Qwen90M step 585000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 580000,
+  "step": 585000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "86da1ad7765ecdf63d4c21ed8d3083e166f543b8c584d4e1ad8585460b37a1c2",
+    "model.safetensors": "2b447c14881264552ab420a8708a59aa948aecc2bafff1047401930b2ef52981",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "1330a58909395b69dd9e933634473173544669f01ca72451c097b4fe93ebb964",
+    "trainer.pt": "92d7ebabc11fd9cdaa8f8eeb61f310d2c4a21ca61ea41e252ac9c8c24070f6e8",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667",
-    "shard_scheduler.json": "4c7cecc17496ee854ce7e9db16f9fc5fe111120cc8b45911284ec118a736e4bb",
-    "coverage.json": "e8919debdd5781a63f54d50ff90e0325881c51f855c15fdd97780f35b3c257bc",
+    "shard_scheduler.json": "0c12b8bca40ac662ed2a947c032a8d65d4cbe1bb3c37e3f37268d66836f48587",
+    "coverage.json": "cf63695a4947eed2d72d78bf215c300aafb9cd742c85d0c9a809906a58a8ff57",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 580000,
+  "step": 585000,
   "stats": {
-    "climbmix/blocks": 20880000,
-    "climbmix/targets": 34903766814,
-    "climbmix/input_tokens": 34905978789,
-    "ultra_qa/blocks": 7830000,
-    "ultra_qa/targets": 12773998197,
-    "ultra_qa/input_tokens": 12774022841,
-    "cortex/blocks": 7830000,
-    "cortex/targets": 297744461,
-    "cortex/input_tokens": 12446411636,
-    "ultra_style/blocks": 7830000,
-    "ultra_style/targets": 12384628359,
-    "ultra_style/input_tokens": 12385277999,
-    "openmath/blocks": 5220000,
-    "openmath/targets": 8065835620,
-    "openmath/input_tokens": 8068521894,
-    "rewrite6/blocks": 2610000,
-    "rewrite6/targets": 4278710471,
-    "rewrite6/input_tokens": 4283401756
+    "climbmix/blocks": 21060000,
+    "climbmix/targets": 35204728526,
+    "climbmix/input_tokens": 35206959792,
+    "ultra_qa/blocks": 7897500,
+    "ultra_qa/targets": 12884016978,
+    "ultra_qa/input_tokens": 12884041832,
+    "cortex/blocks": 7897500,
+    "cortex/targets": 300307032,
+    "cortex/input_tokens": 12553655276,
+    "ultra_style/blocks": 7897500,
+    "ultra_style/targets": 12491928833,
+    "ultra_style/input_tokens": 12492584132,
+    "openmath/blocks": 5265000,
+    "openmath/targets": 8135369806,
+    "openmath/input_tokens": 8138079413,
+    "rewrite6/blocks": 2632500,
+    "rewrite6/targets": 4315618350,
+    "rewrite6/input_tokens": 4320349938
   },
   "sources": {
     "climbmix": {
-      "page": 2859,
-      "row": 2966,
+      "page": 2884,
+      "row": 2144,
       "start": {
-        "file": 138,
+        "file": 139,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 29,
+          "row_group": 47,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 10240,
-      "documents": 11713430
+      "offset": 2048,
+      "documents": 11815008
     },
     "rewrite6": {
-      "page": 26,
-      "row": 3974,
+      "page": 41,
+      "row": 2,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 106496
+          "row": 167936
         }
       },
       "epoch": 32,
       "offset": 0,
-      "documents": 6683398
+      "documents": 6740866
     },
     "openmath": {
-      "page": 1786,
-      "row": 3533,
+      "page": 1802,
+      "row": 1223,
       "start": {
         "file": 16,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 329,
-          "row": 56
+          "row_group": 394,
+          "row": 592
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 7318989
+      "documents": 7382215
     },
     "ultra_qa": {
-      "page": 772,
-      "row": 2970,
+      "page": 779,
+      "row": 1531,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 44355
+          "row_group": 1,
+          "row": 8520
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3165082
+      "documents": 3192315
     },
     "ultra_style": {
-      "page": 1179,
-      "row": 1648,
+      "page": 1189,
+      "row": 2100,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 0,
-          "row": 34590
+          "row": 75550
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4830832
+      "documents": 4872244
     },
     "cortex": {
-      "page": 151857,
-      "row": 6,
+      "page": 153166,
+      "row": 102,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 19437702
+      "documents": 19605350
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 580000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "86da1ad7765ecdf63d4c21ed8d3083e166f543b8c584d4e1ad8585460b37a1c2", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "1330a58909395b69dd9e933634473173544669f01ca72451c097b4fe93ebb964", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667", "shard_scheduler.json": "4c7cecc17496ee854ce7e9db16f9fc5fe111120cc8b45911284ec118a736e4bb", "coverage.json": "e8919debdd5781a63f54d50ff90e0325881c51f855c15fdd97780f35b3c257bc", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000580000"}
+{"step": 585000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "2b447c14881264552ab420a8708a59aa948aecc2bafff1047401930b2ef52981", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "92d7ebabc11fd9cdaa8f8eeb61f310d2c4a21ca61ea41e252ac9c8c24070f6e8", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667", "shard_scheduler.json": "0c12b8bca40ac662ed2a947c032a8d65d4cbe1bb3c37e3f37268d66836f48587", "coverage.json": "cf63695a4947eed2d72d78bf215c300aafb9cd742c85d0c9a809906a58a8ff57", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000585000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86da1ad7765ecdf63d4c21ed8d3083e166f543b8c584d4e1ad8585460b37a1c2
+oid sha256:2b447c14881264552ab420a8708a59aa948aecc2bafff1047401930b2ef52981
 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": 76844.82269293365,
-      "required_pages_per_second_upper_estimate": 0.0053163031384513715
+      "required_token_positions_per_second": 76453.60105814453,
+      "required_pages_per_second_upper_estimate": 0.005289237517997402
     },
     {
       "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": 9605.602836616707,
-      "required_pages_per_second_upper_estimate": 0.004021665319894588
+      "required_token_positions_per_second": 9556.700132268066,
+      "required_pages_per_second_upper_estimate": 0.0040011907787884035
     },
     {
       "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": 19211.205673233413,
-      "required_pages_per_second_upper_estimate": 0.004001187508171285
+      "required_token_positions_per_second": 19113.400264536132,
+      "required_pages_per_second_upper_estimate": 0.003980817220841719
     },
     {
       "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": 28816.808509850118,
-      "required_pages_per_second_upper_estimate": 0.0017275236694466443
+      "required_token_positions_per_second": 28670.1003968042,
+      "required_pages_per_second_upper_estimate": 0.0017187287420798595
     },
     {
       "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": 28816.808509850118,
-      "required_pages_per_second_upper_estimate": 0.0022403801098957493
+      "required_token_positions_per_second": 28670.1003968042,
+      "required_pages_per_second_upper_estimate": 0.0022289741994072207
     },
     {
       "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": 28816.808509850118,
-      "required_pages_per_second_upper_estimate": 0.3696310784860394
+      "required_token_positions_per_second": 28670.1003968042,
+      "required_pages_per_second_upper_estimate": 0.36774926433478533
     }
   ],
   "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": 704566,
-    "producer_batch_seconds": 0.0661616753671126,
-    "producer_p95_seconds": 0.24438537600144628,
-    "gpu_update_seconds": 0.9594400431452845,
-    "producer_batches_per_second": 15.11449029141539,
-    "gpu_batches_per_second": 1.0422746133481668,
-    "producer_headroom": 14.501447217314565,
+    "producer_batch_seconds": 0.12654482389845612,
+    "producer_p95_seconds": 0.2847857099986868,
+    "gpu_update_seconds": 0.9643496052452565,
+    "producer_batches_per_second": 7.902338232359738,
+    "gpu_batches_per_second": 1.0369683303242259,
+    "producer_headroom": 7.620616754890611,
     "sustainable": true,
-    "buffer_seconds": 60.444722718152924,
+    "buffer_seconds": 60.75402513045116,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 406720.83803339687,
-    "upload_bytes_per_second": 5366265.038987921,
+    "generation_bytes_per_second": 354955.35036680783,
+    "upload_bytes_per_second": 5937659.932236937,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 87.2728238654273,
+    "estimated_full_shards_per_hour": 76.16515525105645,
     "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": 1.2003755249988899,
-      "bytes": 10273667,
-      "pages": 44,
+      "seconds": 1.1792509940023592,
+      "bytes": 9839539,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0975800400010485,
-      "bytes": 2683103,
-      "pages": 36,
+      "seconds": 1.0725541200008593,
+      "bytes": 2601868,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1853528239989828,
-      "bytes": 5804023,
+      "seconds": 1.330655336001655,
+      "bytes": 11531295,
       "pages": 28,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.138313076000486,
-      "bytes": 10599364,
-      "pages": 39,
+      "seconds": 1.5732356919979793,
+      "bytes": 10605987,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 0.9668768979991,
-      "bytes": 2353349,
-      "pages": 29,
+      "seconds": 0.9885955910031043,
+      "bytes": 2380823,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4940221720007685,
-      "bytes": 12237361,
-      "pages": 41,
-      "shards": 3
+      "seconds": 0.9730731119998381,
+      "bytes": 6090581,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.8858112369998707,
-      "bytes": 2416252,
-      "pages": 28,
+      "seconds": 1.1632558510027593,
+      "bytes": 7851564,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2992984630000137,
-      "bytes": 10017513,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.0146159530013392,
+      "bytes": 5636542,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 4.1799396490005165,
-      "bytes": 5932760,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.0724632399978873,
+      "bytes": 12061908,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.8289485320001404,
-      "bytes": 2513176,
-      "pages": 33,
+      "seconds": 0.9588815639981476,
+      "bytes": 2343317,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9106573579992983,
-      "bytes": 5965787,
+      "seconds": 1.1246145429977332,
+      "bytes": 6175304,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.575513838999541,
-      "bytes": 18103820,
-      "pages": 39,
-      "shards": 5
+      "seconds": 1.2072375589996227,
+      "bytes": 11138063,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 0.8145781550010724,
-      "bytes": 930729,
-      "pages": 26,
+      "seconds": 0.9064097570008016,
+      "bytes": 1171302,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.3511633730013273,
-      "bytes": 6255780,
+      "seconds": 0.9584431869989203,
+      "bytes": 6298065,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.031140414001129,
-      "bytes": 2762787,
-      "pages": 38,
-      "shards": 2
+      "seconds": 0.972795835998113,
+      "bytes": 8254778,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0752210870014096,
-      "bytes": 5610482,
-      "pages": 19,
-      "shards": 2
+      "seconds": 1.3151332330016885,
+      "bytes": 10337557,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.425927022999531,
-      "bytes": 10079783,
-      "pages": 46,
+      "seconds": 1.029109121998772,
+      "bytes": 7643162,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1401308700005757,
-      "bytes": 8309975,
-      "pages": 34,
-      "shards": 3
+      "seconds": 0.8460896450014843,
+      "bytes": 884554,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.2373918849989423,
-      "bytes": 6097119,
-      "pages": 33,
+      "seconds": 1.0631770070031052,
+      "bytes": 6064387,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.1592294109977956,
-      "bytes": 6491968,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.2800072050013114,
+      "bytes": 12000660,
+      "pages": 43,
+      "shards": 4
     },
     {
-      "seconds": 1.6240387940015353,
-      "bytes": 5730323,
-      "pages": 27,
+      "seconds": 1.2026952199994412,
+      "bytes": 6686271,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.3926623539991851,
-      "bytes": 6408754,
+      "seconds": 1.3011887590000697,
+      "bytes": 6177443,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.260699045000365,
-      "bytes": 2670941,
-      "pages": 37,
+      "seconds": 1.0062806160021864,
+      "bytes": 2394266,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.4827766799971869,
-      "bytes": 15946512,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.1657910530011577,
+      "bytes": 10819661,
+      "pages": 44,
+      "shards": 3
     },
     {
-      "seconds": 1.33145204000175,
-      "bytes": 7515329,
-      "pages": 38,
+      "seconds": 1.1555115019982622,
+      "bytes": 7463395,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.9712299829989206,
-      "bytes": 867774,
+      "seconds": 0.8593025919981301,
+      "bytes": 778515,
       "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.1689961140000378,
-      "bytes": 10341916,
-      "pages": 38,
+      "seconds": 1.26428677900185,
+      "bytes": 11594708,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 4.051943458998721,
-      "bytes": 2586599,
+      "seconds": 1.083468554999854,
+      "bytes": 6825645,
       "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9690040009991208,
-      "bytes": 5902816,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9907326490028936,
-      "bytes": 186236,
-      "pages": 7,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1790409999994154,
-      "bytes": 12018438,
-      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.0833680409996305,
-      "bytes": 2366981,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2835563219996402,
-      "bytes": 10401632,
-      "pages": 38,
+      "seconds": 1.0758820400005789,
+      "bytes": 6474447,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9978380930006097,
-      "bytes": 912994,
-      "pages": 28,
+      "seconds": 0.8884251060007955,
+      "bytes": 1122963,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1661826020026638,
-      "bytes": 7439707,
-      "pages": 37,
+      "seconds": 1.0163737470029446,
+      "bytes": 7245209,
+      "pages": 27,
       "shards": 3
     },
     {
-      "seconds": 1.0849376400001347,
-      "bytes": 10733196,
-      "pages": 39,
+      "seconds": 1.4228277920010441,
+      "bytes": 10524529,
+      "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 1.2397475959987787,
-      "bytes": 8159623,
+      "seconds": 1.1392915559990797,
+      "bytes": 8097285,
       "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.2689961300020514,
-      "bytes": 6434173,
+      "seconds": 1.2682354990029125,
+      "bytes": 6131054,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3497664780006744,
-      "bytes": 7459096,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.2669715150004777,
+      "bytes": 6065248,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.0805805290001445,
-      "bytes": 1076352,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.0634998090026784,
+      "bytes": 7147043,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1792509940023592,
-      "bytes": 9839539,
+      "seconds": 3.4036351569993712,
+      "bytes": 6159372,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.0725541200008593,
-      "bytes": 2601868,
-      "pages": 38,
-      "shards": 2
+      "seconds": 0.9252640429986059,
+      "bytes": 1101958,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 1.330655336001655,
-      "bytes": 11531295,
-      "pages": 28,
-      "shards": 3
+      "seconds": 1.268539120999776,
+      "bytes": 13168093,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.5732356919979793,
-      "bytes": 10605987,
-      "pages": 41,
+      "seconds": 1.1610033780016238,
+      "bytes": 10490094,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9885955910031043,
-      "bytes": 2380823,
+      "seconds": 1.0719913379980426,
+      "bytes": 2751265,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.124440429997776,
+      "bytes": 5889635,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9730731119998381,
-      "bytes": 6090581,
-      "pages": 37,
+      "seconds": 1.0922325869978522,
+      "bytes": 6027814,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1632558510027593,
-      "bytes": 7851564,
-      "pages": 38,
+      "seconds": 1.0737022979992616,
+      "bytes": 6591396,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.0146159530013392,
-      "bytes": 5636542,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.1665754960013146,
+      "bytes": 11803743,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.0724632399978873,
-      "bytes": 12061908,
+      "seconds": 0.962029407000955,
+      "bytes": 2665696,
       "pages": 38,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1741747699998086,
+      "bytes": 5919198,
+      "pages": 26,
+      "shards": 2
+    },
+    {
+      "seconds": 1.178343194998888,
+      "bytes": 10619543,
+      "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 0.9588815639981476,
-      "bytes": 2343317,
-      "pages": 28,
+      "seconds": 1.036487166002189,
+      "bytes": 2690894,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1246145429977332,
-      "bytes": 6175304,
-      "pages": 36,
+      "seconds": 1.1432905639994715,
+      "bytes": 5745742,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2072375589996227,
-      "bytes": 11138063,
-      "pages": 36,
+      "seconds": 1.232796011998289,
+      "bytes": 13189628,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 0.9064097570008016,
-      "bytes": 1171302,
+      "seconds": 0.943955818998802,
+      "bytes": 5281939,
       "pages": 35,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 0.9584431869989203,
-      "bytes": 6298065,
-      "pages": 36,
+      "seconds": 1.0029552209998656,
+      "bytes": 6176365,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.972795835998113,
-      "bytes": 8254778,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.1109868420026032,
+      "bytes": 2592499,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.3151332330016885,
-      "bytes": 10337557,
-      "pages": 35,
+      "seconds": 1.24596934100191,
+      "bytes": 5407802,
+      "pages": 21,
+      "shards": 2
+    },
+    {
+      "seconds": 1.361496919002093,
+      "bytes": 10822221,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.029109121998772,
-      "bytes": 7643162,
-      "pages": 40,
+      "seconds": 1.3625388020009268,
+      "bytes": 8506674,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 0.8460896450014843,
-      "bytes": 884554,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.141177229997993,
+      "bytes": 6001543,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.0631770070031052,
-      "bytes": 6064387,
-      "pages": 32,
+      "seconds": 0.9294679570011795,
+      "bytes": 2588442,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2800072050013114,
-      "bytes": 12000660,
-      "pages": 43,
+      "seconds": 1.1605264330028149,
+      "bytes": 9613122,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.2026952199994412,
-      "bytes": 6686271,
-      "pages": 28,
+      "seconds": 1.0401736989988422,
+      "bytes": 6081990,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3011887590000697,
-      "bytes": 6177443,
-      "pages": 36,
+      "seconds": 0.9755379390007874,
+      "bytes": 2340646,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0062806160021864,
-      "bytes": 2394266,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.4557110550013022,
+      "bytes": 16684165,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.1657910530011577,
-      "bytes": 10819661,
-      "pages": 44,
+      "seconds": 1.2760817240014148,
+      "bytes": 7386323,
+      "pages": 39,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1330a58909395b69dd9e933634473173544669f01ca72451c097b4fe93ebb964
-size 721773055
+oid sha256:92d7ebabc11fd9cdaa8f8eeb61f310d2c4a21ca61ea41e252ac9c8c24070f6e8
+size 721766911