Qwen90M step 860000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 855000,
+  "step": 860000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "628d1e63b12b188b6ca281f0cf6c430fc964fc9444c4b85a80a780b372b0644b",
+    "model.safetensors": "76d2d488ad46202033ce64137cf12ae9086c284d7e8060eb3f80363676d0e8d0",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "0b9d7502ee7a37b93921e50411dcf10cc48e5c0ea61bf54ea6e2aa3c774d65a7",
+    "trainer.pt": "892eb6396b64cb4fd594382fdc135dcfdab02c54a38e9994701368b84bd73dcd",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367",
-    "shard_scheduler.json": "24de37ac4e4183ecbaeccca24a00bbf84010cf8c976979813ca172c7341fe35d",
-    "coverage.json": "399d603729230593eb96d15af1ccf15d2fed8d8f61569fc076f477232117f990",
+    "shard_scheduler.json": "9a29ea4bdc9cefd2e92ebd24e57d2572330e585a6e74a99d3c512c6e2de2f678",
+    "coverage.json": "7fba50c91c135d7273f46b84ac1af03e10b40c0850608fe8d645d5160ea5fc41",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 855000,
+  "step": 860000,
   "stats": {
-    "climbmix/blocks": 30780000,
-    "climbmix/targets": 51453874788,
-    "climbmix/input_tokens": 51457135364,
-    "ultra_qa/blocks": 11542500,
-    "ultra_qa/targets": 18830986539,
-    "ultra_qa/input_tokens": 18831022740,
-    "cortex/blocks": 11542500,
-    "cortex/targets": 438911204,
-    "cortex/input_tokens": 18347586211,
-    "ultra_style/blocks": 11542500,
-    "ultra_style/targets": 18256746475,
-    "ultra_style/input_tokens": 18257703756,
-    "openmath/blocks": 7695000,
-    "openmath/targets": 11890027768,
-    "openmath/input_tokens": 11893989809,
-    "rewrite6/blocks": 3847500,
-    "rewrite6/targets": 6307501946,
-    "rewrite6/input_tokens": 6314417109
+    "climbmix/blocks": 30960000,
+    "climbmix/targets": 51754676559,
+    "climbmix/input_tokens": 51757956002,
+    "ultra_qa/blocks": 11610000,
+    "ultra_qa/targets": 18941056864,
+    "ultra_qa/input_tokens": 18941093290,
+    "cortex/blocks": 11610000,
+    "cortex/targets": 441483304,
+    "cortex/input_tokens": 18454886279,
+    "ultra_style/blocks": 11610000,
+    "ultra_style/targets": 18363767524,
+    "ultra_style/input_tokens": 18364730231,
+    "openmath/blocks": 7740000,
+    "openmath/targets": 11959461269,
+    "openmath/input_tokens": 11963446369,
+    "rewrite6/blocks": 3870000,
+    "rewrite6/targets": 6344361516,
+    "rewrite6/input_tokens": 6351317040
   },
   "sources": {
     "climbmix": {
-      "page": 4215,
-      "row": 2185,
+      "page": 4240,
+      "row": 396,
       "start": {
-        "file": 204,
+        "file": 205,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
+          "row_group": 20,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 17266825
+      "offset": 0,
+      "documents": 17367436
     },
     "rewrite6": {
-      "page": 48,
-      "row": 1106,
+      "page": 12,
+      "row": 642,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 196608
+          "row": 49152
         }
       },
-      "epoch": 47,
-      "offset": 0,
-      "documents": 9851702
+      "epoch": 48,
+      "offset": 2048,
+      "documents": 9909186
     },
     "openmath": {
-      "page": 2634,
-      "row": 2725,
+      "page": 2650,
+      "row": 161,
       "start": {
         "file": 24,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 309,
-          "row": 265
+          "row_group": 374,
+          "row": 801
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 10791589
+      "documents": 10854561
     },
     "ultra_qa": {
-      "page": 1139,
-      "row": 374,
+      "page": 1145,
+      "row": 3108,
       "start": {
-        "file": 8,
+        "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 7,
-          "row": 56622
+          "row_group": 0,
+          "row": 13278
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 4665718
+      "offset": 4096,
+      "documents": 4693028
     },
     "ultra_style": {
-      "page": 1738,
-      "row": 2213,
+      "page": 1748,
+      "row": 2510,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 14940
+          "row": 55900
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 7121061
+      "documents": 7162318
     },
     "cortex": {
-      "page": 223850,
-      "row": 23,
+      "page": 225159,
+      "row": 87,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 28652823
+      "documents": 28820439
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 855000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "628d1e63b12b188b6ca281f0cf6c430fc964fc9444c4b85a80a780b372b0644b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "0b9d7502ee7a37b93921e50411dcf10cc48e5c0ea61bf54ea6e2aa3c774d65a7", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367", "shard_scheduler.json": "24de37ac4e4183ecbaeccca24a00bbf84010cf8c976979813ca172c7341fe35d", "coverage.json": "399d603729230593eb96d15af1ccf15d2fed8d8f61569fc076f477232117f990", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000855000"}
+{"step": 860000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "76d2d488ad46202033ce64137cf12ae9086c284d7e8060eb3f80363676d0e8d0", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "892eb6396b64cb4fd594382fdc135dcfdab02c54a38e9994701368b84bd73dcd", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367", "shard_scheduler.json": "9a29ea4bdc9cefd2e92ebd24e57d2572330e585a6e74a99d3c512c6e2de2f678", "coverage.json": "7fba50c91c135d7273f46b84ac1af03e10b40c0850608fe8d645d5160ea5fc41", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000860000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:628d1e63b12b188b6ca281f0cf6c430fc964fc9444c4b85a80a780b372b0644b
+oid sha256:76d2d488ad46202033ce64137cf12ae9086c284d7e8060eb3f80363676d0e8d0
 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": 77256.40416056558,
-      "required_pages_per_second_upper_estimate": 0.005344777299382726
+      "required_token_positions_per_second": 76235.28940132579,
+      "required_pages_per_second_upper_estimate": 0.0052741342110258485
     },
     {
       "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": 9657.050520070698,
-      "required_pages_per_second_upper_estimate": 0.004043205390606975
+      "required_token_positions_per_second": 9529.411175165724,
+      "required_pages_per_second_upper_estimate": 0.003989765462307878
     },
     {
       "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": 19314.101040141395,
-      "required_pages_per_second_upper_estimate": 0.004022617899564916
+      "required_token_positions_per_second": 19058.822350331448,
+      "required_pages_per_second_upper_estimate": 0.003969450080629271
     },
     {
       "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": 28971.151560212093,
-      "required_pages_per_second_upper_estimate": 0.0017367762996476524
+      "required_token_positions_per_second": 28588.23352549717,
+      "required_pages_per_second_upper_estimate": 0.0017138209481484775
     },
     {
       "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": 28971.151560212093,
-      "required_pages_per_second_upper_estimate": 0.002252379603178061
+      "required_token_positions_per_second": 28588.23352549717,
+      "required_pages_per_second_upper_estimate": 0.00222260941025741
     },
     {
       "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": 28971.151560212093,
-      "required_pages_per_second_upper_estimate": 0.37161082541542684
+      "required_token_positions_per_second": 28588.23352549717,
+      "required_pages_per_second_upper_estimate": 0.36669916401145664
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,408 +158,408 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.07687787735935103,
-    "producer_p95_seconds": 0.20161003999965033,
-    "gpu_update_seconds": 0.9543286514703385,
-    "producer_batches_per_second": 13.00764321738086,
-    "gpu_batches_per_second": 1.0478570442785045,
-    "producer_headroom": 12.41356661045037,
+    "producer_batch_seconds": 0.06858718393758068,
+    "producer_p95_seconds": 0.22669146000043838,
+    "gpu_update_seconds": 0.9671111709417584,
+    "producer_batches_per_second": 14.579983352430284,
+    "gpu_batches_per_second": 1.0340072889719751,
+    "producer_headroom": 14.100464772280196,
     "sustainable": true,
-    "buffer_seconds": 60.122705042631324,
+    "buffer_seconds": 60.92800376933078,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 84,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 446772.5461731995,
-    "upload_bytes_per_second": 4802594.46749234,
+    "generation_bytes_per_second": 389341.2469068264,
+    "upload_bytes_per_second": 3447999.215230209,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 95.8669880761813,
+    "estimated_full_shards_per_hour": 83.54356818584054,
     "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": 1.2396700680001231,
-      "bytes": 7470065,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.6093791190005504,
-      "bytes": 874866,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.182235641999796,
+      "bytes": 6603515,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.1941245199996047,
-      "bytes": 7420221,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.285518773000149,
+      "bytes": 6002809,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.6006398999998055,
-      "bytes": 9959535,
-      "pages": 29,
+      "seconds": 1.7619018259993027,
+      "bytes": 7674402,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2196721769996657,
-      "bytes": 7030636,
-      "pages": 39,
+      "seconds": 1.6144984039983683,
+      "bytes": 4728422,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.0609669329996905,
-      "bytes": 7430334,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.6315553869990254,
+      "bytes": 5703905,
+      "pages": 24,
+      "shards": 2
     },
     {
-      "seconds": 1.0616652130001967,
-      "bytes": 6074717,
+      "seconds": 1.11825536699871,
+      "bytes": 2656311,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3232113710000704,
-      "bytes": 5040655,
-      "pages": 31,
+      "seconds": 1.9071076060008636,
+      "bytes": 12136915,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.5191957560000446,
-      "bytes": 7567101,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.7720820229988021,
+      "bytes": 11800483,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.065770677000728,
-      "bytes": 897206,
-      "pages": 26,
+      "seconds": 1.0689491579996684,
+      "bytes": 1154476,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.459628982999675,
-      "bytes": 13449017,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.7027444539999124,
+      "bytes": 6185388,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.560673630000565,
-      "bytes": 10527927,
-      "pages": 42,
+      "seconds": 1.5859740610012523,
+      "bytes": 2563401,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2778493570003775,
+      "bytes": 10237665,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.384273941000174,
-      "bytes": 1166498,
-      "pages": 35,
+      "seconds": 2.496101438000551,
+      "bytes": 13297971,
+      "pages": 33,
+      "shards": 4
+    },
+    {
+      "seconds": 1.2942262079995999,
+      "bytes": 1101377,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.245850141000119,
-      "bytes": 7339507,
-      "pages": 30,
+      "seconds": 1.815276637998977,
+      "bytes": 10307802,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0799733499998183,
-      "bytes": 6017108,
-      "pages": 34,
+      "seconds": 1.1417645319997973,
+      "bytes": 2370286,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2428365799996755,
-      "bytes": 12889604,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 1.21177975899991,
-      "bytes": 6176322,
-      "pages": 36,
+      "seconds": 1.1171746460004215,
+      "bytes": 6111968,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4487905369996952,
-      "bytes": 925527,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.2683859379994828,
+      "bytes": 7618531,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.6813211230000888,
-      "bytes": 7642707,
-      "pages": 39,
+      "seconds": 1.7501807870012271,
+      "bytes": 11262011,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.8502533870005209,
-      "bytes": 10875382,
+      "seconds": 1.2981046400000196,
+      "bytes": 6199716,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.247339760999239,
-      "bytes": 856828,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.6966266469989932,
+      "bytes": 7696834,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.4432682090000526,
-      "bytes": 13357057,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.3060271180002019,
+      "bytes": 1241981,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.271395830000074,
-      "bytes": 6214888,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.5251753980010108,
+      "bytes": 9924170,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.4851975340006902,
-      "bytes": 6638444,
-      "pages": 30,
+      "seconds": 1.1737768090006284,
+      "bytes": 7614180,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.5820424499997898,
-      "bytes": 5864908,
-      "pages": 27,
+      "seconds": 1.2979090160006308,
+      "bytes": 6818365,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.0337117700000817,
-      "bytes": 1192866,
-      "pages": 35,
+      "seconds": 0.9478092970002763,
+      "bytes": 253929,
+      "pages": 8,
       "shards": 1
     },
     {
-      "seconds": 1.3218913410000823,
-      "bytes": 7396091,
+      "seconds": 1.6955216110000038,
+      "bytes": 7314599,
       "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1367835250002827,
-      "bytes": 5040157,
-      "pages": 37,
+      "seconds": 1.2749757900000986,
+      "bytes": 5237846,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.2845682920014951,
-      "bytes": 11638802,
-      "pages": 28,
-      "shards": 3
+      "seconds": 1.5798416210000141,
+      "bytes": 6237294,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.3405057710006076,
-      "bytes": 7718368,
+      "seconds": 1.5424033700001019,
+      "bytes": 7705972,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.021054885999547,
-      "bytes": 1163402,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.9795733470000414,
+      "bytes": 923741,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.532974641999317,
-      "bytes": 11431908,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.5211821389984834,
+      "bytes": 18268501,
+      "pages": 39,
+      "shards": 5
     },
     {
-      "seconds": 1.541119751000224,
-      "bytes": 6183503,
+      "seconds": 1.3536671069996373,
+      "bytes": 6151717,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.4027249899991148,
-      "bytes": 1181995,
-      "pages": 34,
+      "seconds": 1.1601749330002349,
+      "bytes": 883373,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.3459127579990309,
-      "bytes": 13166079,
-      "pages": 31,
-      "shards": 4
+      "seconds": 1.1801181450009608,
+      "bytes": 7664644,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.166742791001525,
-      "bytes": 5284765,
-      "pages": 34,
+      "seconds": 1.0472080479994474,
+      "bytes": 5473896,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.747680902999491,
-      "bytes": 7697898,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.548700262001148,
+      "bytes": 13538856,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 0.9798488249998627,
-      "bytes": 6153509,
-      "pages": 36,
+      "seconds": 1.3191421729989088,
+      "bytes": 5997715,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.0631558050008607,
-      "bytes": 1018193,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.765130539000893,
+      "bytes": 937778,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 1.8303398019997985,
-      "bytes": 12284688,
-      "pages": 40,
+      "seconds": 2.7900099470007262,
+      "bytes": 11587181,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.182235641999796,
-      "bytes": 6603515,
-      "pages": 27,
-      "shards": 2
+      "seconds": 0.987754063000466,
+      "bytes": 1220161,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.285518773000149,
-      "bytes": 6002809,
+      "seconds": 1.716705255999841,
+      "bytes": 6235687,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.7619018259993027,
-      "bytes": 7674402,
-      "pages": 38,
+      "seconds": 33.88602663399979,
+      "bytes": 7355362,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.6144984039983683,
-      "bytes": 4728422,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.5573600560001069,
+      "bytes": 11648033,
+      "pages": 46,
+      "shards": 3
     },
     {
-      "seconds": 1.6315553869990254,
-      "bytes": 5703905,
-      "pages": 24,
-      "shards": 2
+      "seconds": 1.2928704839996499,
+      "bytes": 7315946,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.11825536699871,
-      "bytes": 2656311,
-      "pages": 35,
+      "seconds": 1.4081739580014982,
+      "bytes": 5891309,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.9071076060008636,
-      "bytes": 12136915,
+      "seconds": 1.286470338000072,
+      "bytes": 2778002,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.7720820229988021,
-      "bytes": 11800483,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1679330120005034,
+      "bytes": 10820666,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.0689491579996684,
-      "bytes": 1154476,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.4827527400011604,
+      "bytes": 6647922,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.7027444539999124,
-      "bytes": 6185388,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.248109399999521,
+      "bytes": 7422482,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.5859740610012523,
-      "bytes": 2563401,
-      "pages": 35,
+      "seconds": 1.451942487999986,
+      "bytes": 6077412,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.2778493570003775,
-      "bytes": 10237665,
-      "pages": 33,
+      "seconds": 1.4631003799986502,
+      "bytes": 7075109,
+      "pages": 43,
       "shards": 3
     },
     {
-      "seconds": 2.496101438000551,
-      "bytes": 13297971,
-      "pages": 33,
-      "shards": 4
+      "seconds": 2.129564337999909,
+      "bytes": 5884538,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.2942262079995999,
-      "bytes": 1101377,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1728028199995606,
+      "bytes": 6147166,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.815276637998977,
-      "bytes": 10307802,
-      "pages": 39,
+      "seconds": 1.2317090710002958,
+      "bytes": 8202394,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1417645319997973,
-      "bytes": 2370286,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.6838970150001842,
+      "bytes": 9992623,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.1171746460004215,
-      "bytes": 6111968,
+      "seconds": 1.2029927740004496,
+      "bytes": 6068424,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2683859379994828,
-      "bytes": 7618531,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1870869580016006,
+      "bytes": 2666964,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.7501807870012271,
-      "bytes": 11262011,
-      "pages": 28,
+      "seconds": 1.4462591800001974,
+      "bytes": 10549433,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2981046400000196,
-      "bytes": 6199716,
+      "seconds": 1.0944702749984572,
+      "bytes": 2560030,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.6966266469989932,
-      "bytes": 7696834,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.3531785280028998,
+      "bytes": 11787584,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.3060271180002019,
-      "bytes": 1241981,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.0967427190007584,
+      "bytes": 6096101,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.5251753980010108,
-      "bytes": 9924170,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.3933399300003657,
+      "bytes": 2445888,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.1737768090006284,
-      "bytes": 7614180,
+      "seconds": 1.9226194889997714,
+      "bytes": 9744014,
       "pages": 39,
       "shards": 3
     }

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0b9d7502ee7a37b93921e50411dcf10cc48e5c0ea61bf54ea6e2aa3c774d65a7
-size 721794431
+oid sha256:892eb6396b64cb4fd594382fdc135dcfdab02c54a38e9994701368b84bd73dcd
+size 721770367