Qwen90M step 65000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 60000,
+  "step": 65000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "a86621d512304974e4d03bef262524ec95edf0e52d000b45faebb196bf0ad813",
+    "model.safetensors": "e33da1dad32ae2433103ec21715c2d542c98dc7e7caa2f82997e62b6fe8342d2",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "09ea04c8a1fd2002db31adafc765f00a01b0176621104639d3a70a8010a648f4",
+    "trainer.pt": "e3e3dd01665bea5cf593b73f4417e92f7e07fdb4e1949ec1cfe309333ab911ee",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f",
-    "shard_scheduler.json": "502ba4fb1878cbdcbe2c170d26ff10458bed187e056b37f6ad7fa2afaa31cc12",
-    "coverage.json": "558c9d8fbc6f9cdc4f9f8ba67a6152d0f15e7f96f74c310caabb313dfccd148e",
+    "shard_scheduler.json": "fb85ef21cda081906908c01fe4de71830aa788514ed6534050e7da541b10e3e2",
+    "coverage.json": "f8f337b02bacae59e68d5dbc5eb4b6339c704cf53f376b91e8950e701871a03e",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 60000,
+  "step": 65000,
   "stats": {
-    "climbmix/blocks": 2160000,
-    "climbmix/targets": 3609663868,
-    "climbmix/input_tokens": 3609892580,
-    "ultra_qa/blocks": 810000,
-    "ultra_qa/targets": 1321742315,
-    "ultra_qa/input_tokens": 1321744914,
-    "cortex/blocks": 810000,
-    "cortex/targets": 30814012,
-    "cortex/input_tokens": 1287629180,
-    "ultra_style/blocks": 810000,
-    "ultra_style/targets": 1281312518,
-    "ultra_style/input_tokens": 1281379443,
-    "openmath/blocks": 540000,
-    "openmath/targets": 834463745,
-    "openmath/input_tokens": 834742126,
-    "rewrite6/blocks": 270000,
-    "rewrite6/targets": 442578795,
-    "rewrite6/input_tokens": 443064258
+    "climbmix/blocks": 2340000,
+    "climbmix/targets": 3910377700,
+    "climbmix/input_tokens": 3910625311,
+    "ultra_qa/blocks": 877500,
+    "ultra_qa/targets": 1431685811,
+    "ultra_qa/input_tokens": 1431688622,
+    "cortex/blocks": 877500,
+    "cortex/targets": 33383362,
+    "cortex/input_tokens": 1394824403,
+    "ultra_style/blocks": 877500,
+    "ultra_style/targets": 1387878431,
+    "ultra_style/input_tokens": 1387950984,
+    "openmath/blocks": 585000,
+    "openmath/targets": 903936284,
+    "openmath/input_tokens": 904237672,
+    "rewrite6/blocks": 292500,
+    "rewrite6/targets": 479420381,
+    "rewrite6/input_tokens": 479946209
   },
   "sources": {
     "climbmix": {
-      "page": 295,
-      "row": 3347,
+      "page": 320,
+      "row": 1789,
       "start": {
-        "file": 14,
+        "file": 15,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 22,
+          "row_group": 39,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1211667
+      "documents": 1312509
     },
     "rewrite6": {
-      "page": 18,
-      "row": 1596,
+      "page": 32,
+      "row": 1861,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 73728
+          "row": 131072
         }
       },
       "epoch": 3,
-      "offset": 2048,
-      "documents": 691536
+      "offset": 0,
+      "documents": 749145
     },
     "openmath": {
-      "page": 184,
-      "row": 3842,
+      "page": 200,
+      "row": 1224,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 317,
-          "row": 14
+          "row_group": 382,
+          "row": 550
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 757506
+      "documents": 820424
     },
     "ultra_qa": {
-      "page": 79,
-      "row": 4028,
+      "page": 86,
+      "row": 2631,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 847
+          "row": 29519
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 327612
+      "documents": 354887
     },
     "ultra_style": {
-      "page": 121,
-      "row": 3563,
+      "page": 132,
+      "row": 470,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 5,
-          "row": 69812
+          "row_group": 6,
+          "row": 30335
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 499179
+      "documents": 541142
     },
     "cortex": {
-      "page": 15706,
-      "row": 87,
+      "page": 17014,
+      "row": 1,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 2010455
+      "documents": 2177793
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 60000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "a86621d512304974e4d03bef262524ec95edf0e52d000b45faebb196bf0ad813", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "09ea04c8a1fd2002db31adafc765f00a01b0176621104639d3a70a8010a648f4", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "502ba4fb1878cbdcbe2c170d26ff10458bed187e056b37f6ad7fa2afaa31cc12", "coverage.json": "558c9d8fbc6f9cdc4f9f8ba67a6152d0f15e7f96f74c310caabb313dfccd148e", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000060000"}
+{"step": 65000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e33da1dad32ae2433103ec21715c2d542c98dc7e7caa2f82997e62b6fe8342d2", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "e3e3dd01665bea5cf593b73f4417e92f7e07fdb4e1949ec1cfe309333ab911ee", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "fb85ef21cda081906908c01fe4de71830aa788514ed6534050e7da541b10e3e2", "coverage.json": "f8f337b02bacae59e68d5dbc5eb4b6339c704cf53f376b91e8950e701871a03e", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000065000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a86621d512304974e4d03bef262524ec95edf0e52d000b45faebb196bf0ad813
+oid sha256:e33da1dad32ae2433103ec21715c2d542c98dc7e7caa2f82997e62b6fe8342d2
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 1062967.6168812835,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 75551.01885780718,
-      "required_pages_per_second_upper_estimate": 0.006548570832288048
+      "required_token_positions_per_second": 75748.15161257412,
+      "required_pages_per_second_upper_estimate": 0.006565657799842846
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 1206672.263377472,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9443.877357225898,
-      "required_pages_per_second_upper_estimate": 0.0037673038763466962
+      "required_token_positions_per_second": 9468.518951571765,
+      "required_pages_per_second_upper_estimate": 0.0037771337767559298
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 410042.7581134458,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 18887.754714451796,
-      "required_pages_per_second_upper_estimate": 0.004369349640058767
+      "required_token_positions_per_second": 18937.03790314353,
+      "required_pages_per_second_upper_estimate": 0.0043807504384081025
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 2163486.5624068812,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28331.632071677694,
-      "required_pages_per_second_upper_estimate": 0.0017252477485795504
+      "required_token_positions_per_second": 28405.556854715294,
+      "required_pages_per_second_upper_estimate": 0.0017297493800131785
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 1546694.2737633225,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28331.632071677694,
-      "required_pages_per_second_upper_estimate": 0.002181047463854873
+      "required_token_positions_per_second": 28405.556854715294,
+      "required_pages_per_second_upper_estimate": 0.0021867383982900044
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 1015177.8395898147,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28331.632071677694,
-      "required_pages_per_second_upper_estimate": 0.3517840504572767
+      "required_token_positions_per_second": 28405.556854715294,
+      "required_pages_per_second_upper_estimate": 0.3527019488522703
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.09136518590602805,
-    "producer_p95_seconds": 0.29452696799853584,
-    "gpu_update_seconds": 0.9758703603820585,
-    "producer_batches_per_second": 10.945088001337089,
-    "gpu_batches_per_second": 1.0247262757406574,
-    "producer_headroom": 10.68098697227817,
+    "producer_batch_seconds": 0.11656077062494319,
+    "producer_p95_seconds": 0.29636224299974856,
+    "gpu_update_seconds": 0.9733306810850448,
+    "producer_batches_per_second": 8.579215757055119,
+    "gpu_batches_per_second": 1.0274000598493669,
+    "producer_headroom": 8.350413915990007,
     "sustainable": true,
-    "buffer_seconds": 60.50396234368763,
+    "buffer_seconds": 60.34650222727278,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 87,
+    "upload_pages_per_commit": 86,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 384384.7746639473,
-    "upload_bytes_per_second": 6418706.338644549,
+    "generation_bytes_per_second": 512873.7689111375,
+    "upload_bytes_per_second": 4674834.667993678,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 82.48002462328733,
+    "estimated_full_shards_per_hour": 110.05077171803086,
     "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": 1.1064888179989794,
-      "bytes": 5541578,
-      "pages": 33,
+      "seconds": 1.406286874000216,
+      "bytes": 18261180,
+      "pages": 44,
+      "shards": 5
+    },
+    {
+      "seconds": 1.0190579750014876,
+      "bytes": 5885831,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9643621880004503,
-      "bytes": 7410728,
-      "pages": 34,
+      "seconds": 0.8101317319997179,
+      "bytes": 713378,
+      "pages": 25,
+      "shards": 2
+    },
+    {
+      "seconds": 1.03739196099923,
+      "bytes": 7802740,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.8883736560001125,
-      "bytes": 6293157,
-      "pages": 36,
+      "seconds": 1.1620692619981128,
+      "bytes": 6079962,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.9868076469992957,
-      "bytes": 927962,
-      "pages": 26,
-      "shards": 1
+      "seconds": 0.8016836730021168,
+      "bytes": 4650466,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.2888746650005487,
-      "bytes": 18374728,
-      "pages": 40,
-      "shards": 5
+      "seconds": 1.2261960830001044,
+      "bytes": 13273679,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 1.1437757880012214,
-      "bytes": 6056576,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.7825120570014406,
+      "bytes": 1093241,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.102731920998849,
-      "bytes": 2598092,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.4853517330011528,
+      "bytes": 11092996,
+      "pages": 24,
+      "shards": 4
     },
     {
-      "seconds": 1.1366644360005012,
-      "bytes": 5837515,
-      "pages": 27,
+      "seconds": 0.8748619340003643,
+      "bytes": 6384371,
+      "pages": 42,
       "shards": 2
     },
     {
-      "seconds": 0.8290303609992407,
-      "bytes": 5336251,
-      "pages": 39,
-      "shards": 2
+      "seconds": 0.9224505880010838,
+      "bytes": 1117951,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.0252480340004695,
-      "bytes": 7371694,
-      "pages": 29,
+      "seconds": 1.0108905939996475,
+      "bytes": 7845615,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1864866850010003,
-      "bytes": 11535625,
+      "seconds": 1.27624071000173,
+      "bytes": 16075425,
       "pages": 35,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 0.9566465869993408,
-      "bytes": 1137762,
-      "pages": 35,
+      "seconds": 0.8457192399982887,
+      "bytes": 2628229,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0941519280004286,
-      "bytes": 11375746,
-      "pages": 34,
-      "shards": 4
+      "seconds": 0.9226963619985327,
+      "bytes": 5611919,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.7203741030007222,
-      "bytes": 1052913,
+      "seconds": 0.7924021649996575,
+      "bytes": 1122256,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.0976181589994667,
-      "bytes": 7695785,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.1671633029982331,
+      "bytes": 11363880,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.024126566000632,
-      "bytes": 5953181,
+      "seconds": 0.9065282010014926,
+      "bytes": 6177542,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.452807390000089,
-      "bytes": 11684781,
-      "pages": 40,
+      "seconds": 1.0971184539994283,
+      "bytes": 8482200,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.8971567329990648,
-      "bytes": 7415356,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.1112909670009685,
+      "bytes": 6011206,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 0.9046226490008848,
-      "bytes": 1040439,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.9119941530007054,
+      "bytes": 5747516,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.043625083999359,
-      "bytes": 6212867,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.016742787000112,
+      "bytes": 7467282,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.3019511900001817,
-      "bytes": 12069002,
-      "pages": 36,
-      "shards": 4
+      "seconds": 0.8858778819994768,
+      "bytes": 5985432,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.8234439619991463,
-      "bytes": 1074472,
-      "pages": 32,
+      "seconds": 0.7830291239988583,
+      "bytes": 1207474,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.0881272910010011,
-      "bytes": 13091178,
-      "pages": 33,
+      "seconds": 1.2103096939972602,
+      "bytes": 12749523,
+      "pages": 22,
       "shards": 4
     },
     {
-      "seconds": 1.0808892179993563,
-      "bytes": 6450126,
+      "seconds": 0.8322358400000667,
+      "bytes": 4694947,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 0.8593272979996982,
+      "bytes": 6225685,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8513057910004136,
-      "bytes": 341762,
-      "pages": 9,
-      "shards": 1
+      "seconds": 0.9578338470018934,
+      "bytes": 2306768,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.0242165819981892,
-      "bytes": 5354140,
+      "seconds": 0.8912073430001328,
+      "bytes": 6007286,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 0.845649806999063,
-      "bytes": 8068507,
-      "pages": 38,
+      "seconds": 1.088943154001754,
+      "bytes": 10590685,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.9696599159997277,
-      "bytes": 6181281,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.9887308050019783,
+      "bytes": 8510426,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.0534396790008032,
-      "bytes": 2069289,
-      "pages": 21,
+      "seconds": 1.0075569920009002,
+      "bytes": 5909692,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.112150704000669,
-      "bytes": 16576455,
-      "pages": 44,
-      "shards": 4
+      "seconds": 0.7540812590013957,
+      "bytes": 2508470,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 0.7744886969994695,
-      "bytes": 1014054,
+      "seconds": 1.3975003769992327,
+      "bytes": 10375015,
       "pages": 34,
-      "shards": 1
+      "shards": 4
     },
     {
-      "seconds": 0.9544328729989502,
-      "bytes": 7444287,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.0634488309988228,
+      "bytes": 6386527,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0936733489998005,
-      "bytes": 876493,
-      "pages": 25,
-      "shards": 1
+      "seconds": 14.660640815000079,
+      "bytes": 2712693,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.364482594999572,
-      "bytes": 12283477,
-      "pages": 38,
+      "seconds": 3.9525413480005227,
+      "bytes": 15333188,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.1743630839991965,
-      "bytes": 5760431,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9926189179968787,
-      "bytes": 7099055,
-      "pages": 43,
-      "shards": 2
+      "seconds": 2.024789704999421,
+      "bytes": 1170220,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.0609590189997107,
-      "bytes": 7221766,
-      "pages": 30,
+      "seconds": 1.2876270810011192,
+      "bytes": 7494644,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9498501529997156,
-      "bytes": 6032509,
-      "pages": 30,
+      "seconds": 0.8472818170012033,
+      "bytes": 6269698,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.024946067998826,
-      "bytes": 5141639,
-      "pages": 39,
-      "shards": 2
+      "seconds": 0.9710190369987686,
+      "bytes": 6743419,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.2369695059969672,
-      "bytes": 7463645,
+      "seconds": 1.0164730639990012,
+      "bytes": 6038957,
       "pages": 35,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 0.9742800499989244,
-      "bytes": 772407,
-      "pages": 25,
+      "seconds": 0.9485057740021148,
+      "bytes": 880447,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.406286874000216,
-      "bytes": 18261180,
-      "pages": 44,
-      "shards": 5
-    },
-    {
-      "seconds": 1.0190579750014876,
-      "bytes": 5885831,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8101317319997179,
-      "bytes": 713378,
-      "pages": 25,
-      "shards": 2
+      "seconds": 1.0207911869983946,
+      "bytes": 13578331,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.03739196099923,
-      "bytes": 7802740,
-      "pages": 38,
+      "seconds": 1.2704945900004532,
+      "bytes": 9798343,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1620692619981128,
-      "bytes": 6079962,
-      "pages": 34,
+      "seconds": 0.7868170459987596,
+      "bytes": 2512813,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8016836730021168,
-      "bytes": 4650466,
-      "pages": 32,
+      "seconds": 1.0122808699998131,
+      "bytes": 6141256,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2261960830001044,
-      "bytes": 13273679,
-      "pages": 37,
-      "shards": 4
-    },
-    {
-      "seconds": 0.7825120570014406,
-      "bytes": 1093241,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.8550289739978325,
+      "bytes": 5835711,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.4853517330011528,
-      "bytes": 11092996,
-      "pages": 24,
-      "shards": 4
+      "seconds": 1.051623090999783,
+      "bytes": 7071478,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 0.8748619340003643,
-      "bytes": 6384371,
-      "pages": 42,
-      "shards": 2
+      "seconds": 1.1867464049973933,
+      "bytes": 12128568,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9224505880010838,
-      "bytes": 1117951,
+      "seconds": 0.8681122560010408,
+      "bytes": 1227302,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.0108905939996475,
-      "bytes": 7845615,
-      "pages": 36,
+      "seconds": 1.205836907000048,
+      "bytes": 7335325,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.27624071000173,
-      "bytes": 16075425,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.085818534000282,
+      "bytes": 10112362,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 0.8457192399982887,
-      "bytes": 2628229,
+      "seconds": 0.927039637001144,
+      "bytes": 2682847,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9226963619985327,
-      "bytes": 5611919,
-      "pages": 27,
+      "seconds": 0.8612236650005798,
+      "bytes": 6230072,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.7924021649996575,
-      "bytes": 1122256,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0601522930010105,
+      "bytes": 12072218,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.1671633029982331,
-      "bytes": 11363880,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.126622554998903,
+      "bytes": 6643361,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.9065282010014926,
-      "bytes": 6177542,
-      "pages": 28,
+      "seconds": 0.8856843470020976,
+      "bytes": 5828359,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.0971184539994283,
-      "bytes": 8482200,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0851010650003445,
+      "bytes": 1256653,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.1112909670009685,
-      "bytes": 6011206,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.3800626919983188,
+      "bytes": 7827681,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.9119941530007054,
-      "bytes": 5747516,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.5110167620005086,
+      "bytes": 10255842,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.016742787000112,
-      "bytes": 7467282,
+      "seconds": 8.627843940001185,
+      "bytes": 8388512,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.8858778819994768,
-      "bytes": 5985432,
-      "pages": 27,
+      "seconds": 1.2330220080002618,
+      "bytes": 6137294,
+      "pages": 36,
       "shards": 2
+    },
+    {
+      "seconds": 1.6848288789988146,
+      "bytes": 823777,
+      "pages": 25,
+      "shards": 1
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:09ea04c8a1fd2002db31adafc765f00a01b0176621104639d3a70a8010a648f4
-size 721780607
+oid sha256:e3e3dd01665bea5cf593b73f4417e92f7e07fdb4e1949ec1cfe309333ab911ee
+size 721787647