+      "bytes": 5044470,
       "pages": 31,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.9699245899973903,
-      "bytes": 5017188,
-      "pages": 6,
-      "shards": 2
+      "seconds": 1.4609920459988643,
+      "bytes": 7624505,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 0.8813080820000323,
-      "bytes": 1258193,
+      "seconds": 1.1959701189989573,
+      "bytes": 1136148,
       "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1000836000021081,
-      "bytes": 6163961,
-      "pages": 36,
+      "seconds": 1.5574687859989353,
+      "bytes": 11997417,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1141700879998098,
-      "bytes": 11677393,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.3358469599988894,
+      "bytes": 7186422,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 0.971025193000969,
-      "bytes": 6720640,
-      "pages": 27,
+      "seconds": 1.527223689998209,
+      "bytes": 6243254,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.2639794309980061,
-      "bytes": 7601943,
-      "pages": 40,
+      "seconds": 1.4163199819995498,
+      "bytes": 7360891,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1646632219999447,
-      "bytes": 931521,
-      "pages": 28,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2082039209999493,
-      "bytes": 10963858,
+      "seconds": 1.1906762940016051,
+      "bytes": 1163380,
       "pages": 35,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 1.5337231229968893,
-      "bytes": 7962330,
-      "pages": 40,
+      "seconds": 1.3281080369997653,
+      "bytes": 9814293,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.0160535649993108,
-      "bytes": 1074791,
+      "seconds": 1.3582085800007917,
+      "bytes": 8243145,
       "pages": 31,
-      "shards": 1
-    },
-    {
-      "seconds": 1.7156042310016346,
-      "bytes": 13288348,
-      "pages": 39,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.1277745370025514,
-      "bytes": 5019078,
-      "pages": 31,
+      "seconds": 1.29941092199806,
+      "bytes": 6125022,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.113636714999302,
-      "bytes": 5851090,
-      "pages": 31,
+      "seconds": 1.4454559629994037,
+      "bytes": 6189473,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4330363720000605,
-      "bytes": 7577948,
-      "pages": 40,
+      "seconds": 1.604715366000164,
+      "bytes": 6773206,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.8521174329980568,
-      "bytes": 865962,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0794645450005191,
-      "bytes": 10910956,
+      "seconds": 1.4528134669999417,
+      "bytes": 6484124,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.1706882670005143,
-      "bytes": 13271387,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.400596306000807,
+      "bytes": 7658131,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.8168831449984282,
-      "bytes": 1231448,
-      "pages": 36,
+      "seconds": 1.01357181700223,
+      "bytes": 644212,
+      "pages": 18,
       "shards": 1
     },
     {
-      "seconds": 1.1650056679973204,
-      "bytes": 7481151,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9575125699993805,
-      "bytes": 1008516,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.5190746830012358,
+      "bytes": 16798507,
+      "pages": 50,
+      "shards": 4
     },
     {
-      "seconds": 1.5694887619974907,
-      "bytes": 10308114,
-      "pages": 44,
-      "shards": 3
+      "seconds": 1.2172141400005785,
+      "bytes": 2428421,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.0191816589976952,
-      "bytes": 7135611,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.3105249759973958,
+      "bytes": 6190634,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1397755830003007,
-      "bytes": 6913587,
-      "pages": 33,
+      "seconds": 1.3242703979994985,
+      "bytes": 5961993,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4925930749996041,
-      "bytes": 12021057,
-      "pages": 40,
+      "seconds": 1.7475975829947856,
+      "bytes": 6736460,
+      "pages": 41,
       "shards": 4
     },
     {
-      "seconds": 1.3497725520028325,
-      "bytes": 6087212,
-      "pages": 37,
+      "seconds": 1.445422744996904,
+      "bytes": 6026824,
+      "pages": 21,
       "shards": 2
     },
     {
-      "seconds": 0.784925423002278,
-      "bytes": 818137,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.5578130739959306,
+      "bytes": 13753578,
+      "pages": 49,
+      "shards": 4
     },
     {
-      "seconds": 1.161659340999904,
-      "bytes": 8019004,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.2824578229992767,
+      "bytes": 5780729,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.0203383750013018,
-      "bytes": 5011407,
-      "pages": 29,
+      "seconds": 1.419093739001255,
+      "bytes": 6340288,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1040821489987138,
-      "bytes": 11902249,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.3758785059981165,
+      "bytes": 2329519,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.1162976039995556,
-      "bytes": 7875044,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.3759045859987964,
+      "bytes": 6265583,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1005420340006822,
-      "bytes": 1290047,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.4603665260001435,
+      "bytes": 2072280,
+      "pages": 20,
+      "shards": 2
     },
     {
-      "seconds": 1.504874573998677,
-      "bytes": 12022604,
-      "pages": 34,
+      "seconds": 1.5149503700013156,
+      "bytes": 16156661,
+      "pages": 43,
       "shards": 4
     },
     {
-      "seconds": 1.3143507820022933,
-      "bytes": 918492,
-      "pages": 27,
+      "seconds": 1.2590785899956245,
+      "bytes": 6075199,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.5125747689999116,
-      "bytes": 6481526,
-      "pages": 37,
+      "seconds": 1.3119273870033794,
+      "bytes": 2422074,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.7741823629985447,
-      "bytes": 13400964,
-      "pages": 38,
+      "seconds": 1.4814246340029058,
+      "bytes": 5980643,
+      "pages": 34,
+      "shards": 2
+    },
+    {
+      "seconds": 1.4380262719932944,
+      "bytes": 12216092,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.2863249110014294,
-      "bytes": 4967385,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.1287659260051441,
+      "bytes": 715029,
+      "pages": 21,
+      "shards": 1
     },
     {
-      "seconds": 1.6688778630013985,
-      "bytes": 7600081,
-      "pages": 39,
+      "seconds": 1.51385825300531,
+      "bytes": 12055505,
+      "pages": 47,
       "shards": 3
     },
     {
-      "seconds": 1.2229772380014765,
-      "bytes": 1028666,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.379372390001663,
+      "bytes": 7458410,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.4301322219980648,
-      "bytes": 6037828,
+      "seconds": 1.3072389689987176,
+      "bytes": 4796842,
       "pages": 34,
       "shards": 2
-    },
-    {
-      "seconds": 1.568642534999526,
-      "bytes": 11958930,
-      "pages": 32,
-      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a95f31929a892cce3d118cf8d73e75858c30e57cad5a3ec036a4b699b0452b6
-size 721772159
+oid sha256:2e88f84323ca4bdc4a8a0fa1302e353f3a442dcb7967001b5b87a164915124f0
+size 721776191