Qwen90M step 430000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 425000,
+  "step": 430000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "e6a8119feebeb9b4f48e3d9e2dc5415b553b928301f4bcae69173e0b92d62b00",
+    "model.safetensors": "55726e70d1f08038bb9239098dc07428f7dbb69a155603511a8d493d3b3d1bb8",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "b280b0d64c4eb0d14a1021d0c9cdbedc4c0bf8eb245a63c69257180b96af6158",
+    "trainer.pt": "ae57f71e3f57e49a3027c7cc6ec8a729344948717d72a54e4f742c154a0c26a4",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18",
-    "shard_scheduler.json": "53c18fdc60e1935cc354da354050e76668d361282678ac8d9bee2b8f0811a722",
-    "coverage.json": "3b07084bb276dfdd5fdda54e8af8ef1840ed097f7c487c84187259f480a569a9",
+    "shard_scheduler.json": "ea42642d1a5ca11fe1d3250a4c5ef108721b3514388692e82cb5ad01efa41e05",
+    "coverage.json": "5ff68de276024a1ae2e7817b08e75a238d264d40560ae7e48097661cad1ef6c8",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 425000,
+  "step": 430000,
   "stats": {
-    "climbmix/blocks": 15300000,
-    "climbmix/targets": 25576792778,
-    "climbmix/input_tokens": 25578412852,
-    "ultra_qa/blocks": 5737500,
-    "ultra_qa/targets": 9359436204,
-    "ultra_qa/input_tokens": 9359454150,
-    "cortex/blocks": 5737500,
-    "cortex/targets": 218211819,
-    "cortex/input_tokens": 9120392161,
-    "ultra_style/blocks": 5737500,
-    "ultra_style/targets": 9075454296,
-    "ultra_style/input_tokens": 9075929780,
-    "openmath/blocks": 3825000,
-    "openmath/targets": 5910673230,
-    "openmath/input_tokens": 5912642746,
-    "rewrite6/blocks": 1912500,
-    "rewrite6/targets": 3135273649,
-    "rewrite6/input_tokens": 3138711114
+    "climbmix/blocks": 15480000,
+    "climbmix/targets": 25877789085,
+    "climbmix/input_tokens": 25879428024,
+    "ultra_qa/blocks": 5805000,
+    "ultra_qa/targets": 9469588245,
+    "ultra_qa/input_tokens": 9469606395,
+    "cortex/blocks": 5805000,
+    "cortex/targets": 220770662,
+    "cortex/input_tokens": 9227785254,
+    "ultra_style/blocks": 5805000,
+    "ultra_style/targets": 9181956738,
+    "ultra_style/input_tokens": 9182438095,
+    "openmath/blocks": 3870000,
+    "openmath/targets": 5980155459,
+    "openmath/input_tokens": 5982148264,
+    "rewrite6/blocks": 1935000,
+    "rewrite6/targets": 3172177358,
+    "rewrite6/input_tokens": 3175655197
   },
   "sources": {
     "climbmix": {
-      "page": 2095,
-      "row": 1309,
+      "page": 2119,
+      "row": 3598,
       "start": {
-        "file": 101,
+        "file": 102,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 34,
+          "row_group": 47,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 8582429
+      "documents": 8683022
     },
     "rewrite6": {
-      "page": 42,
-      "row": 888,
+      "page": 6,
+      "row": 373,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 172032
+          "row": 24576
         }
       },
-      "epoch": 23,
-      "offset": 0,
-      "documents": 4897212
+      "epoch": 24,
+      "offset": 4096,
+      "documents": 4954645
     },
     "openmath": {
-      "page": 1309,
-      "row": 2292,
+      "page": 1325,
+      "row": 120,
       "start": {
         "file": 12,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 121,
-          "row": 864
+          "row_group": 187,
+          "row": 400
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 5363956
+      "documents": 5427320
     },
     "ultra_qa": {
-      "page": 566,
-      "row": 874,
+      "page": 572,
+      "row": 3580,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 46098
+          "row_group": 4,
+          "row": 6126
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 2319210
+      "offset": 4096,
+      "documents": 2346492
     },
     "ultra_style": {
-      "page": 864,
-      "row": 304,
+      "page": 874,
+      "row": 1747,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 29158
+          "row": 70118
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3539248
+      "documents": 3581651
     },
     "cortex": {
-      "page": 111278,
-      "row": 123,
+      "page": 112589,
+      "row": 61,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 14243707
+      "documents": 14411453
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 425000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e6a8119feebeb9b4f48e3d9e2dc5415b553b928301f4bcae69173e0b92d62b00", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b280b0d64c4eb0d14a1021d0c9cdbedc4c0bf8eb245a63c69257180b96af6158", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "53c18fdc60e1935cc354da354050e76668d361282678ac8d9bee2b8f0811a722", "coverage.json": "3b07084bb276dfdd5fdda54e8af8ef1840ed097f7c487c84187259f480a569a9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000425000"}
+{"step": 430000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "55726e70d1f08038bb9239098dc07428f7dbb69a155603511a8d493d3b3d1bb8", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ae57f71e3f57e49a3027c7cc6ec8a729344948717d72a54e4f742c154a0c26a4", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "ea42642d1a5ca11fe1d3250a4c5ef108721b3514388692e82cb5ad01efa41e05", "coverage.json": "5ff68de276024a1ae2e7817b08e75a238d264d40560ae7e48097661cad1ef6c8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000430000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e6a8119feebeb9b4f48e3d9e2dc5415b553b928301f4bcae69173e0b92d62b00
+oid sha256:55726e70d1f08038bb9239098dc07428f7dbb69a155603511a8d493d3b3d1bb8
 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": 76980.58360649772,
-      "required_pages_per_second_upper_estimate": 0.005325695393460453
+      "required_token_positions_per_second": 76292.77819935197,
+      "required_pages_per_second_upper_estimate": 0.005278111419465689
     },
     {
       "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": 9622.572950812215,
-      "required_pages_per_second_upper_estimate": 0.004028770352331965
+      "required_token_positions_per_second": 9536.597274918997,
+      "required_pages_per_second_upper_estimate": 0.003992774132211747
     },
     {
       "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": 19245.14590162443,
-      "required_pages_per_second_upper_estimate": 0.004008256362681121
+      "required_token_positions_per_second": 19073.194549837994,
+      "required_pages_per_second_upper_estimate": 0.0039724434307660955
     },
     {
       "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": 28867.718852436643,
-      "required_pages_per_second_upper_estimate": 0.0017305756667491143
+      "required_token_positions_per_second": 28609.79182475699,
+      "required_pages_per_second_upper_estimate": 0.0017151133352714869
     },
     {
       "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": 28867.718852436643,
-      "required_pages_per_second_upper_estimate": 0.00224433816510092
+      "required_token_positions_per_second": 28609.79182475699,
+      "required_pages_per_second_upper_estimate": 0.0022242854731999244
     },
     {
       "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": 28867.718852436643,
-      "required_pages_per_second_upper_estimate": 0.3702841016974724
+      "required_token_positions_per_second": 28609.79182475699,
+      "required_pages_per_second_upper_estimate": 0.3669756907268633
     }
   ],
   "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": 704566,
-    "producer_batch_seconds": 0.4470424027734907,
-    "producer_p95_seconds": 0.30381212399879587,
-    "gpu_update_seconds": 0.9577480001564553,
-    "producer_batches_per_second": 2.23692426891926,
-    "gpu_batches_per_second": 1.0441159885863949,
-    "producer_headroom": 2.142409745058862,
+    "producer_batch_seconds": 0.09154915285955667,
+    "producer_p95_seconds": 0.3261282250023214,
+    "gpu_update_seconds": 0.9663824249177262,
+    "producer_batches_per_second": 10.923093974818922,
+    "gpu_batches_per_second": 1.034787030698676,
+    "producer_headroom": 10.555886042989714,
     "sustainable": true,
-    "buffer_seconds": 60.338124009856685,
+    "buffer_seconds": 60.88209276981675,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 77,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 399755.21465372964,
-    "upload_bytes_per_second": 5963845.868077777,
+    "generation_bytes_per_second": 336081.2992223133,
+    "upload_bytes_per_second": 5716296.3033110965,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 85.77816323956411,
+    "estimated_full_shards_per_hour": 72.1152232408719,
     "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": 1.0381844509975053,
-      "bytes": 7741032,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5200233909999952,
-      "bytes": 6160062,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1740434719977202,
-      "bytes": 8468925,
-      "pages": 41,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0248841799984802,
-      "bytes": 5963815,
-      "pages": 27,
+      "seconds": 0.8803229269979056,
+      "bytes": 4831371,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.231629100999271,
-      "bytes": 4969209,
+      "seconds": 1.1803598539991071,
+      "bytes": 6473769,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.249853739001992,
-      "bytes": 7694709,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.0211126419999346,
+      "bytes": 2557759,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.2034739760019875,
-      "bytes": 5697150,
+      "seconds": 1.475009352001507,
+      "bytes": 6011684,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9529857010020351,
-      "bytes": 6624046,
-      "pages": 40,
+      "seconds": 1.40803722399869,
+      "bytes": 10317954,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1070565920017543,
-      "bytes": 11831537,
-      "pages": 32,
+      "seconds": 1.058006204002595,
+      "bytes": 8327609,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.0998464429976593,
-      "bytes": 6314908,
-      "pages": 36,
+      "seconds": 1.1640828560011869,
+      "bytes": 6135465,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0093370079994202,
-      "bytes": 2319025,
-      "pages": 29,
+      "seconds": 0.8637312579994614,
+      "bytes": 2324435,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.116415386000881,
-      "bytes": 10937725,
+      "seconds": 1.1669898459986143,
+      "bytes": 10131671,
       "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 0.8731399669995881,
-      "bytes": 778500,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.0318543029970897,
+      "bytes": 5757385,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.033372367000993,
-      "bytes": 7683491,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9449253849998058,
+      "bytes": 2649946,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.4393485709988454,
-      "bytes": 11721944,
-      "pages": 36,
+      "seconds": 1.2970311710014357,
+      "bytes": 11562430,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.162564084999758,
-      "bytes": 5892604,
-      "pages": 23,
-      "shards": 3
+      "seconds": 1.4141549599989958,
+      "bytes": 12336426,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.082973058000789,
-      "bytes": 6647343,
-      "pages": 45,
-      "shards": 2
+      "seconds": 0.8797184569993988,
+      "bytes": 870056,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.1072141639997426,
-      "bytes": 6205762,
+      "seconds": 1.3180106779982452,
+      "bytes": 6077154,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9200315849993785,
-      "bytes": 2298271,
-      "pages": 29,
+      "seconds": 0.8500619599981292,
+      "bytes": 2427201,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.584766658001172,
-      "bytes": 10894182,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.7313189029991918,
+      "bytes": 10259217,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.1065202139980101,
-      "bytes": 8432247,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.5030511410004692,
+      "bytes": 13228107,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.2434194350025791,
-      "bytes": 6179718,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8756032579985913,
+      "bytes": 1232264,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.03145402500013,
-      "bytes": 6170934,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.2522269889996096,
+      "bytes": 10696457,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 0.9970344739995198,
-      "bytes": 6228444,
-      "pages": 33,
+      "seconds": 1.1242637430004834,
+      "bytes": 7345030,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.8807951420021709,
-      "bytes": 343455,
-      "pages": 9,
+      "seconds": 0.8663610330004303,
+      "bytes": 1077995,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1079528180016496,
-      "bytes": 5894448,
-      "pages": 27,
-      "shards": 3
+      "seconds": 0.9089211409991549,
+      "bytes": 5857565,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.0535939100009273,
-      "bytes": 7640568,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.1926017650002905,
+      "bytes": 13608076,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 0.9312373259999731,
-      "bytes": 6794347,
-      "pages": 32,
+      "seconds": 0.8606232550009736,
+      "bytes": 4749519,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.2384252730007574,
-      "bytes": 10759915,
-      "pages": 34,
+      "seconds": 1.311676024000917,
+      "bytes": 7627867,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.8831225600006292,
-      "bytes": 2598975,
-      "pages": 38,
-      "shards": 2
+      "seconds": 0.9741381140011072,
+      "bytes": 986302,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 1.0244675720023224,
-      "bytes": 5791149,
-      "pages": 27,
+      "seconds": 1.1853109039984702,
+      "bytes": 6001312,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.2756652919997578,
-      "bytes": 7797831,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.160805725998216,
+      "bytes": 12260097,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 0.9431873399989854,
-      "bytes": 5431625,
+      "seconds": 0.9388873019997845,
+      "bytes": 6924644,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.3104591960000107,
-      "bytes": 11823249,
+      "seconds": 1.2151130479978747,
+      "bytes": 5819277,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1440478279982926,
+      "bytes": 2552467,
       "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1846777530008694,
+      "bytes": 10218604,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1533994220008026,
-      "bytes": 7669461,
-      "pages": 39,
+      "seconds": 1.1278007279979647,
+      "bytes": 7339680,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.8288239209978201,
-      "bytes": 961455,
-      "pages": 27,
+      "seconds": 1.694962838999345,
+      "bytes": 1221938,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.3137797929994122,
-      "bytes": 10288628,
-      "pages": 39,
+      "seconds": 1.4893900960014435,
+      "bytes": 11508997,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.1022804700005508,
-      "bytes": 2387118,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.178049504000228,
+      "bytes": 7502276,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.1524969450001663,
-      "bytes": 6155701,
-      "pages": 35,
+      "seconds": 0.9951034849982534,
+      "bytes": 5795502,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4659404199992423,
-      "bytes": 13390262,
+      "seconds": 0.9211728519985627,
+      "bytes": 7602156,
       "pages": 39,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.8803229269979056,
-      "bytes": 4831371,
-      "pages": 29,
-      "shards": 2
+      "seconds": 0.7880418419990747,
+      "bytes": 1256366,
+      "pages": 36,
+      "shards": 1
     },
     {
-      "seconds": 1.1803598539991071,
-      "bytes": 6473769,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1164300489981542,
+      "bytes": 9840833,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.0211126419999346,
-      "bytes": 2557759,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.1696946029987885,
+      "bytes": 13219787,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.475009352001507,
-      "bytes": 6011684,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8621246889997565,
+      "bytes": 1265428,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.40803722399869,
-      "bytes": 10317954,
-      "pages": 35,
+      "seconds": 1.4639245990001655,
+      "bytes": 10067426,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.058006204002595,
-      "bytes": 8327609,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.8352084129983268,
+      "bytes": 2708130,
+      "pages": 39,
+      "shards": 2
     },
     {
-      "seconds": 1.1640828560011869,
-      "bytes": 6135465,
+      "seconds": 0.9732167729998764,
+      "bytes": 6207933,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8637312579994614,
-      "bytes": 2324435,
-      "pages": 28,
+      "seconds": 0.9747246619990619,
+      "bytes": 7297014,
+      "pages": 31,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0759720290006953,
+      "bytes": 6305836,
+      "pages": 19,
       "shards": 2
     },
     {
-      "seconds": 1.1669898459986143,
-      "bytes": 10131671,
-      "pages": 41,
+      "seconds": 1.0651401020004414,
+      "bytes": 11486452,
+      "pages": 47,
       "shards": 3
     },
     {
-      "seconds": 1.0318543029970897,
-      "bytes": 5757385,
-      "pages": 28,
+      "seconds": 1.0763056159994449,
+      "bytes": 2441249,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9449253849998058,
-      "bytes": 2649946,
-      "pages": 37,
+      "seconds": 4.102204398001049,
+      "bytes": 6262944,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2970311710014357,
-      "bytes": 11562430,
-      "pages": 33,
+      "seconds": 1.1364409210000304,
+      "bytes": 7535674,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.4141549599989958,
-      "bytes": 12336426,
-      "pages": 42,
-      "shards": 4
+      "seconds": 0.9002165780002542,
+      "bytes": 5078039,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.8797184569993988,
-      "bytes": 870056,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.1330570440004522,
+      "bytes": 12010212,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.3180106779982452,
-      "bytes": 6077154,
-      "pages": 37,
+      "seconds": 1.1789866869985417,
+      "bytes": 2676190,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 0.8500619599981292,
-      "bytes": 2427201,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.2169118900019384,
+      "bytes": 10000231,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.7313189029991918,
-      "bytes": 10259217,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.287032372001704,
+      "bytes": 6056411,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.5030511410004692,
-      "bytes": 13228107,
+      "seconds": 1.0081800109983305,
+      "bytes": 2487942,
       "pages": 34,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 0.8756032579985913,
-      "bytes": 1232264,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1110187749982288,
+      "bytes": 6282739,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.2522269889996096,
-      "bytes": 10696457,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.3052810369990766,
+      "bytes": 12708434,
+      "pages": 30,
+      "shards": 4
     },
     {
-      "seconds": 1.1242637430004834,
-      "bytes": 7345030,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.1760405560016807,
+      "bytes": 6277751,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.8663610330004303,
-      "bytes": 1077995,
+      "seconds": 1.038220692997129,
+      "bytes": 6391937,
       "pages": 35,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 0.9089211409991549,
-      "bytes": 5857565,
-      "pages": 34,
+      "seconds": 0.970617956998467,
+      "bytes": 2311552,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.1926017650002905,
-      "bytes": 13608076,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.171514493998984,
+      "bytes": 9915176,
+      "pages": 34,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b280b0d64c4eb0d14a1021d0c9cdbedc4c0bf8eb245a63c69257180b96af6158
-size 721780415
+oid sha256:ae57f71e3f57e49a3027c7cc6ec8a729344948717d72a54e4f742c154a0c26a4
+size 721760319