appvoid commited on
Commit
d15cbd8
·
verified ·
1 Parent(s): cef6011

Qwen90M step 842507; complete resumable checkpoint

Browse files
Files changed (6) hide show
  1. complete.json +5 -5
  2. coverage.json +47 -47
  3. latest.json +1 -1
  4. model.safetensors +1 -1
  5. shard_scheduler.json +155 -155
  6. trainer.pt +2 -2
complete.json CHANGED
@@ -1,18 +1,18 @@
1
  {
2
- "step": 840000,
3
  "files": {
4
  "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
5
  "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
6
- "model.safetensors": "3e144501faba5106e3f43a2e36d221a8630b9cdfb6837a8f976f58b534e86cfa",
7
  "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
8
  "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
9
  "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
10
  "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
11
- "trainer.pt": "c1114d95865c877e56d4ca2db61e9232398961a900ec723d9fe48b310b10c038",
12
  "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
13
  "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
14
- "shard_scheduler.json": "5d089facc2fe4bc290dfa5a25481443cf9a74f9c2860a1b57795e6bdc0906c9e",
15
- "coverage.json": "abfd3dc053aa9107e89eeb9bfe2735dc1cd78f0a1640546271b969e4322c5588",
16
  "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
17
  },
18
  "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"
 
1
  {
2
+ "step": 842507,
3
  "files": {
4
  "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
5
  "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
6
+ "model.safetensors": "6f8382631e83ae42ed9759e16b99adad61f343fa6de9cb2d60b9dd6c9148bc70",
7
  "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
8
  "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
9
  "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
10
  "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
11
+ "trainer.pt": "a2b248de903e43cc04b64f31262641a2a596e6720f6e8b83622de3cb8a1ad33a",
12
  "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
13
  "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
14
+ "shard_scheduler.json": "76298eb667981f18a15e74c27926fbcd7a9ae27df4f2ad6d655bc8499e520b76",
15
+ "coverage.json": "b6f294dc008ba78304b1fa2017cef88912e1ceaaf314ebcebdb825482582a1b3",
16
  "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
17
  },
18
  "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"
coverage.json CHANGED
@@ -1,107 +1,107 @@
1
  {
2
- "step": 840000,
3
  "stats": {
4
- "climbmix/blocks": 30240000,
5
- "climbmix/targets": 50551290330,
6
- "climbmix/input_tokens": 50554493320,
7
- "ultra_qa/blocks": 11340000,
8
- "ultra_qa/targets": 18500378600,
9
- "ultra_qa/input_tokens": 18500414184,
10
- "cortex/blocks": 11340000,
11
- "cortex/targets": 431213356,
12
- "cortex/input_tokens": 18025811942,
13
- "ultra_style/blocks": 11340000,
14
- "ultra_style/targets": 17936310265,
15
- "ultra_style/input_tokens": 17937251230,
16
- "openmath/blocks": 7560000,
17
- "openmath/targets": 11681504015,
18
- "openmath/input_tokens": 11685396540,
19
- "rewrite6/blocks": 3780000,
20
- "rewrite6/targets": 6196813702,
21
- "rewrite6/input_tokens": 6203607570
22
  },
23
  "sources": {
24
  "climbmix": {
25
- "page": 4141,
26
- "row": 1870,
27
  "start": {
28
- "file": 200,
29
  "reader": {
30
  "format": "parquet-row-v1",
31
- "row_group": 37,
32
  "row": 1024
33
  }
34
  },
35
  "epoch": 0,
36
  "offset": 2048,
37
- "documents": 16963406
38
  },
39
  "rewrite6": {
40
- "page": 6,
41
- "row": 281,
42
  "start": {
43
  "file": 0,
44
  "reader": {
45
  "format": "jsonl-row-v1",
46
- "row": 24576
47
  }
48
  },
49
  "epoch": 47,
50
- "offset": 0,
51
- "documents": 9678845
52
  },
53
  "openmath": {
54
- "page": 2588,
55
- "row": 1820,
56
  "start": {
57
  "file": 24,
58
  "reader": {
59
  "format": "parquet-row-v1",
60
- "row_group": 120,
61
- "row": 849
62
  }
63
  },
64
  "epoch": 0,
65
  "offset": 0,
66
- "documents": 10602268
67
  },
68
  "ultra_qa": {
69
- "page": 1119,
70
- "row": 459,
71
  "start": {
72
  "file": 8,
73
  "reader": {
74
  "format": "parquet-row-v1",
75
  "row_group": 6,
76
- "row": 39179
77
  }
78
  },
79
  "epoch": 0,
80
- "offset": 0,
81
- "documents": 4583883
82
  },
83
  "ultra_style": {
84
- "page": 1708,
85
- "row": 803,
86
  "start": {
87
  "file": 10,
88
  "reader": {
89
  "format": "parquet-row-v1",
90
  "row_group": 1,
91
- "row": 61224
92
  }
93
  },
94
  "epoch": 0,
95
- "offset": 2048,
96
- "documents": 6996771
97
  },
98
  "cortex": {
99
- "page": 219925,
100
- "row": 120,
101
  "start": null,
102
  "epoch": 0,
103
  "offset": 0,
104
- "documents": 28150520
105
  }
106
  }
107
  }
 
1
  {
2
+ "step": 842507,
3
  "stats": {
4
+ "climbmix/blocks": 30330252,
5
+ "climbmix/targets": 50701917572,
6
+ "climbmix/input_tokens": 50705130186,
7
+ "ultra_qa/blocks": 11373844,
8
+ "ultra_qa/targets": 18555701427,
9
+ "ultra_qa/input_tokens": 18555737118,
10
+ "cortex/blocks": 11373845,
11
+ "cortex/targets": 432503890,
12
+ "cortex/input_tokens": 18079643538,
13
+ "ultra_style/blocks": 11373844,
14
+ "ultra_style/targets": 17989747078,
15
+ "ultra_style/input_tokens": 17990690711,
16
+ "openmath/blocks": 7582563,
17
+ "openmath/targets": 11716447202,
18
+ "openmath/input_tokens": 11720351418,
19
+ "rewrite6/blocks": 3791282,
20
+ "rewrite6/targets": 6215294727,
21
+ "rewrite6/input_tokens": 6222108705
22
  },
23
  "sources": {
24
  "climbmix": {
25
+ "page": 4153,
26
+ "row": 3572,
27
  "start": {
28
+ "file": 201,
29
  "reader": {
30
  "format": "parquet-row-v1",
31
+ "row_group": 2,
32
  "row": 1024
33
  }
34
  },
35
  "epoch": 0,
36
  "offset": 2048,
37
+ "documents": 17014260
38
  },
39
  "rewrite6": {
40
+ "page": 13,
41
+ "row": 316,
42
  "start": {
43
  "file": 0,
44
  "reader": {
45
  "format": "jsonl-row-v1",
46
+ "row": 53248
47
  }
48
  },
49
  "epoch": 47,
50
+ "offset": 2048,
51
+ "documents": 9707552
52
  },
53
  "openmath": {
54
+ "page": 2596,
55
+ "row": 750,
56
  "start": {
57
  "file": 24,
58
  "reader": {
59
  "format": "parquet-row-v1",
60
+ "row_group": 153,
61
+ "row": 617
62
  }
63
  },
64
  "epoch": 0,
65
  "offset": 0,
66
+ "documents": 10633966
67
  },
68
  "ultra_qa": {
69
+ "page": 1122,
70
+ "row": 1847,
71
  "start": {
72
  "file": 8,
73
  "reader": {
74
  "format": "parquet-row-v1",
75
  "row_group": 6,
76
+ "row": 51467
77
  }
78
  },
79
  "epoch": 0,
80
+ "offset": 2048,
81
+ "documents": 4597559
82
  },
83
  "ultra_style": {
84
+ "page": 1713,
85
+ "row": 1089,
86
  "start": {
87
  "file": 10,
88
  "reader": {
89
  "format": "parquet-row-v1",
90
  "row_group": 1,
91
+ "row": 81704
92
  }
93
  },
94
  "epoch": 0,
95
+ "offset": 0,
96
+ "documents": 7017537
97
  },
98
  "cortex": {
99
+ "page": 220582,
100
+ "row": 48,
101
  "start": null,
102
  "epoch": 0,
103
  "offset": 0,
104
+ "documents": 28234544
105
  }
106
  }
107
  }
latest.json CHANGED
@@ -1 +1 @@
1
- {"step": 840000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "3e144501faba5106e3f43a2e36d221a8630b9cdfb6837a8f976f58b534e86cfa", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "c1114d95865c877e56d4ca2db61e9232398961a900ec723d9fe48b310b10c038", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "5d089facc2fe4bc290dfa5a25481443cf9a74f9c2860a1b57795e6bdc0906c9e", "coverage.json": "abfd3dc053aa9107e89eeb9bfe2735dc1cd78f0a1640546271b969e4322c5588", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000840000"}
 
1
+ {"step": 842507, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "6f8382631e83ae42ed9759e16b99adad61f343fa6de9cb2d60b9dd6c9148bc70", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a2b248de903e43cc04b64f31262641a2a596e6720f6e8b83622de3cb8a1ad33a", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "76298eb667981f18a15e74c27926fbcd7a9ae27df4f2ad6d655bc8499e520b76", "coverage.json": "b6f294dc008ba78304b1fa2017cef88912e1ceaaf314ebcebdb825482582a1b3", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000842507"}
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3e144501faba5106e3f43a2e36d221a8630b9cdfb6837a8f976f58b534e86cfa
3
  size 360622744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f8382631e83ae42ed9759e16b99adad61f343fa6de9cb2d60b9dd6c9148bc70
3
  size 360622744
shard_scheduler.json CHANGED
@@ -22,8 +22,8 @@
22
  "tokens_per_second": 598315.063954628,
23
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
24
  "estimated_pages_for_buffer": 1,
25
- "required_token_positions_per_second": 77053.7277135633,
26
- "required_pages_per_second_upper_estimate": 0.005330755672504961
27
  },
28
  {
29
  "source": "rewrite6",
@@ -36,8 +36,8 @@
36
  "tokens_per_second": 8533.957213384612,
37
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
38
  "estimated_pages_for_buffer": 1,
39
- "required_token_positions_per_second": 9631.715964195413,
40
- "required_pages_per_second_upper_estimate": 0.0040325983411076795
41
  },
42
  {
43
  "source": "openmath",
@@ -50,8 +50,8 @@
50
  "tokens_per_second": 380144.35761751566,
51
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
52
  "estimated_pages_for_buffer": 1,
53
- "required_token_positions_per_second": 19263.431928390826,
54
- "required_pages_per_second_upper_estimate": 0.004012064859821606
55
  },
56
  {
57
  "source": "ultra_qa",
@@ -64,8 +64,8 @@
64
  "tokens_per_second": 43742.16556775435,
65
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
66
  "estimated_pages_for_buffer": 1,
67
- "required_token_positions_per_second": 28895.147892586243,
68
- "required_pages_per_second_upper_estimate": 0.0017322199958243636
69
  },
70
  {
71
  "source": "ultra_style",
@@ -78,8 +78,8 @@
78
  "tokens_per_second": 22018.86276523721,
79
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
80
  "estimated_pages_for_buffer": 1,
81
- "required_token_positions_per_second": 28895.147892586243,
82
- "required_pages_per_second_upper_estimate": 0.0022464706523249544
83
  },
84
  {
85
  "source": "cortex",
@@ -92,8 +92,8 @@
92
  "tokens_per_second": 910315.184299418,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
  "estimated_pages_for_buffer": 23,
95
- "required_token_positions_per_second": 28895.147892586243,
96
- "required_pages_per_second_upper_estimate": 0.3706359319735027
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,159 +158,27 @@
158
  "prefetch_batches": 63,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 720750,
161
- "producer_batch_seconds": 0.15130693903114434,
162
- "producer_p95_seconds": 0.37083707300189417,
163
- "gpu_update_seconds": 0.956838847227142,
164
- "producer_batches_per_second": 6.609082216607161,
165
- "gpu_batches_per_second": 1.0451080690316203,
166
- "producer_headroom": 6.3238266093678,
167
  "sustainable": true,
168
- "buffer_seconds": 60.280847375309946,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
- "upload_pages_per_commit": 82,
172
  "max_pending_pages": 4096,
173
- "generation_bytes_per_second": 369362.01326889056,
174
- "upload_bytes_per_second": 4183866.9114862867,
175
  "upload_sustainable": true,
176
  "logical_page_rows": 4096,
177
  "logical_cortex_page_rows": 128,
178
- "estimated_full_shards_per_hour": 79.25648973989523,
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": 2.2660662760026753,
184
- "bytes": 11276737,
185
- "pages": 36,
186
- "shards": 4
187
- },
188
- {
189
- "seconds": 1.8010207819970674,
190
- "bytes": 5858104,
191
- "pages": 32,
192
- "shards": 2
193
- },
194
- {
195
- "seconds": 1.3410764899963397,
196
- "bytes": 1019830,
197
- "pages": 31,
198
- "shards": 1
199
- },
200
- {
201
- "seconds": 1.8211984559966368,
202
- "bytes": 13136869,
203
- "pages": 32,
204
- "shards": 4
205
- },
206
- {
207
- "seconds": 1.4420257410019985,
208
- "bytes": 5442386,
209
- "pages": 33,
210
- "shards": 2
211
- },
212
- {
213
- "seconds": 2.6930293979967246,
214
- "bytes": 5991867,
215
- "pages": 29,
216
- "shards": 3
217
- },
218
- {
219
- "seconds": 1.7729905640007928,
220
- "bytes": 7619842,
221
- "pages": 39,
222
- "shards": 3
223
- },
224
- {
225
- "seconds": 1.2987675089971162,
226
- "bytes": 1223490,
227
- "pages": 35,
228
- "shards": 1
229
- },
230
- {
231
- "seconds": 1.546362834997126,
232
- "bytes": 11822887,
233
- "pages": 33,
234
- "shards": 4
235
- },
236
- {
237
- "seconds": 1.5667999549987144,
238
- "bytes": 6923893,
239
- "pages": 35,
240
- "shards": 2
241
- },
242
- {
243
- "seconds": 1.4826895270016394,
244
- "bytes": 6163509,
245
- "pages": 32,
246
- "shards": 2
247
- },
248
- {
249
- "seconds": 1.7263165549957193,
250
- "bytes": 7465725,
251
- "pages": 39,
252
- "shards": 3
253
- },
254
- {
255
- "seconds": 2.012805174999812,
256
- "bytes": 4595806,
257
- "pages": 30,
258
- "shards": 2
259
- },
260
- {
261
- "seconds": 1.6076512380022905,
262
- "bytes": 7510724,
263
- "pages": 40,
264
- "shards": 3
265
- },
266
- {
267
- "seconds": 1.5883049809999648,
268
- "bytes": 5665128,
269
- "pages": 27,
270
- "shards": 2
271
- },
272
- {
273
- "seconds": 2.0745964650050155,
274
- "bytes": 11457524,
275
- "pages": 35,
276
- "shards": 3
277
- },
278
- {
279
- "seconds": 1.6211695970050641,
280
- "bytes": 7634724,
281
- "pages": 40,
282
- "shards": 3
283
- },
284
- {
285
- "seconds": 1.4634708860030514,
286
- "bytes": 987981,
287
- "pages": 29,
288
- "shards": 1
289
- },
290
- {
291
- "seconds": 1.6518487119974452,
292
- "bytes": 7534957,
293
- "pages": 37,
294
- "shards": 3
295
- },
296
- {
297
- "seconds": 1.5817108970004483,
298
- "bytes": 10450177,
299
- "pages": 33,
300
- "shards": 3
301
- },
302
- {
303
- "seconds": 1.328966004999529,
304
- "bytes": 1184587,
305
- "pages": 36,
306
- "shards": 1
307
- },
308
- {
309
- "seconds": 2.0095185120007955,
310
- "bytes": 12961722,
311
- "pages": 36,
312
- "shards": 4
313
- },
314
  {
315
  "seconds": 1.4560003849983332,
316
  "bytes": 6473170,
@@ -562,6 +430,138 @@
562
  "bytes": 7374458,
563
  "pages": 32,
564
  "shards": 3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
565
  }
566
  ],
567
  "cache_key": {
 
22
  "tokens_per_second": 598315.063954628,
23
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
24
  "estimated_pages_for_buffer": 1,
25
+ "required_token_positions_per_second": 76375.6706640414,
26
+ "required_pages_per_second_upper_estimate": 0.005283846112509921
27
  },
28
  {
29
  "source": "rewrite6",
 
36
  "tokens_per_second": 8533.957213384612,
37
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
38
  "estimated_pages_for_buffer": 1,
39
+ "required_token_positions_per_second": 9546.958833005176,
40
+ "required_pages_per_second_upper_estimate": 0.003997112300208492
41
  },
42
  {
43
  "source": "openmath",
 
50
  "tokens_per_second": 380144.35761751566,
51
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
52
  "estimated_pages_for_buffer": 1,
53
+ "required_token_positions_per_second": 19093.91766601035,
54
+ "required_pages_per_second_upper_estimate": 0.003976759509359474
55
  },
56
  {
57
  "source": "ultra_qa",
 
64
  "tokens_per_second": 43742.16556775435,
65
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
66
  "estimated_pages_for_buffer": 1,
67
+ "required_token_positions_per_second": 28640.876499015525,
68
+ "required_pages_per_second_upper_estimate": 0.0017169768140297367
69
  },
70
  {
71
  "source": "ultra_style",
 
78
  "tokens_per_second": 22018.86276523721,
79
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
80
  "estimated_pages_for_buffer": 1,
81
+ "required_token_positions_per_second": 28640.876499015525,
82
+ "required_pages_per_second_upper_estimate": 0.002226702169896493
83
  },
84
  {
85
  "source": "cortex",
 
92
  "tokens_per_second": 910315.184299418,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
  "estimated_pages_for_buffer": 23,
95
+ "required_token_positions_per_second": 28640.876499015525,
96
+ "required_pages_per_second_upper_estimate": 0.36737441155212897
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
 
158
  "prefetch_batches": 63,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 720750,
161
+ "producer_batch_seconds": 0.15020428112444506,
162
+ "producer_p95_seconds": 0.3366278989997227,
163
+ "gpu_update_seconds": 0.9653335854072185,
164
+ "producer_batches_per_second": 6.657599853438895,
165
+ "gpu_batches_per_second": 1.0359113317062907,
166
+ "producer_headroom": 6.426804736726741,
167
  "sustainable": true,
168
+ "buffer_seconds": 60.81601588065477,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
+ "upload_pages_per_commit": 79,
172
  "max_pending_pages": 4096,
173
+ "generation_bytes_per_second": 273762.60107156524,
174
+ "upload_bytes_per_second": 4419757.457076919,
175
  "upload_sustainable": true,
176
  "logical_page_rows": 4096,
177
  "logical_cortex_page_rows": 128,
178
+ "estimated_full_shards_per_hour": 58.74308132276743,
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": 1.4560003849983332,
184
  "bytes": 6473170,
 
430
  "bytes": 7374458,
431
  "pages": 32,
432
  "shards": 3
433
+ },
434
+ {
435
+ "seconds": 1.3394468149999739,
436
+ "bytes": 10758114,
437
+ "pages": 29,
438
+ "shards": 3
439
+ },
440
+ {
441
+ "seconds": 1.5901181910012383,
442
+ "bytes": 260770,
443
+ "pages": 9,
444
+ "shards": 1
445
+ },
446
+ {
447
+ "seconds": 1.5258419460005825,
448
+ "bytes": 5807963,
449
+ "pages": 29,
450
+ "shards": 2
451
+ },
452
+ {
453
+ "seconds": 1.2447194750056951,
454
+ "bytes": 7832230,
455
+ "pages": 41,
456
+ "shards": 4
457
+ },
458
+ {
459
+ "seconds": 0.9895193189950078,
460
+ "bytes": 741196,
461
+ "pages": 23,
462
+ "shards": 1
463
+ },
464
+ {
465
+ "seconds": 1.489227063000726,
466
+ "bytes": 11724055,
467
+ "pages": 40,
468
+ "shards": 4
469
+ },
470
+ {
471
+ "seconds": 0.9323788800029433,
472
+ "bytes": 1111114,
473
+ "pages": 34,
474
+ "shards": 1
475
+ },
476
+ {
477
+ "seconds": 1.5997961780012702,
478
+ "bytes": 11835967,
479
+ "pages": 35,
480
+ "shards": 3
481
+ },
482
+ {
483
+ "seconds": 1.3631727270039846,
484
+ "bytes": 2146631,
485
+ "pages": 21,
486
+ "shards": 2
487
+ },
488
+ {
489
+ "seconds": 1.2327626959959161,
490
+ "bytes": 11052547,
491
+ "pages": 48,
492
+ "shards": 3
493
+ },
494
+ {
495
+ "seconds": 1.4408944379974855,
496
+ "bytes": 7298137,
497
+ "pages": 31,
498
+ "shards": 3
499
+ },
500
+ {
501
+ "seconds": 1.4694450449969736,
502
+ "bytes": 1110429,
503
+ "pages": 36,
504
+ "shards": 1
505
+ },
506
+ {
507
+ "seconds": 1.0526011850015493,
508
+ "bytes": 6120517,
509
+ "pages": 31,
510
+ "shards": 2
511
+ },
512
+ {
513
+ "seconds": 1.7904460849968018,
514
+ "bytes": 12133874,
515
+ "pages": 26,
516
+ "shards": 4
517
+ },
518
+ {
519
+ "seconds": 1.269238109998696,
520
+ "bytes": 6277214,
521
+ "pages": 37,
522
+ "shards": 2
523
+ },
524
+ {
525
+ "seconds": 1.02439347699692,
526
+ "bytes": 5979878,
527
+ "pages": 29,
528
+ "shards": 2
529
+ },
530
+ {
531
+ "seconds": 1.1111059189934167,
532
+ "bytes": 2867400,
533
+ "pages": 39,
534
+ "shards": 2
535
+ },
536
+ {
537
+ "seconds": 1.2084345140028745,
538
+ "bytes": 10491799,
539
+ "pages": 30,
540
+ "shards": 3
541
+ },
542
+ {
543
+ "seconds": 0.8811795219953638,
544
+ "bytes": 776994,
545
+ "pages": 26,
546
+ "shards": 1
547
+ },
548
+ {
549
+ "seconds": 1.837297693004075,
550
+ "bytes": 13040260,
551
+ "pages": 38,
552
+ "shards": 4
553
+ },
554
+ {
555
+ "seconds": 1.5299789529963164,
556
+ "bytes": 6406874,
557
+ "pages": 38,
558
+ "shards": 2
559
+ },
560
+ {
561
+ "seconds": 1.7032136210036697,
562
+ "bytes": 6726184,
563
+ "pages": 31,
564
+ "shards": 3
565
  }
566
  ],
567
  "cache_key": {
trainer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c1114d95865c877e56d4ca2db61e9232398961a900ec723d9fe48b310b10c038
3
- size 721765631
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2b248de903e43cc04b64f31262641a2a596e6720f6e8b83622de3cb8a1ad33a
3
+ size 721782015