appvoid commited on
Commit
1295dc9
·
verified ·
1 Parent(s): 905147d

Qwen90M step 81921; complete resumable checkpoint

Browse files
Files changed (7) hide show
  1. complete.json +6 -6
  2. coverage.json +44 -44
  3. latest.json +1 -1
  4. model.safetensors +1 -1
  5. shard_scheduler.json +127 -127
  6. trainer.pt +2 -2
  7. training_config.json +2 -2
complete.json CHANGED
@@ -1,18 +1,18 @@
1
  {
2
- "step": 80000,
3
  "files": {
4
  "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
5
  "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
6
- "model.safetensors": "016fae2e58ad750ac0b4abde700bbdbea9199273ac137f7e182d0fe7874da0dc",
7
  "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
8
  "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
9
  "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
10
  "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
11
- "trainer.pt": "63c9c8c99731fc20714a54a44377517fe0f1f3043beea6e7f6b8a0c370badb1e",
12
- "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
13
  "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f",
14
- "shard_scheduler.json": "ea08386ca957ff57a36e47f5f3cdc24c23475683922434b0de7b9c8abf2744b1",
15
- "coverage.json": "42736094b47b18e0027534773acb49639d163b78174444895fdb7d300580166a",
16
  "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
17
  },
18
  "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"
 
1
  {
2
+ "step": 81921,
3
  "files": {
4
  "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
5
  "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
6
+ "model.safetensors": "d9bd504db1f2f3b85533bffea41681c8a3dcb48377b09ac460343140c40f9cb4",
7
  "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
8
  "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
9
  "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
10
  "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
11
+ "trainer.pt": "869335f7fb0864cabf919da1e915a66ed080d75a072a00e376ba282d840dc755",
12
+ "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf",
13
  "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f",
14
+ "shard_scheduler.json": "c8e2fa7804bf590b627c334969e9b37bcabea7c8cf567c273a2065d57ec93b49",
15
+ "coverage.json": "99bf4554196cda94eb5f816b7c0fb3e79e4c86b59ed55da03a4438f66fb4eec3",
16
  "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
17
  },
18
  "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"
coverage.json CHANGED
@@ -1,107 +1,107 @@
1
  {
2
- "step": 80000,
3
  "stats": {
4
- "climbmix/blocks": 2880000,
5
- "climbmix/targets": 4812534066,
6
- "climbmix/input_tokens": 4812838946,
7
- "ultra_qa/blocks": 1080000,
8
- "ultra_qa/targets": 1761872361,
9
- "ultra_qa/input_tokens": 1761875805,
10
- "cortex/blocks": 1080000,
11
- "cortex/targets": 41089673,
12
- "cortex/input_tokens": 1716618096,
13
- "ultra_style/blocks": 1080000,
14
- "ultra_style/targets": 1707744164,
15
- "ultra_style/input_tokens": 1707833604,
16
- "openmath/blocks": 720000,
17
- "openmath/targets": 1112712435,
18
- "openmath/input_tokens": 1113083661,
19
- "rewrite6/blocks": 360000,
20
- "rewrite6/targets": 590150425,
21
- "rewrite6/input_tokens": 590797964
22
  },
23
  "sources": {
24
  "climbmix": {
25
- "page": 394,
26
- "row": 2302,
27
  "start": {
28
  "file": 19,
29
  "reader": {
30
  "format": "parquet-row-v1",
31
- "row_group": 3,
32
  "row": 1024
33
  }
34
  },
35
  "epoch": 0,
36
  "offset": 2048,
37
- "documents": 1616126
38
  },
39
  "rewrite6": {
40
- "page": 24,
41
- "row": 2442,
42
  "start": {
43
  "file": 0,
44
  "reader": {
45
  "format": "jsonl-row-v1",
46
- "row": 98304
47
  }
48
  },
49
  "epoch": 4,
50
- "offset": 2048,
51
- "documents": 922362
52
  },
53
  "openmath": {
54
- "page": 246,
55
- "row": 2574,
56
  "start": {
57
  "file": 2,
58
  "reader": {
59
  "format": "parquet-row-v1",
60
- "row_group": 134,
61
- "row": 316
62
  }
63
  },
64
  "epoch": 0,
65
  "offset": 0,
66
- "documents": 1010190
67
  },
68
  "ultra_qa": {
69
- "page": 106,
70
- "row": 2578,
71
  "start": {
72
  "file": 0,
73
  "reader": {
74
  "format": "parquet-row-v1",
75
  "row_group": 6,
76
- "row": 46661
77
  }
78
  },
79
  "epoch": 0,
80
  "offset": 4096,
81
- "documents": 436754
82
  },
83
  "ultra_style": {
84
- "page": 162,
85
- "row": 2876,
86
  "start": {
87
  "file": 0,
88
  "reader": {
89
  "format": "parquet-row-v1",
90
  "row_group": 7,
91
- "row": 68428
92
  }
93
  },
94
  "epoch": 0,
95
  "offset": 2048,
96
- "documents": 666428
97
  },
98
  "cortex": {
99
- "page": 20940,
100
- "row": 66,
101
  "start": null,
102
  "epoch": 0,
103
  "offset": 0,
104
- "documents": 2680386
105
  }
106
  }
107
  }
 
1
  {
2
+ "step": 81921,
3
  "stats": {
4
+ "climbmix/blocks": 2949156,
5
+ "climbmix/targets": 4928102633,
6
+ "climbmix/input_tokens": 4928414803,
7
+ "ultra_qa/blocks": 1105933,
8
+ "ultra_qa/targets": 1804133074,
9
+ "ultra_qa/input_tokens": 1804136584,
10
+ "cortex/blocks": 1105934,
11
+ "cortex/targets": 42079455,
12
+ "cortex/input_tokens": 1757883587,
13
+ "ultra_style/blocks": 1105933,
14
+ "ultra_style/targets": 1748747349,
15
+ "ultra_style/input_tokens": 1748838948,
16
+ "openmath/blocks": 737289,
17
+ "openmath/targets": 1139400794,
18
+ "openmath/input_tokens": 1139780972,
19
+ "rewrite6/blocks": 368645,
20
+ "rewrite6/targets": 604257798,
21
+ "rewrite6/input_tokens": 604920589
22
  },
23
  "sources": {
24
  "climbmix": {
25
+ "page": 404,
26
+ "row": 101,
27
  "start": {
28
  "file": 19,
29
  "reader": {
30
  "format": "parquet-row-v1",
31
+ "row_group": 43,
32
  "row": 1024
33
  }
34
  },
35
  "epoch": 0,
36
  "offset": 2048,
37
+ "documents": 1654885
38
  },
39
  "rewrite6": {
40
+ "page": 29,
41
+ "row": 3813,
42
  "start": {
43
  "file": 0,
44
  "reader": {
45
  "format": "jsonl-row-v1",
46
+ "row": 118784
47
  }
48
  },
49
  "epoch": 4,
50
+ "offset": 0,
51
+ "documents": 944213
52
  },
53
  "openmath": {
54
+ "page": 252,
55
+ "row": 2358,
56
  "start": {
57
  "file": 2,
58
  "reader": {
59
  "format": "parquet-row-v1",
60
+ "row_group": 158,
61
+ "row": 892
62
  }
63
  },
64
  "epoch": 0,
65
  "offset": 0,
66
+ "documents": 1034550
67
  },
68
  "ultra_qa": {
69
+ "page": 109,
70
+ "row": 758,
71
  "start": {
72
  "file": 0,
73
  "reader": {
74
  "format": "parquet-row-v1",
75
  "row_group": 6,
76
+ "row": 58949
77
  }
78
  },
79
  "epoch": 0,
80
  "offset": 4096,
81
+ "documents": 447222
82
  },
83
  "ultra_style": {
84
+ "page": 166,
85
+ "row": 2532,
86
  "start": {
87
  "file": 0,
88
  "reader": {
89
  "format": "parquet-row-v1",
90
  "row_group": 7,
91
+ "row": 84812
92
  }
93
  },
94
  "epoch": 0,
95
  "offset": 2048,
96
+ "documents": 682468
97
  },
98
  "cortex": {
99
+ "page": 21444,
100
+ "row": 62,
101
  "start": null,
102
  "epoch": 0,
103
  "offset": 0,
104
+ "documents": 2744894
105
  }
106
  }
107
  }
latest.json CHANGED
@@ -1 +1 @@
1
- {"step": 80000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "016fae2e58ad750ac0b4abde700bbdbea9199273ac137f7e182d0fe7874da0dc", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "63c9c8c99731fc20714a54a44377517fe0f1f3043beea6e7f6b8a0c370badb1e", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "ea08386ca957ff57a36e47f5f3cdc24c23475683922434b0de7b9c8abf2744b1", "coverage.json": "42736094b47b18e0027534773acb49639d163b78174444895fdb7d300580166a", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000080000"}
 
1
+ {"step": 81921, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "d9bd504db1f2f3b85533bffea41681c8a3dcb48377b09ac460343140c40f9cb4", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "869335f7fb0864cabf919da1e915a66ed080d75a072a00e376ba282d840dc755", "training_config.json": "2122ad6f8105987f9b87dd55d13b632084853908566b081c829ffe801259e2bf", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "c8e2fa7804bf590b627c334969e9b37bcabea7c8cf567c273a2065d57ec93b49", "coverage.json": "99bf4554196cda94eb5f816b7c0fb3e79e4c86b59ed55da03a4438f66fb4eec3", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000081921"}
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:016fae2e58ad750ac0b4abde700bbdbea9199273ac137f7e182d0fe7874da0dc
3
  size 360622744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9bd504db1f2f3b85533bffea41681c8a3dcb48377b09ac460343140c40f9cb4
3
  size 360622744
shard_scheduler.json CHANGED
@@ -22,8 +22,8 @@
22
  "tokens_per_second": 1062967.6168812835,
23
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
24
  "estimated_pages_for_buffer": 1,
25
- "required_token_positions_per_second": 75133.13345379168,
26
- "required_pages_per_second_upper_estimate": 0.006512349584588858
27
  },
28
  {
29
  "source": "rewrite6",
@@ -36,8 +36,8 @@
36
  "tokens_per_second": 1206672.263377472,
37
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
38
  "estimated_pages_for_buffer": 1,
39
- "required_token_positions_per_second": 9391.64168172396,
40
- "required_pages_per_second_upper_estimate": 0.003746466284396027
41
  },
42
  {
43
  "source": "openmath",
@@ -50,8 +50,8 @@
50
  "tokens_per_second": 410042.7581134458,
51
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
52
  "estimated_pages_for_buffer": 1,
53
- "required_token_positions_per_second": 18783.28336344792,
54
- "required_pages_per_second_upper_estimate": 0.004345182031637001
55
  },
56
  {
57
  "source": "ultra_qa",
@@ -64,8 +64,8 @@
64
  "tokens_per_second": 2163486.5624068812,
65
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
66
  "estimated_pages_for_buffer": 1,
67
- "required_token_positions_per_second": 28174.925045171884,
68
- "required_pages_per_second_upper_estimate": 0.0017157051128435733
69
  },
70
  {
71
  "source": "ultra_style",
@@ -78,8 +78,8 @@
78
  "tokens_per_second": 1546694.2737633225,
79
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
80
  "estimated_pages_for_buffer": 1,
81
- "required_token_positions_per_second": 28174.925045171884,
82
- "required_pages_per_second_upper_estimate": 0.0021689837231616427
83
  },
84
  {
85
  "source": "cortex",
@@ -91,9 +91,9 @@
91
  "read_seconds": 0.042780299000014566,
92
  "tokens_per_second": 1015177.8395898147,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
- "estimated_pages_for_buffer": 22,
95
- "required_token_positions_per_second": 28174.925045171884,
96
- "required_pages_per_second_upper_estimate": 0.34983827365275444
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,132 +155,30 @@
155
  "shard_benchmark_scope": "Local packing/read measurements; upload capacity measured only on real data commits. Targets larger than available pages are sample-limited.",
156
  "estimated_fresh_generation_batch_seconds": 0.1434574083060457,
157
  "selected": {
158
- "prefetch_batches": 62,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 688946,
161
- "producer_batch_seconds": 0.12603624823447035,
162
- "producer_p95_seconds": 0.43727277699508704,
163
- "gpu_update_seconds": 0.9812980852894166,
164
- "producer_batches_per_second": 7.934225383634551,
165
- "gpu_batches_per_second": 1.0190583422009507,
166
- "producer_headroom": 7.785840177215272,
167
  "sustainable": true,
168
- "buffer_seconds": 60.84048128794383,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
- "upload_pages_per_commit": 83,
172
  "max_pending_pages": 4096,
173
- "generation_bytes_per_second": 287303.50039527926,
174
- "upload_bytes_per_second": 6345416.31829655,
175
  "upload_sustainable": true,
176
  "logical_page_rows": 4096,
177
  "logical_cortex_page_rows": 128,
178
- "estimated_full_shards_per_hour": 61.64864310163291,
179
  "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
180
  },
181
  "upload_measurements": [
182
- {
183
- "seconds": 0.9040151730005164,
184
- "bytes": 8396980,
185
- "pages": 35,
186
- "shards": 3
187
- },
188
- {
189
- "seconds": 1.1415953719988465,
190
- "bytes": 5808415,
191
- "pages": 28,
192
- "shards": 2
193
- },
194
- {
195
- "seconds": 1.0399577659991337,
196
- "bytes": 7078524,
197
- "pages": 40,
198
- "shards": 3
199
- },
200
- {
201
- "seconds": 0.9194039719986904,
202
- "bytes": 5850558,
203
- "pages": 28,
204
- "shards": 2
205
- },
206
- {
207
- "seconds": 0.8959461580016068,
208
- "bytes": 6085583,
209
- "pages": 36,
210
- "shards": 3
211
- },
212
- {
213
- "seconds": 0.9613766149996081,
214
- "bytes": 2539274,
215
- "pages": 33,
216
- "shards": 2
217
- },
218
- {
219
- "seconds": 1.3802653160018963,
220
- "bytes": 12066636,
221
- "pages": 34,
222
- "shards": 3
223
- },
224
- {
225
- "seconds": 0.828995023999596,
226
- "bytes": 4743589,
227
- "pages": 35,
228
- "shards": 2
229
- },
230
- {
231
- "seconds": 0.9685989429999609,
232
- "bytes": 7522789,
233
- "pages": 29,
234
- "shards": 3
235
- },
236
- {
237
- "seconds": 0.7894636139972135,
238
- "bytes": 5936931,
239
- "pages": 37,
240
- "shards": 2
241
- },
242
- {
243
- "seconds": 1.0263739810034167,
244
- "bytes": 6580240,
245
- "pages": 34,
246
- "shards": 3
247
- },
248
- {
249
- "seconds": 1.228623751005216,
250
- "bytes": 6179585,
251
- "pages": 37,
252
- "shards": 2
253
- },
254
- {
255
- "seconds": 1.3004778859976796,
256
- "bytes": 12330619,
257
- "pages": 35,
258
- "shards": 3
259
- },
260
- {
261
- "seconds": 1.576678247001837,
262
- "bytes": 2288077,
263
- "pages": 28,
264
- "shards": 2
265
- },
266
- {
267
- "seconds": 1.2293803360007587,
268
- "bytes": 10631332,
269
- "pages": 34,
270
- "shards": 3
271
- },
272
- {
273
- "seconds": 0.8141304599994328,
274
- "bytes": 2723440,
275
- "pages": 38,
276
- "shards": 2
277
- },
278
- {
279
- "seconds": 0.9625181489973329,
280
- "bytes": 5689945,
281
- "pages": 27,
282
- "shards": 2
283
- },
284
  {
285
  "seconds": 0.9489649809984257,
286
  "bytes": 5862208,
@@ -562,6 +460,108 @@
562
  "bytes": 6142895,
563
  "pages": 38,
564
  "shards": 3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
565
  }
566
  ],
567
  "cache_key": {
 
22
  "tokens_per_second": 1062967.6168812835,
23
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
24
  "estimated_pages_for_buffer": 1,
25
+ "required_token_positions_per_second": 74488.08882560692,
26
+ "required_pages_per_second_upper_estimate": 0.006456438751068466
27
  },
28
  {
29
  "source": "rewrite6",
 
36
  "tokens_per_second": 1206672.263377472,
37
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
38
  "estimated_pages_for_buffer": 1,
39
+ "required_token_positions_per_second": 9311.011103200864,
40
+ "required_pages_per_second_upper_estimate": 0.0037143015410885847
41
  },
42
  {
43
  "source": "openmath",
 
50
  "tokens_per_second": 410042.7581134458,
51
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
52
  "estimated_pages_for_buffer": 1,
53
+ "required_token_positions_per_second": 18622.02220640173,
54
+ "required_pages_per_second_upper_estimate": 0.004307877101053795
55
  },
56
  {
57
  "source": "ultra_qa",
 
64
  "tokens_per_second": 2163486.5624068812,
65
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
66
  "estimated_pages_for_buffer": 1,
67
+ "required_token_positions_per_second": 27933.03330960259,
68
+ "required_pages_per_second_upper_estimate": 0.001700975175255258
69
  },
70
  {
71
  "source": "ultra_style",
 
78
  "tokens_per_second": 1546694.2737633225,
79
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
80
  "estimated_pages_for_buffer": 1,
81
+ "required_token_positions_per_second": 27933.03330960259,
82
+ "required_pages_per_second_upper_estimate": 0.002150362227758338
83
  },
84
  {
85
  "source": "cortex",
 
91
  "read_seconds": 0.042780299000014566,
92
  "tokens_per_second": 1015177.8395898147,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
+ "estimated_pages_for_buffer": 21,
95
+ "required_token_positions_per_second": 27933.03330960259,
96
+ "required_pages_per_second_upper_estimate": 0.3468347878565453
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
 
155
  "shard_benchmark_scope": "Local packing/read measurements; upload capacity measured only on real data commits. Targets larger than available pages are sample-limited.",
156
  "estimated_fresh_generation_batch_seconds": 0.1434574083060457,
157
  "selected": {
158
+ "prefetch_batches": 61,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 688946,
161
+ "producer_batch_seconds": 0.144511708648281,
162
+ "producer_p95_seconds": 0.5217355390050216,
163
+ "gpu_update_seconds": 0.9897958339703621,
164
+ "producer_batches_per_second": 6.919854518043547,
165
+ "gpu_batches_per_second": 1.0103093644966215,
166
+ "producer_headroom": 6.849243173640491,
167
  "sustainable": true,
168
+ "buffer_seconds": 60.37754587219209,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
+ "upload_pages_per_commit": 78,
172
  "max_pending_pages": 4096,
173
+ "generation_bytes_per_second": 246534.60347499957,
174
+ "upload_bytes_per_second": 6254719.971833577,
175
  "upload_sustainable": true,
176
  "logical_page_rows": 4096,
177
  "logical_cortex_page_rows": 128,
178
+ "estimated_full_shards_per_hour": 52.90058687388888,
179
  "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
180
  },
181
  "upload_measurements": [
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
182
  {
183
  "seconds": 0.9489649809984257,
184
  "bytes": 5862208,
 
460
  "bytes": 6142895,
461
  "pages": 38,
462
  "shards": 3
463
+ },
464
+ {
465
+ "seconds": 1.3159294670040254,
466
+ "bytes": 6394810,
467
+ "pages": 28,
468
+ "shards": 2
469
+ },
470
+ {
471
+ "seconds": 0.8037503040031879,
472
+ "bytes": 917701,
473
+ "pages": 27,
474
+ "shards": 1
475
+ },
476
+ {
477
+ "seconds": 1.0914277900010347,
478
+ "bytes": 7011453,
479
+ "pages": 24,
480
+ "shards": 3
481
+ },
482
+ {
483
+ "seconds": 0.9372564279983635,
484
+ "bytes": 5878408,
485
+ "pages": 36,
486
+ "shards": 3
487
+ },
488
+ {
489
+ "seconds": 1.3057175280046067,
490
+ "bytes": 11562438,
491
+ "pages": 32,
492
+ "shards": 3
493
+ },
494
+ {
495
+ "seconds": 1.3249238560019876,
496
+ "bytes": 7783504,
497
+ "pages": 39,
498
+ "shards": 3
499
+ },
500
+ {
501
+ "seconds": 0.6935778650004067,
502
+ "bytes": 919630,
503
+ "pages": 27,
504
+ "shards": 1
505
+ },
506
+ {
507
+ "seconds": 1.1119171959944651,
508
+ "bytes": 9959227,
509
+ "pages": 34,
510
+ "shards": 3
511
+ },
512
+ {
513
+ "seconds": 1.0288992829955532,
514
+ "bytes": 2624793,
515
+ "pages": 37,
516
+ "shards": 2
517
+ },
518
+ {
519
+ "seconds": 1.2585268050024752,
520
+ "bytes": 5926427,
521
+ "pages": 33,
522
+ "shards": 2
523
+ },
524
+ {
525
+ "seconds": 1.132453666999936,
526
+ "bytes": 13402287,
527
+ "pages": 32,
528
+ "shards": 4
529
+ },
530
+ {
531
+ "seconds": 1.2154031860045507,
532
+ "bytes": 5549196,
533
+ "pages": 34,
534
+ "shards": 2
535
+ },
536
+ {
537
+ "seconds": 0.9452024689962855,
538
+ "bytes": 5836559,
539
+ "pages": 31,
540
+ "shards": 2
541
+ },
542
+ {
543
+ "seconds": 1.0610048369999276,
544
+ "bytes": 2712570,
545
+ "pages": 38,
546
+ "shards": 2
547
+ },
548
+ {
549
+ "seconds": 0.991505464000511,
550
+ "bytes": 5901383,
551
+ "pages": 29,
552
+ "shards": 2
553
+ },
554
+ {
555
+ "seconds": 1.330521811003564,
556
+ "bytes": 11596177,
557
+ "pages": 39,
558
+ "shards": 4
559
+ },
560
+ {
561
+ "seconds": 1.1032275320030749,
562
+ "bytes": 6673144,
563
+ "pages": 28,
564
+ "shards": 2
565
  }
566
  ],
567
  "cache_key": {
trainer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:63c9c8c99731fc20714a54a44377517fe0f1f3043beea6e7f6b8a0c370badb1e
3
- size 721775039
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:869335f7fb0864cabf919da1e915a66ed080d75a072a00e376ba282d840dc755
3
+ size 721766143
training_config.json CHANGED
@@ -46,8 +46,8 @@
46
  "cpu_threads": 4,
47
  "page_rows": 4096,
48
  "cortex_page_rows": 128,
49
- "prefetch_batches": 62,
50
- "prefetch_low_water": 31,
51
  "memory_pages": 12,
52
  "cache_gb": 20,
53
  "force_rebenchmark": false,
 
46
  "cpu_threads": 4,
47
  "page_rows": 4096,
48
  "cortex_page_rows": 128,
49
+ "prefetch_batches": 61,
50
+ "prefetch_low_water": 30,
51
  "memory_pages": 12,
52
  "cache_gb": 20,
53
  "force_rebenchmark": false,