Qwen90M step 760000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 755000,
+  "step": 760000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "75cc8e7df2946f47615bba735e2412ba7d26acf35ff8d177274e2b278117ccf7",
+    "model.safetensors": "a860e83f899b4e90ca04decae15751893c788a0a68e4bf4ef98f0325ea54d165",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "04389db27054d0deb18e21ea9848cc49ae3250a62872254d899b5d9fb8463946",
+    "trainer.pt": "1b1f259fdc0e4e7c034af4b926b3e23d2a45e4a91f9e75f8c00269d04f243d83",
     "training_config.json": "75248af0a321738b405c3d51d1d55c7040dbc0785759d95f8dbdf6cd071a6687",
     "performance.json": "48e3aa7938b07d5e08810d60bc208cc2cb9133c4c39a24151758b79306675379",
-    "shard_scheduler.json": "6b4bb13b0a72c739a75dbb27075aa6cb85d8a6601c3ec79525308daa070d0808",
-    "coverage.json": "550b419d22a23b028c27e10458eb1b3ca5486992854436adfd61260ee070e83f",
+    "shard_scheduler.json": "8546a039abf7332253379c1f06f6d9b77cec29d6925ca8a503635e01f245c443",
+    "coverage.json": "e5698dd1fdd8d06e82449a48c972062af4c83576c8da391fd360e5358f8f05e2",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "13510bfc04d8fed598fd0d09c1f778fbaebf19cefdd81f23f5ef3b909ef94a63"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 755000,
+  "step": 760000,
   "stats": {
-    "climbmix/blocks": 27180000,
-    "climbmix/targets": 45436583708,
-    "climbmix/input_tokens": 45439462550,
-    "ultra_qa/blocks": 10192500,
-    "ultra_qa/targets": 16627829240,
-    "ultra_qa/input_tokens": 16627861293,
-    "cortex/blocks": 10192500,
-    "cortex/targets": 387543241,
-    "cortex/input_tokens": 16201531922,
-    "ultra_style/blocks": 10192500,
-    "ultra_style/targets": 16121676984,
-    "ultra_style/input_tokens": 16122522199,
-    "openmath/blocks": 6795000,
-    "openmath/targets": 10499419459,
-    "openmath/input_tokens": 10502917378,
-    "rewrite6/blocks": 3397500,
-    "rewrite6/targets": 5569717202,
-    "rewrite6/input_tokens": 5575823779
+    "climbmix/blocks": 27360000,
+    "climbmix/targets": 45737235053,
+    "climbmix/input_tokens": 45740133081,
+    "ultra_qa/blocks": 10260000,
+    "ultra_qa/targets": 16737918320,
+    "ultra_qa/input_tokens": 16737950563,
+    "cortex/blocks": 10260000,
+    "cortex/targets": 390103024,
+    "cortex/input_tokens": 16308780977,
+    "ultra_style/blocks": 10260000,
+    "ultra_style/targets": 16228024880,
+    "ultra_style/input_tokens": 16228875947,
+    "openmath/blocks": 6840000,
+    "openmath/targets": 10568964438,
+    "openmath/input_tokens": 10572485614,
+    "rewrite6/blocks": 3420000,
+    "rewrite6/targets": 5606566408,
+    "rewrite6/input_tokens": 5612713435
   },
   "sources": {
     "climbmix": {
-      "page": 3722,
-      "row": 698,
+      "page": 3746,
+      "row": 3857,
       "start": {
-        "file": 180,
+        "file": 181,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 14,
+          "row_group": 28,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 15246010
+      "documents": 15347473
     },
     "rewrite6": {
-      "page": 17,
-      "row": 3070,
+      "page": 31,
+      "row": 3430,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 69632
+          "row": 126976
         }
       },
       "epoch": 42,
       "offset": 0,
-      "documents": 8699670
+      "documents": 8757374
     },
     "openmath": {
-      "page": 2326,
-      "row": 1294,
+      "page": 2341,
+      "row": 3059,
       "start": {
         "file": 21,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 357,
-          "row": 646
+          "row_group": 419,
+          "row": 86
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 9528590
+      "documents": 9591795
     },
     "ultra_qa": {
-      "page": 1005,
-      "row": 3733,
+      "page": 1012,
+      "row": 2311,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 26944
+          "row": 55616
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 4120213
+      "offset": 2048,
+      "documents": 4147463
     },
     "ultra_style": {
-      "page": 1535,
-      "row": 679,
+      "page": 1545,
+      "row": 1981,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 37522
+          "row": 78482
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6288039
+      "documents": 6330301
     },
     "cortex": {
-      "page": 197671,
-      "row": 117,
+      "page": 198979,
+      "row": 69,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 25302005
+      "documents": 25469381
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 755000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "75cc8e7df2946f47615bba735e2412ba7d26acf35ff8d177274e2b278117ccf7", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "04389db27054d0deb18e21ea9848cc49ae3250a62872254d899b5d9fb8463946", "training_config.json": "75248af0a321738b405c3d51d1d55c7040dbc0785759d95f8dbdf6cd071a6687", "performance.json": "48e3aa7938b07d5e08810d60bc208cc2cb9133c4c39a24151758b79306675379", "shard_scheduler.json": "6b4bb13b0a72c739a75dbb27075aa6cb85d8a6601c3ec79525308daa070d0808", "coverage.json": "550b419d22a23b028c27e10458eb1b3ca5486992854436adfd61260ee070e83f", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "13510bfc04d8fed598fd0d09c1f778fbaebf19cefdd81f23f5ef3b909ef94a63", "folder": "step-000000755000"}
+{"step": 760000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "a860e83f899b4e90ca04decae15751893c788a0a68e4bf4ef98f0325ea54d165", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "1b1f259fdc0e4e7c034af4b926b3e23d2a45e4a91f9e75f8c00269d04f243d83", "training_config.json": "75248af0a321738b405c3d51d1d55c7040dbc0785759d95f8dbdf6cd071a6687", "performance.json": "48e3aa7938b07d5e08810d60bc208cc2cb9133c4c39a24151758b79306675379", "shard_scheduler.json": "8546a039abf7332253379c1f06f6d9b77cec29d6925ca8a503635e01f245c443", "coverage.json": "e5698dd1fdd8d06e82449a48c972062af4c83576c8da391fd360e5358f8f05e2", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "13510bfc04d8fed598fd0d09c1f778fbaebf19cefdd81f23f5ef3b909ef94a63", "folder": "step-000000760000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75cc8e7df2946f47615bba735e2412ba7d26acf35ff8d177274e2b278117ccf7
+oid sha256:a860e83f899b4e90ca04decae15751893c788a0a68e4bf4ef98f0325ea54d165
 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": 76927.36162945612,
-      "required_pages_per_second_upper_estimate": 0.005322013373596714
+      "required_token_positions_per_second": 76663.29501248694,
+      "required_pages_per_second_upper_estimate": 0.005303744632315819
     },
     {
       "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": 9615.920203682015,
-      "required_pages_per_second_upper_estimate": 0.0040259849860337085
+      "required_token_positions_per_second": 9582.911876560867,
+      "required_pages_per_second_upper_estimate": 0.004012165088760336
     },
     {
       "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": 19231.84040736403,
-      "required_pages_per_second_upper_estimate": 0.00400548517911616
+      "required_token_positions_per_second": 19165.823753121735,
+      "required_pages_per_second_upper_estimate": 0.0039917356510137374
     },
     {
       "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": 28847.760611046047,
-      "required_pages_per_second_upper_estimate": 0.0017293792006522182
+      "required_token_positions_per_second": 28748.735629682604,
+      "required_pages_per_second_upper_estimate": 0.0017234428042218715
     },
     {
       "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": 28847.760611046047,
-      "required_pages_per_second_upper_estimate": 0.002242786499619828
+      "required_token_positions_per_second": 28748.735629682604,
+      "required_pages_per_second_upper_estimate": 0.002235087742883681
     },
     {
       "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": 28847.760611046047,
-      "required_pages_per_second_upper_estimate": 0.3700280988064038
+      "required_token_positions_per_second": 28748.735629682604,
+      "required_pages_per_second_upper_estimate": 0.3687579126702146
     }
   ],
   "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": 0.11758859633584962,
-    "producer_p95_seconds": 0.2860190520004835,
-    "gpu_update_seconds": 0.9584106153949902,
-    "producer_batches_per_second": 8.504226014773227,
-    "gpu_batches_per_second": 1.0433941193231353,
-    "producer_headroom": 8.150540488276892,
+    "producer_batch_seconds": 0.09382273250031403,
+    "producer_p95_seconds": 0.3345595279970439,
+    "gpu_update_seconds": 0.9617118594757916,
+    "producer_batches_per_second": 10.65839773955265,
+    "gpu_batches_per_second": 1.039812486605997,
+    "producer_headroom": 10.250307509137754,
     "sustainable": true,
-    "buffer_seconds": 60.37986876988438,
+    "buffer_seconds": 60.58784714697487,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 86,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 451012.4654282901,
-    "upload_bytes_per_second": 5097647.062415216,
+    "generation_bytes_per_second": 256556.76260388948,
+    "upload_bytes_per_second": 4861160.122505256,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 96.77677604805496,
+    "estimated_full_shards_per_hour": 55.05110891902459,
     "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": 1.0973008010005287,
-      "bytes": 11080849,
+      "seconds": 0.9614488920014992,
+      "bytes": 6537093,
       "pages": 33,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1580557140005112,
-      "bytes": 6155625,
-      "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8461053749997518,
-      "bytes": 1155308,
-      "pages": 32,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.2314064579986734,
-      "bytes": 7429916,
-      "pages": 31,
+      "seconds": 1.2357967999996617,
+      "bytes": 11997860,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.4280758529966988,
-      "bytes": 11440670,
-      "pages": 39,
+      "seconds": 1.7381740189994161,
+      "bytes": 7938212,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.364633949000563,
-      "bytes": 7290084,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.2239695549978933,
+      "bytes": 5214335,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.0805251250021684,
-      "bytes": 6054722,
-      "pages": 36,
+      "seconds": 1.2570640209996782,
+      "bytes": 6323177,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 2.882830210000975,
-      "bytes": 1047147,
+      "seconds": 1.1513970990017697,
+      "bytes": 2421895,
       "pages": 31,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.6690692930023943,
-      "bytes": 11069831,
-      "pages": 40,
+      "seconds": 0.9733352420007577,
+      "bytes": 6126927,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3508422129998507,
+      "bytes": 16469825,
+      "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 2.282538455998292,
-      "bytes": 6276829,
-      "pages": 36,
+      "seconds": 1.220427556996583,
+      "bytes": 2558602,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.4166184950008756,
-      "bytes": 8285024,
-      "pages": 35,
+      "seconds": 1.1153945629994269,
+      "bytes": 6138686,
+      "pages": 33,
+      "shards": 2
+    },
+    {
+      "seconds": 1.4000027559959562,
+      "bytes": 7623395,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.271756786001788,
-      "bytes": 10114603,
-      "pages": 34,
+      "seconds": 1.0437422140021226,
+      "bytes": 887331,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3929757290025009,
+      "bytes": 11170736,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.2068859079990943,
-      "bytes": 6031296,
-      "pages": 36,
+      "seconds": 1.5524120170011884,
+      "bytes": 8373554,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0317238650022773,
-      "bytes": 2334152,
-      "pages": 28,
+      "seconds": 1.806914691995189,
+      "bytes": 5833325,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.198212211002101,
-      "bytes": 5956669,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1171102930020425,
+      "bytes": 1254155,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.182239486999606,
-      "bytes": 5356316,
-      "pages": 28,
+      "seconds": 1.3684268549986882,
+      "bytes": 11656046,
+      "pages": 37,
+      "shards": 4
+    },
+    {
+      "seconds": 1.130367224002839,
+      "bytes": 6468621,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.588657971999055,
-      "bytes": 13335657,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.1125211710022995,
+      "bytes": 2380554,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.8740047409992258,
-      "bytes": 947216,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.3700068609978189,
+      "bytes": 12013278,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.4317332190003071,
-      "bytes": 7796978,
-      "pages": 38,
+      "seconds": 1.232138457999099,
+      "bytes": 9991874,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.0188206910024746,
-      "bytes": 812811,
-      "pages": 23,
-      "shards": 1
+      "seconds": 1.27796818799834,
+      "bytes": 2573475,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.1825939109985484,
-      "bytes": 10557366,
-      "pages": 46,
-      "shards": 3
+      "seconds": 1.1166499749961076,
+      "bytes": 6085922,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.6627466680001817,
-      "bytes": 7279837,
-      "pages": 30,
-      "shards": 3
+      "seconds": 0.8856082540005445,
+      "bytes": 2405776,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.578546805998485,
-      "bytes": 6985484,
+      "seconds": 1.0989994560004561,
+      "bytes": 10674348,
       "pages": 38,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.4114699799974915,
-      "bytes": 10209990,
-      "pages": 36,
+      "seconds": 1.3233692020003218,
+      "bytes": 11880619,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0779238770010124,
-      "bytes": 1827152,
-      "pages": 11,
+      "seconds": 0.9542889080039458,
+      "bytes": 2355430,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9918526799992833,
-      "bytes": 6095255,
+      "seconds": 1.3224051930010319,
+      "bytes": 6398826,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0435619140007475,
-      "bytes": 6434451,
-      "pages": 37,
+      "seconds": 1.916404732997762,
+      "bytes": 5130348,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1101431559982302,
-      "bytes": 7100100,
-      "pages": 31,
+      "seconds": 1.503245963001973,
+      "bytes": 7636802,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0310441840010753,
-      "bytes": 5933626,
-      "pages": 36,
+      "seconds": 1.082158730001538,
+      "bytes": 6031838,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.339776704000542,
-      "bytes": 13443293,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.433015736998641,
+      "bytes": 8073393,
+      "pages": 27,
+      "shards": 3
     },
     {
-      "seconds": 1.7319635569983802,
-      "bytes": 934829,
-      "pages": 27,
+      "seconds": 1.7437503740002285,
+      "bytes": 10517926,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3904311470032553,
+      "bytes": 899587,
+      "pages": 31,
       "shards": 1
     },
     {
-      "seconds": 1.0800791440015018,
-      "bytes": 6107947,
-      "pages": 36,
+      "seconds": 1.836203224993369,
+      "bytes": 7625671,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 1.115198325001984,
+      "bytes": 5971301,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2829358429989952,
-      "bytes": 6479810,
-      "pages": 33,
+      "seconds": 1.9192805860002409,
+      "bytes": 6532250,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.002878891002183,
-      "bytes": 6148473,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.5230046290016617,
+      "bytes": 11668778,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.581660068997735,
-      "bytes": 6199462,
-      "pages": 37,
+      "seconds": 2.0409051120004733,
+      "bytes": 6281809,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3033671679986583,
-      "bytes": 13020043,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.1541477030041278,
+      "bytes": 6906912,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1211934040002234,
-      "bytes": 5810769,
+      "seconds": 1.7220873259939253,
+      "bytes": 6147366,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2197048900015943,
-      "bytes": 2772365,
+      "seconds": 3.3549275529949227,
+      "bytes": 2682623,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.9406323449984484,
-      "bytes": 5788082,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5139052369995625,
-      "bytes": 6128565,
+      "seconds": 1.4113562120037386,
+      "bytes": 6081250,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9614488920014992,
-      "bytes": 6537093,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2357967999996617,
-      "bytes": 11997860,
-      "pages": 35,
+      "seconds": 1.8620396420010366,
+      "bytes": 10625025,
+      "pages": 23,
       "shards": 3
     },
     {
-      "seconds": 1.7381740189994161,
-      "bytes": 7938212,
-      "pages": 40,
+      "seconds": 1.826674156000081,
+      "bytes": 7716672,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.2239695549978933,
-      "bytes": 5214335,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2570640209996782,
-      "bytes": 6323177,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1513970990017697,
-      "bytes": 2421895,
+      "seconds": 1.3658487110005808,
+      "bytes": 5996229,
       "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 0.9733352420007577,
-      "bytes": 6126927,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.6789640829956625,
+      "bytes": 1167744,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.3508422129998507,
-      "bytes": 16469825,
-      "pages": 33,
+      "seconds": 1.3386350959990523,
+      "bytes": 11890624,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 1.220427556996583,
-      "bytes": 2558602,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1153945629994269,
-      "bytes": 6138686,
-      "pages": 33,
+      "seconds": 1.2723760490043787,
+      "bytes": 6238095,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.4000027559959562,
-      "bytes": 7623395,
-      "pages": 39,
+      "seconds": 1.1314651699940441,
+      "bytes": 8436407,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0437422140021226,
-      "bytes": 887331,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.675168465000752,
+      "bytes": 5970136,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.3929757290025009,
-      "bytes": 11170736,
+      "seconds": 1.2014365330032888,
+      "bytes": 5687823,
       "pages": 35,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.5524120170011884,
-      "bytes": 8373554,
+      "seconds": 1.1749995130012394,
+      "bytes": 7282459,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.806914691995189,
-      "bytes": 5833325,
-      "pages": 30,
+      "seconds": 1.3900761400000192,
+      "bytes": 5687350,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1171102930020425,
-      "bytes": 1254155,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0886611489986535,
+      "bytes": 2750948,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.3684268549986882,
-      "bytes": 11656046,
-      "pages": 37,
+      "seconds": 2.3056074749983964,
+      "bytes": 15312112,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.130367224002839,
-      "bytes": 6468621,
-      "pages": 37,
+      "seconds": 1.2607366750016809,
+      "bytes": 5997714,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.1125211710022995,
-      "bytes": 2380554,
-      "pages": 28,
+      "seconds": 1.251247890999366,
+      "bytes": 2744476,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3700068609978189,
-      "bytes": 12013278,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.2621945279970532,
+      "bytes": 5911926,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.232138457999099,
-      "bytes": 9991874,
-      "pages": 36,
+      "seconds": 1.4384337969968328,
+      "bytes": 10703677,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.27796818799834,
-      "bytes": 2573475,
-      "pages": 32,
+      "seconds": 1.1966409989981912,
+      "bytes": 2669509,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1166499749961076,
-      "bytes": 6085922,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.7882141709997086,
+      "bytes": 11758267,
+      "pages": 28,
+      "shards": 3
     },
     {
-      "seconds": 0.8856082540005445,
-      "bytes": 2405776,
-      "pages": 29,
+      "seconds": 1.0542783380005858,
+      "bytes": 2768181,
+      "pages": 35,
       "shards": 2
+    },
+    {
+      "seconds": 1.5631635209938395,
+      "bytes": 10280055,
+      "pages": 36,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:04389db27054d0deb18e21ea9848cc49ae3250a62872254d899b5d9fb8463946
-size 721760703
+oid sha256:1b1f259fdc0e4e7c034af4b926b3e23d2a45e4a91f9e75f8c00269d04f243d83
+size 721774335