Qwen90M step 620000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 615000,
+  "step": 620000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "5dafb78e58d2434052a2f73226b28c4135549c3ad84f91855ab29a59004932b9",
+    "model.safetensors": "5e1c8846d93cc736a471c45613d78007b35b8e94c1285a2904ac37057a5c3356",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "8a7925c71cd513cd3d5f53037b00418d6c7a7f8c7b424e1f49846780b2108ebe",
+    "trainer.pt": "9933156547ab18b657069aa60eb52e3b669e25faa068a6484883abec26f0f1a7",
     "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb",
     "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e",
-    "shard_scheduler.json": "9c4ffe74747c4cccb687d247aca424092ee0f5baab8450b599ecbfb62af1ef3d",
-    "coverage.json": "caa1ddfdddd2441adccd3fe63c7891dea469bcdb5d7199604e60c94b258433d9",
+    "shard_scheduler.json": "b2d31325d348832dedfe0eeb8c5de7f5606e001abaa7474b129c45c2b3901479",
+    "coverage.json": "bb539a2055773d2d68de8915bf0ebe63f0b3e8e118e9bf2703f2ab7066ea0769",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 615000,
+  "step": 620000,
   "stats": {
-    "climbmix/blocks": 22140000,
-    "climbmix/targets": 37010537230,
-    "climbmix/input_tokens": 37012882752,
-    "ultra_qa/blocks": 8302500,
-    "ultra_qa/targets": 13544842671,
-    "ultra_qa/input_tokens": 13544868823,
-    "cortex/blocks": 8302500,
-    "cortex/targets": 315671192,
-    "cortex/input_tokens": 13197429410,
-    "ultra_style/blocks": 8302500,
-    "ultra_style/targets": 13131878532,
-    "ultra_style/input_tokens": 13132567915,
-    "openmath/blocks": 5535000,
-    "openmath/targets": 8552607269,
-    "openmath/input_tokens": 8555456133,
-    "rewrite6/blocks": 2767500,
-    "rewrite6/targets": 4536948321,
-    "rewrite6/input_tokens": 4541922859
+    "climbmix/blocks": 22320000,
+    "climbmix/targets": 37312355817,
+    "climbmix/input_tokens": 37314720200,
+    "ultra_qa/blocks": 8370000,
+    "ultra_qa/targets": 13655080912,
+    "ultra_qa/input_tokens": 13655107298,
+    "cortex/blocks": 8370000,
+    "cortex/targets": 318257872,
+    "cortex/input_tokens": 13304792682,
+    "ultra_style/blocks": 8370000,
+    "ultra_style/targets": 13238688750,
+    "ultra_style/input_tokens": 13239383690,
+    "openmath/blocks": 5580000,
+    "openmath/targets": 8622121150,
+    "openmath/input_tokens": 8624993272,
+    "rewrite6/blocks": 2790000,
+    "rewrite6/targets": 4573810627,
+    "rewrite6/input_tokens": 4578825235
   },
   "sources": {
     "climbmix": {
-      "page": 3032,
-      "row": 1301,
+      "page": 3056,
+      "row": 2715,
       "start": {
-        "file": 146,
+        "file": 147,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 62,
+          "row_group": 75,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 12420373
+      "offset": 2048,
+      "documents": 12520091
     },
     "rewrite6": {
-      "page": 25,
-      "row": 724,
+      "page": 39,
+      "row": 610,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 102400
+          "row": 159744
         }
       },
       "epoch": 34,
       "offset": 0,
-      "documents": 7086860
+      "documents": 7144090
     },
     "openmath": {
-      "page": 1894,
-      "row": 3308,
+      "page": 1910,
+      "row": 970,
       "start": {
         "file": 17,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 334,
-          "row": 774
+          "row_group": 400,
+          "row": 310
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 7761132
+      "offset": 2048,
+      "documents": 7824330
     },
     "ultra_qa": {
-      "page": 819,
-      "row": 1438,
+      "page": 826,
+      "row": 27,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 43319
+          "row_group": 4,
+          "row": 7499
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 3356062
+      "offset": 2048,
+      "documents": 3383323
     },
     "ultra_style": {
-      "page": 1250,
-      "row": 3208,
+      "page": 1260,
+      "row": 3840,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 68659
+          "row_group": 4,
+          "row": 25036
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 5123208
+      "offset": 0,
+      "documents": 5164800
     },
     "cortex": {
-      "page": 161021,
-      "row": 86,
+      "page": 162331,
+      "row": 125,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 20610774
+      "documents": 20778493
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 615000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "5dafb78e58d2434052a2f73226b28c4135549c3ad84f91855ab29a59004932b9", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "8a7925c71cd513cd3d5f53037b00418d6c7a7f8c7b424e1f49846780b2108ebe", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e", "shard_scheduler.json": "9c4ffe74747c4cccb687d247aca424092ee0f5baab8450b599ecbfb62af1ef3d", "coverage.json": "caa1ddfdddd2441adccd3fe63c7891dea469bcdb5d7199604e60c94b258433d9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000615000"}
+{"step": 620000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "5e1c8846d93cc736a471c45613d78007b35b8e94c1285a2904ac37057a5c3356", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "9933156547ab18b657069aa60eb52e3b669e25faa068a6484883abec26f0f1a7", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e", "shard_scheduler.json": "b2d31325d348832dedfe0eeb8c5de7f5606e001abaa7474b129c45c2b3901479", "coverage.json": "bb539a2055773d2d68de8915bf0ebe63f0b3e8e118e9bf2703f2ab7066ea0769", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000620000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5dafb78e58d2434052a2f73226b28c4135549c3ad84f91855ab29a59004932b9
+oid sha256:5e1c8846d93cc736a471c45613d78007b35b8e94c1285a2904ac37057a5c3356
 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": 76847.5007153862,
-      "required_pages_per_second_upper_estimate": 0.005316488410258506
+      "required_token_positions_per_second": 76942.96702956906,
+      "required_pages_per_second_upper_estimate": 0.005323092991386044
     },
     {
       "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": 9605.937589423274,
-      "required_pages_per_second_upper_estimate": 0.004021805473904264
+      "required_token_positions_per_second": 9617.870878696132,
+      "required_pages_per_second_upper_estimate": 0.0040268016929273925
     },
     {
       "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": 19211.87517884655,
-      "required_pages_per_second_upper_estimate": 0.004001326948534451
+      "required_token_positions_per_second": 19235.741757392265,
+      "required_pages_per_second_upper_estimate": 0.004006297727441522
     },
     {
       "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": 28817.812768269825,
-      "required_pages_per_second_upper_estimate": 0.0017275838732054918
+      "required_token_positions_per_second": 28853.612636088397,
+      "required_pages_per_second_upper_estimate": 0.0017297300206179822
     },
     {
       "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": 28817.812768269825,
-      "required_pages_per_second_upper_estimate": 0.0022404581865706156
+      "required_token_positions_per_second": 28853.612636088397,
+      "required_pages_per_second_upper_estimate": 0.0022432414688265327
     },
     {
       "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": 28817.812768269825,
-      "required_pages_per_second_upper_estimate": 0.36964396003475875
+      "required_token_positions_per_second": 28853.612636088397,
+      "required_pages_per_second_upper_estimate": 0.37010316230023216
     }
   ],
   "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": 720750,
-    "producer_batch_seconds": 0.11640952138283467,
-    "producer_p95_seconds": 0.20131237399982638,
-    "gpu_update_seconds": 0.9594066080699275,
-    "producer_batches_per_second": 8.590362610557529,
-    "gpu_batches_per_second": 1.042310936352352,
-    "producer_headroom": 8.241650654285726,
+    "producer_batch_seconds": 0.18642643738270692,
+    "producer_p95_seconds": 0.23165708300075494,
+    "gpu_update_seconds": 0.9582162327021578,
+    "producer_batches_per_second": 5.364046076507607,
+    "gpu_batches_per_second": 1.0436057811085213,
+    "producer_headroom": 5.139916023471909,
     "sustainable": true,
-    "buffer_seconds": 60.442616308405434,
+    "buffer_seconds": 60.367622660235945,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
     "upload_pages_per_commit": 81,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 310821.5374057803,
-    "upload_bytes_per_second": 5657201.142948646,
+    "generation_bytes_per_second": 480694.27939603303,
+    "upload_bytes_per_second": 5198415.335503838,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 66.69506637220437,
+    "estimated_full_shards_per_hour": 103.14580236826652,
     "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": 1.4455456679997951,
-      "bytes": 5746642,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.2262302330000239,
+      "bytes": 12002858,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 0.88688044599985,
-      "bytes": 6234900,
+      "seconds": 0.976691177000248,
+      "bytes": 6244802,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1657635070000651,
-      "bytes": 12844748,
-      "pages": 34,
-      "shards": 5
+      "seconds": 1.5793915419999394,
+      "bytes": 781594,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.0483822090000103,
-      "bytes": 6048562,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.4397608880008193,
+      "bytes": 13440043,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.1226819749999777,
-      "bytes": 7652419,
+      "seconds": 1.215214173998902,
+      "bytes": 10529948,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3251570170000377,
-      "bytes": 1162985,
-      "pages": 33,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4578549020002356,
-      "bytes": 10422176,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.3451746110004024,
+      "bytes": 2254637,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.1834071509997557,
-      "bytes": 2468071,
-      "pages": 31,
+      "seconds": 1.1764542460005032,
+      "bytes": 6191157,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1340931629997613,
-      "bytes": 11786170,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.0847412589992018,
+      "bytes": 1126127,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 0.9738086980005392,
-      "bytes": 6055828,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2304532270009076,
+      "bytes": 11321849,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.397671879999507,
-      "bytes": 6213114,
-      "pages": 33,
+      "seconds": 1.4404141169998184,
+      "bytes": 12120623,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0006439850003517,
-      "bytes": 6195776,
-      "pages": 35,
+      "seconds": 0.8672846369991021,
+      "bytes": 2691618,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0036822970005232,
-      "bytes": 2451346,
-      "pages": 32,
+      "seconds": 0.9980244999987917,
+      "bytes": 5788373,
+      "pages": 22,
       "shards": 2
     },
     {
-      "seconds": 1.0700342689997342,
-      "bytes": 10599218,
-      "pages": 33,
-      "shards": 3
+      "seconds": 3.3476856620000035,
+      "bytes": 5904691,
+      "pages": 43,
+      "shards": 2
     },
     {
-      "seconds": 1.020371795999381,
-      "bytes": 12084005,
-      "pages": 40,
+      "seconds": 1.2678156700003456,
+      "bytes": 7637773,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9868787969999175,
-      "bytes": 2703362,
-      "pages": 37,
+      "seconds": 1.1486175610007194,
+      "bytes": 5966085,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1407470789999934,
-      "bytes": 6061047,
-      "pages": 27,
+      "seconds": 1.0020495059998211,
+      "bytes": 6673858,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.7620159190000777,
-      "bytes": 5407000,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.1948771370007307,
+      "bytes": 12132833,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.0307093159999567,
-      "bytes": 7655434,
+      "seconds": 1.2424271460004093,
+      "bytes": 1081337,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3408090630000515,
+      "bytes": 7658166,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0102191360001598,
-      "bytes": 6169276,
-      "pages": 31,
+      "seconds": 0.8867853440006002,
+      "bytes": 5785267,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.2401527040001383,
-      "bytes": 8401681,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.081288327000948,
+      "bytes": 5787210,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.6231297679996715,
-      "bytes": 9980497,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.1759565290012688,
+      "bytes": 13132586,
+      "pages": 31,
+      "shards": 4
     },
     {
-      "seconds": 1.1220608109997556,
-      "bytes": 7908917,
+      "seconds": 1.004212905998429,
+      "bytes": 6157452,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9769788749999861,
-      "bytes": 875600,
+      "seconds": 1.3965238540004066,
+      "bytes": 817171,
       "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 0.8441450039999836,
-      "bytes": 578644,
-      "pages": 17,
-      "shards": 1
-    },
-    {
-      "seconds": 1.364240728000368,
-      "bytes": 10856561,
+      "seconds": 1.115737056999933,
+      "bytes": 11496202,
       "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2959960469997895,
-      "bytes": 13446052,
-      "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 0.8787024550001661,
-      "bytes": 837425,
-      "pages": 26,
+      "seconds": 1.037322108000808,
+      "bytes": 1047345,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.2931989250000697,
-      "bytes": 7658868,
-      "pages": 39,
+      "seconds": 1.0591198959991743,
+      "bytes": 7189815,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2650147490003292,
-      "bytes": 6116564,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.272207456000615,
+      "bytes": 11979936,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.0610393269998895,
-      "bytes": 5051210,
-      "pages": 35,
+      "seconds": 1.1375752950007154,
+      "bytes": 4905532,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.4254050959998494,
-      "bytes": 7441828,
-      "pages": 29,
+      "seconds": 1.533432411000831,
+      "bytes": 7357007,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1509858810004516,
-      "bytes": 6964137,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.1559075860004668,
+      "bytes": 1149523,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.1934967530005451,
-      "bytes": 5866491,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.0760069800016936,
+      "bytes": 5947734,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.2554875369996807,
-      "bytes": 11949834,
-      "pages": 44,
+      "seconds": 1.2793344790006813,
+      "bytes": 12370460,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 0.8095207870001104,
-      "bytes": 1155021,
-      "pages": 34,
+      "seconds": 0.8282303630003298,
+      "bytes": 647721,
+      "pages": 20,
       "shards": 1
     },
     {
-      "seconds": 1.4986395109999648,
-      "bytes": 7663876,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.3377682740010641,
+      "bytes": 13564102,
+      "pages": 48,
+      "shards": 4
     },
     {
-      "seconds": 0.9853112940008941,
-      "bytes": 5162186,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.359001171998898,
+      "bytes": 1213849,
+      "pages": 32,
+      "shards": 1
     },
     {
-      "seconds": 1.2030387599997994,
-      "bytes": 11763818,
-      "pages": 37,
+      "seconds": 1.1420412400002533,
+      "bytes": 10074433,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0410187759998735,
-      "bytes": 7475284,
-      "pages": 38,
+      "seconds": 1.157611688000543,
+      "bytes": 7621104,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.175038504999975,
-      "bytes": 877584,
+      "seconds": 1.0118182780006464,
+      "bytes": 802933,
       "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.2262302330000239,
-      "bytes": 12002858,
-      "pages": 41,
-      "shards": 4
-    },
-    {
-      "seconds": 0.976691177000248,
-      "bytes": 6244802,
-      "pages": 36,
+      "seconds": 1.0449377190016094,
+      "bytes": 6509260,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5793915419999394,
-      "bytes": 781594,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4397608880008193,
-      "bytes": 13440043,
-      "pages": 39,
+      "seconds": 1.2692650000008143,
+      "bytes": 13237909,
+      "pages": 41,
       "shards": 4
     },
     {
-      "seconds": 1.215214173998902,
-      "bytes": 10529948,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.122997641999973,
+      "bytes": 6774100,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.3451746110004024,
-      "bytes": 2254637,
+      "seconds": 2.5239626219990896,
+      "bytes": 2432130,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1764542460005032,
-      "bytes": 6191157,
+      "seconds": 1.0105871389987442,
+      "bytes": 6118542,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0847412589992018,
-      "bytes": 1126127,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.3655256990005,
+      "bytes": 10029580,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.2304532270009076,
-      "bytes": 11321849,
-      "pages": 33,
-      "shards": 4
+      "seconds": 0.8411326719997305,
+      "bytes": 2272552,
+      "pages": 26,
+      "shards": 2
     },
     {
-      "seconds": 1.4404141169998184,
-      "bytes": 12120623,
+      "seconds": 1.4776745150011266,
+      "bytes": 12319008,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1878520110003592,
+      "bytes": 6586003,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.8672846369991021,
-      "bytes": 2691618,
-      "pages": 37,
+      "seconds": 4.034771163000187,
+      "bytes": 5791973,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 0.9980244999987917,
-      "bytes": 5788373,
-      "pages": 22,
+      "seconds": 1.1227300080008717,
+      "bytes": 6451625,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 3.3476856620000035,
-      "bytes": 5904691,
-      "pages": 43,
+      "seconds": 1.4268143489989598,
+      "bytes": 2671684,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2678156700003456,
-      "bytes": 7637773,
+      "seconds": 1.154910495999502,
+      "bytes": 10565027,
+      "pages": 23,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3083442860006471,
+      "bytes": 8430845,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1486175610007194,
-      "bytes": 5966085,
+      "seconds": 1.0249396090002847,
+      "bytes": 5941897,
+      "pages": 36,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2488192199998593,
+      "bytes": 5819043,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0020495059998211,
-      "bytes": 6673858,
-      "pages": 32,
+      "seconds": 0.903354562999084,
+      "bytes": 2640360,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1948771370007307,
-      "bytes": 12132833,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.1839327219986444,
+      "bytes": 10186345,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.2424271460004093,
-      "bytes": 1081337,
-      "pages": 34,
-      "shards": 1
+      "seconds": 3.6534681639986957,
+      "bytes": 12129367,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.3408090630000515,
-      "bytes": 7658166,
-      "pages": 39,
+      "seconds": 0.9705111450002732,
+      "bytes": 2305053,
+      "pages": 28,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3458375579994026,
+      "bytes": 11053239,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.8867853440006002,
-      "bytes": 5785267,
-      "pages": 27,
+      "seconds": 1.1655933139991248,
+      "bytes": 2619678,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.081288327000948,
-      "bytes": 5787210,
-      "pages": 36,
+      "seconds": 1.0188674239998363,
+      "bytes": 6227206,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1759565290012688,
-      "bytes": 13132586,
-      "pages": 31,
-      "shards": 4
+      "seconds": 0.9700920809991658,
+      "bytes": 6347868,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.004212905998429,
-      "bytes": 6157452,
-      "pages": 38,
+      "seconds": 1.00576553800056,
+      "bytes": 8295904,
+      "pages": 33,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a7925c71cd513cd3d5f53037b00418d6c7a7f8c7b424e1f49846780b2108ebe
-size 721771711
+oid sha256:9933156547ab18b657069aa60eb52e3b669e25faa068a6484883abec26f0f1a7
+size 721780991