kevinoli commited on
Commit
cf1d350
1 Parent(s): 6743a65

Training in progress, step 19500, checkpoint

Browse files
checkpoint-19500/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6ab80d6629ca6dccb5ab90bd1f4a4a09f13e81d83a3e524b7e52d5ca985d4ae9
3
  size 1711848436
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5365ba9173e945a87b80106b0b53090205abbca802ab495e2c8cd0ee6baa3b3d
3
  size 1711848436
checkpoint-19500/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9baf6b22ae74ec59cd3b43289d181cae1bf46e8ec8cd14cd9eee923df193101c
3
  size 3424043887
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:decad4770499ab7a04e2136087e80a44824686ba17de20a906e6130712f9291c
3
  size 3424043887
checkpoint-19500/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3539aa6e95b2141f42894398485aadda8c64ea1db50724b6372df41842f8b2d0
3
  size 623
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eff1530934bb27035b7ae18b3d0970b8a99d7eec272d0d97cb90e21d19a52a0b
3
  size 623
checkpoint-19500/trainer_state.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "best_metric": 0.5568962693214417,
3
- "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l57-l/checkpoint-18500",
4
- "epoch": 2.1952043228638973,
5
  "eval_steps": 500,
6
  "global_step": 19500,
7
  "is_hyper_param_search": false,
@@ -9,595 +9,595 @@
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
- "epoch": 0.056287290329843524,
13
- "grad_norm": 150.1642303466797,
14
- "learning_rate": 4.929640887087696e-07,
15
- "loss": 0.3486,
16
  "step": 500
17
  },
18
  {
19
- "epoch": 0.056287290329843524,
20
- "eval_loss": 1.1265727281570435,
21
- "eval_runtime": 123.7806,
22
- "eval_samples_per_second": 15.948,
23
- "eval_steps_per_second": 1.995,
24
  "step": 500
25
  },
26
  {
27
- "epoch": 0.11257458065968705,
28
- "grad_norm": 366.6913146972656,
29
- "learning_rate": 4.859281774175391e-07,
30
- "loss": 0.2733,
31
  "step": 1000
32
  },
33
  {
34
- "epoch": 0.11257458065968705,
35
- "eval_loss": 0.9742079377174377,
36
- "eval_runtime": 125.8047,
37
- "eval_samples_per_second": 15.691,
38
- "eval_steps_per_second": 1.963,
39
  "step": 1000
40
  },
41
  {
42
- "epoch": 0.16886187098953057,
43
- "grad_norm": 555.5016479492188,
44
- "learning_rate": 4.788922661263087e-07,
45
- "loss": 0.1851,
46
  "step": 1500
47
  },
48
  {
49
- "epoch": 0.16886187098953057,
50
- "eval_loss": 0.9162159562110901,
51
- "eval_runtime": 126.7606,
52
- "eval_samples_per_second": 15.573,
53
- "eval_steps_per_second": 1.949,
54
  "step": 1500
55
  },
56
  {
57
- "epoch": 0.2251491613193741,
58
- "grad_norm": 0.423260897397995,
59
- "learning_rate": 4.7185635483507824e-07,
60
- "loss": 0.1973,
61
  "step": 2000
62
  },
63
  {
64
- "epoch": 0.2251491613193741,
65
- "eval_loss": 0.8716973662376404,
66
- "eval_runtime": 127.6955,
67
- "eval_samples_per_second": 15.459,
68
- "eval_steps_per_second": 1.934,
69
  "step": 2000
70
  },
71
  {
72
- "epoch": 0.2814364516492176,
73
- "grad_norm": 6.840139389038086,
74
- "learning_rate": 4.6482044354384774e-07,
75
- "loss": 0.1881,
76
  "step": 2500
77
  },
78
  {
79
- "epoch": 0.2814364516492176,
80
- "eval_loss": 0.830047070980072,
81
- "eval_runtime": 128.0771,
82
- "eval_samples_per_second": 15.413,
83
- "eval_steps_per_second": 1.929,
84
  "step": 2500
85
  },
86
  {
87
- "epoch": 0.33772374197906113,
88
- "grad_norm": 729.8377685546875,
89
- "learning_rate": 4.577845322526173e-07,
90
- "loss": 0.1695,
91
  "step": 3000
92
  },
93
  {
94
- "epoch": 0.33772374197906113,
95
- "eval_loss": 0.8265627026557922,
96
- "eval_runtime": 128.052,
97
- "eval_samples_per_second": 15.416,
98
- "eval_steps_per_second": 1.929,
99
  "step": 3000
100
  },
101
  {
102
- "epoch": 0.39401103230890466,
103
- "grad_norm": 7.96999938756926e-06,
104
- "learning_rate": 4.507486209613869e-07,
105
- "loss": 0.155,
106
  "step": 3500
107
  },
108
  {
109
- "epoch": 0.39401103230890466,
110
- "eval_loss": 0.7721038460731506,
111
- "eval_runtime": 128.1871,
112
- "eval_samples_per_second": 15.399,
113
- "eval_steps_per_second": 1.927,
114
  "step": 3500
115
  },
116
  {
117
- "epoch": 0.4502983226387482,
118
- "grad_norm": 0.006555848754942417,
119
- "learning_rate": 4.4371270967015645e-07,
120
- "loss": 0.1216,
121
  "step": 4000
122
  },
123
  {
124
- "epoch": 0.4502983226387482,
125
- "eval_loss": 0.7490401864051819,
126
- "eval_runtime": 127.5516,
127
- "eval_samples_per_second": 15.476,
128
- "eval_steps_per_second": 1.936,
129
  "step": 4000
130
  },
131
  {
132
- "epoch": 0.5065856129685917,
133
- "grad_norm": 0.0004515685432124883,
134
- "learning_rate": 4.36676798378926e-07,
135
- "loss": 0.1531,
136
  "step": 4500
137
  },
138
  {
139
- "epoch": 0.5065856129685917,
140
- "eval_loss": 0.7533182501792908,
141
- "eval_runtime": 127.4797,
142
- "eval_samples_per_second": 15.485,
143
- "eval_steps_per_second": 1.938,
144
  "step": 4500
145
  },
146
  {
147
- "epoch": 0.5628729032984352,
148
- "grad_norm": 4.4633176003117114e-05,
149
- "learning_rate": 4.2964088708769556e-07,
150
- "loss": 0.1154,
151
  "step": 5000
152
  },
153
  {
154
- "epoch": 0.5628729032984352,
155
- "eval_loss": 0.747530460357666,
156
- "eval_runtime": 127.8675,
157
- "eval_samples_per_second": 15.438,
158
- "eval_steps_per_second": 1.932,
159
  "step": 5000
160
  },
161
  {
162
- "epoch": 0.6191601936282788,
163
- "grad_norm": 4.158839702606201,
164
- "learning_rate": 4.2260497579646517e-07,
165
- "loss": 0.1407,
166
  "step": 5500
167
  },
168
  {
169
- "epoch": 0.6191601936282788,
170
- "eval_loss": 0.724854588508606,
171
- "eval_runtime": 125.1629,
172
- "eval_samples_per_second": 15.771,
173
- "eval_steps_per_second": 1.973,
174
  "step": 5500
175
  },
176
  {
177
- "epoch": 0.6754474839581223,
178
- "grad_norm": 0.748630166053772,
179
- "learning_rate": 4.155690645052347e-07,
180
- "loss": 0.1447,
181
  "step": 6000
182
  },
183
  {
184
- "epoch": 0.6754474839581223,
185
- "eval_loss": 0.7091771960258484,
186
- "eval_runtime": 125.2549,
187
- "eval_samples_per_second": 15.76,
188
- "eval_steps_per_second": 1.972,
189
  "step": 6000
190
  },
191
  {
192
- "epoch": 0.7317347742879657,
193
- "grad_norm": 2.8120924980612472e-05,
194
- "learning_rate": 4.0853315321400427e-07,
195
- "loss": 0.1146,
196
  "step": 6500
197
  },
198
  {
199
- "epoch": 0.7317347742879657,
200
- "eval_loss": 0.7105826735496521,
201
- "eval_runtime": 127.7786,
202
- "eval_samples_per_second": 15.449,
203
- "eval_steps_per_second": 1.933,
204
  "step": 6500
205
  },
206
  {
207
- "epoch": 0.7880220646178093,
208
- "grad_norm": 1.4741635823156685e-07,
209
- "learning_rate": 4.014972419227738e-07,
210
- "loss": 0.171,
211
  "step": 7000
212
  },
213
  {
214
- "epoch": 0.7880220646178093,
215
- "eval_loss": 0.6978534460067749,
216
- "eval_runtime": 125.2567,
217
- "eval_samples_per_second": 15.76,
218
- "eval_steps_per_second": 1.972,
219
  "step": 7000
220
  },
221
  {
222
- "epoch": 0.8443093549476528,
223
- "grad_norm": 0.002975167240947485,
224
- "learning_rate": 3.944613306315434e-07,
225
- "loss": 0.1584,
226
  "step": 7500
227
  },
228
  {
229
- "epoch": 0.8443093549476528,
230
- "eval_loss": 0.6923494935035706,
231
- "eval_runtime": 125.3384,
232
- "eval_samples_per_second": 15.749,
233
- "eval_steps_per_second": 1.971,
234
  "step": 7500
235
  },
236
  {
237
- "epoch": 0.9005966452774964,
238
- "grad_norm": 9.225498797604814e-05,
239
- "learning_rate": 3.8742541934031293e-07,
240
- "loss": 0.1384,
241
  "step": 8000
242
  },
243
  {
244
- "epoch": 0.9005966452774964,
245
- "eval_loss": 0.6872902512550354,
246
- "eval_runtime": 128.114,
247
- "eval_samples_per_second": 15.408,
248
- "eval_steps_per_second": 1.928,
249
  "step": 8000
250
  },
251
  {
252
- "epoch": 0.9568839356073399,
253
- "grad_norm": 1.9474915902151224e-08,
254
- "learning_rate": 3.803895080490825e-07,
255
- "loss": 0.1243,
256
  "step": 8500
257
  },
258
  {
259
- "epoch": 0.9568839356073399,
260
- "eval_loss": 0.6788680553436279,
261
- "eval_runtime": 125.0525,
262
- "eval_samples_per_second": 15.785,
263
- "eval_steps_per_second": 1.975,
264
  "step": 8500
265
  },
266
  {
267
- "epoch": 1.0131712259371835,
268
- "grad_norm": 5.165647506713867,
269
- "learning_rate": 3.7335359675785204e-07,
270
- "loss": 0.1102,
271
  "step": 9000
272
  },
273
  {
274
- "epoch": 1.0131712259371835,
275
- "eval_loss": 0.6788864135742188,
276
- "eval_runtime": 124.7777,
277
- "eval_samples_per_second": 15.82,
278
- "eval_steps_per_second": 1.98,
279
  "step": 9000
280
  },
281
  {
282
- "epoch": 1.069458516267027,
283
- "grad_norm": 0.0005840375670231879,
284
- "learning_rate": 3.6631768546662164e-07,
285
- "loss": 0.0523,
286
  "step": 9500
287
  },
288
  {
289
- "epoch": 1.069458516267027,
290
- "eval_loss": 0.6628187894821167,
291
- "eval_runtime": 127.8018,
292
- "eval_samples_per_second": 15.446,
293
- "eval_steps_per_second": 1.933,
294
  "step": 9500
295
  },
296
  {
297
- "epoch": 1.1257458065968704,
298
- "grad_norm": 0.001191094284877181,
299
- "learning_rate": 3.592817741753912e-07,
300
- "loss": 0.0617,
301
  "step": 10000
302
  },
303
  {
304
- "epoch": 1.1257458065968704,
305
- "eval_loss": 0.6507942080497742,
306
- "eval_runtime": 125.484,
307
- "eval_samples_per_second": 15.731,
308
- "eval_steps_per_second": 1.968,
309
  "step": 10000
310
  },
311
  {
312
- "epoch": 1.1820330969267139,
313
- "grad_norm": 0.0003092484548687935,
314
- "learning_rate": 3.5224586288416075e-07,
315
- "loss": 0.0598,
316
  "step": 10500
317
  },
318
  {
319
- "epoch": 1.1820330969267139,
320
- "eval_loss": 0.6480408906936646,
321
- "eval_runtime": 125.7661,
322
- "eval_samples_per_second": 15.696,
323
- "eval_steps_per_second": 1.964,
324
  "step": 10500
325
  },
326
  {
327
- "epoch": 1.2383203872565574,
328
- "grad_norm": 5.508670710696606e-07,
329
- "learning_rate": 3.452099515929303e-07,
330
- "loss": 0.0782,
331
  "step": 11000
332
  },
333
  {
334
- "epoch": 1.2383203872565574,
335
- "eval_loss": 0.6403237581253052,
336
- "eval_runtime": 128.3947,
337
- "eval_samples_per_second": 15.374,
338
- "eval_steps_per_second": 1.924,
339
  "step": 11000
340
  },
341
  {
342
- "epoch": 1.294607677586401,
343
- "grad_norm": 0.1470593810081482,
344
- "learning_rate": 3.3817404030169986e-07,
345
- "loss": 0.0616,
346
  "step": 11500
347
  },
348
  {
349
- "epoch": 1.294607677586401,
350
- "eval_loss": 0.6376460790634155,
351
- "eval_runtime": 125.5391,
352
- "eval_samples_per_second": 15.724,
353
- "eval_steps_per_second": 1.968,
354
  "step": 11500
355
  },
356
  {
357
- "epoch": 1.3508949679162445,
358
- "grad_norm": 1.8021532014245167e-05,
359
- "learning_rate": 3.3113812901046946e-07,
360
- "loss": 0.0723,
361
  "step": 12000
362
  },
363
  {
364
- "epoch": 1.3508949679162445,
365
- "eval_loss": 0.6378474831581116,
366
- "eval_runtime": 125.5914,
367
- "eval_samples_per_second": 15.718,
368
- "eval_steps_per_second": 1.967,
369
  "step": 12000
370
  },
371
  {
372
- "epoch": 1.407182258246088,
373
- "grad_norm": 0.00013483635848388076,
374
- "learning_rate": 3.2410221771923896e-07,
375
- "loss": 0.0459,
376
  "step": 12500
377
  },
378
  {
379
- "epoch": 1.407182258246088,
380
- "eval_loss": 0.6378260850906372,
381
- "eval_runtime": 128.2725,
382
- "eval_samples_per_second": 15.389,
383
- "eval_steps_per_second": 1.926,
384
  "step": 12500
385
  },
386
  {
387
- "epoch": 1.4634695485759315,
388
- "grad_norm": 4.757418707868055e-07,
389
- "learning_rate": 3.170663064280085e-07,
390
- "loss": 0.0571,
391
  "step": 13000
392
  },
393
  {
394
- "epoch": 1.4634695485759315,
395
- "eval_loss": 0.6366899609565735,
396
- "eval_runtime": 125.6986,
397
- "eval_samples_per_second": 15.704,
398
- "eval_steps_per_second": 1.965,
399
  "step": 13000
400
  },
401
  {
402
- "epoch": 1.5197568389057752,
403
- "grad_norm": 0.00028420978924259543,
404
- "learning_rate": 3.1003039513677807e-07,
405
- "loss": 0.0499,
406
  "step": 13500
407
  },
408
  {
409
- "epoch": 1.5197568389057752,
410
- "eval_loss": 0.6286123394966125,
411
- "eval_runtime": 127.9165,
412
- "eval_samples_per_second": 15.432,
413
- "eval_steps_per_second": 1.931,
414
  "step": 13500
415
  },
416
  {
417
- "epoch": 1.5760441292356187,
418
- "grad_norm": 2.1999912291903456e-07,
419
- "learning_rate": 3.0299448384554767e-07,
420
- "loss": 0.0679,
421
  "step": 14000
422
  },
423
  {
424
- "epoch": 1.5760441292356187,
425
- "eval_loss": 0.634914755821228,
426
- "eval_runtime": 125.0396,
427
- "eval_samples_per_second": 15.787,
428
- "eval_steps_per_second": 1.975,
429
  "step": 14000
430
  },
431
  {
432
- "epoch": 1.6323314195654621,
433
- "grad_norm": 0.00023505109129473567,
434
- "learning_rate": 2.959585725543172e-07,
435
- "loss": 0.0554,
436
  "step": 14500
437
  },
438
  {
439
- "epoch": 1.6323314195654621,
440
- "eval_loss": 0.623322069644928,
441
- "eval_runtime": 125.2429,
442
- "eval_samples_per_second": 15.761,
443
- "eval_steps_per_second": 1.972,
444
  "step": 14500
445
  },
446
  {
447
- "epoch": 1.6886187098953056,
448
- "grad_norm": 11.346132278442383,
449
- "learning_rate": 2.889226612630868e-07,
450
- "loss": 0.0617,
451
  "step": 15000
452
  },
453
  {
454
- "epoch": 1.6886187098953056,
455
- "eval_loss": 0.6115825772285461,
456
- "eval_runtime": 127.8813,
457
- "eval_samples_per_second": 15.436,
458
- "eval_steps_per_second": 1.931,
459
  "step": 15000
460
  },
461
  {
462
- "epoch": 1.744906000225149,
463
- "grad_norm": 308.43804931640625,
464
- "learning_rate": 2.8188674997185633e-07,
465
- "loss": 0.0561,
466
  "step": 15500
467
  },
468
  {
469
- "epoch": 1.744906000225149,
470
- "eval_loss": 0.5951005220413208,
471
- "eval_runtime": 125.2491,
472
- "eval_samples_per_second": 15.761,
473
- "eval_steps_per_second": 1.972,
474
  "step": 15500
475
  },
476
  {
477
- "epoch": 1.8011932905549926,
478
- "grad_norm": 6.91558055265773e-09,
479
- "learning_rate": 2.7485083868062594e-07,
480
- "loss": 0.0672,
481
  "step": 16000
482
  },
483
  {
484
- "epoch": 1.8011932905549926,
485
- "eval_loss": 0.5809568166732788,
486
- "eval_runtime": 125.2557,
487
- "eval_samples_per_second": 15.76,
488
- "eval_steps_per_second": 1.972,
489
  "step": 16000
490
  },
491
  {
492
- "epoch": 1.8574805808848363,
493
- "grad_norm": 2.2250075915053458e-07,
494
- "learning_rate": 2.678149273893955e-07,
495
- "loss": 0.0812,
496
  "step": 16500
497
  },
498
  {
499
- "epoch": 1.8574805808848363,
500
- "eval_loss": 0.5691242218017578,
501
- "eval_runtime": 128.3032,
502
- "eval_samples_per_second": 15.385,
503
- "eval_steps_per_second": 1.925,
504
  "step": 16500
505
  },
506
  {
507
- "epoch": 1.9137678712146797,
508
- "grad_norm": 1.1724663972854614,
509
- "learning_rate": 2.6077901609816504e-07,
510
- "loss": 0.0614,
511
  "step": 17000
512
  },
513
  {
514
- "epoch": 1.9137678712146797,
515
- "eval_loss": 0.5643230080604553,
516
- "eval_runtime": 125.7191,
517
- "eval_samples_per_second": 15.702,
518
- "eval_steps_per_second": 1.965,
519
  "step": 17000
520
  },
521
  {
522
- "epoch": 1.9700551615445232,
523
- "grad_norm": 0.00030226208036765456,
524
- "learning_rate": 2.5374310480693454e-07,
525
- "loss": 0.0462,
526
  "step": 17500
527
  },
528
  {
529
- "epoch": 1.9700551615445232,
530
- "eval_loss": 0.5657238364219666,
531
- "eval_runtime": 124.9923,
532
- "eval_samples_per_second": 15.793,
533
- "eval_steps_per_second": 1.976,
534
  "step": 17500
535
  },
536
  {
537
- "epoch": 2.026342451874367,
538
- "grad_norm": 1.949754704355655e-07,
539
- "learning_rate": 2.4670719351570415e-07,
540
- "loss": 0.0565,
541
  "step": 18000
542
  },
543
  {
544
- "epoch": 2.026342451874367,
545
- "eval_loss": 0.5607953667640686,
546
- "eval_runtime": 127.7187,
547
- "eval_samples_per_second": 15.456,
548
- "eval_steps_per_second": 1.934,
549
  "step": 18000
550
  },
551
  {
552
- "epoch": 2.0826297422042104,
553
- "grad_norm": 14.2212553024292,
554
- "learning_rate": 2.396712822244737e-07,
555
- "loss": 0.0549,
556
  "step": 18500
557
  },
558
  {
559
- "epoch": 2.0826297422042104,
560
- "eval_loss": 0.5568962693214417,
561
- "eval_runtime": 125.0503,
562
- "eval_samples_per_second": 15.786,
563
- "eval_steps_per_second": 1.975,
564
  "step": 18500
565
  },
566
  {
567
- "epoch": 2.138917032534054,
568
- "grad_norm": 1.2045973539352417,
569
- "learning_rate": 2.3263537093324326e-07,
570
- "loss": 0.0245,
571
  "step": 19000
572
  },
573
  {
574
- "epoch": 2.138917032534054,
575
- "eval_loss": 0.5595241785049438,
576
- "eval_runtime": 124.8527,
577
- "eval_samples_per_second": 15.811,
578
- "eval_steps_per_second": 1.978,
579
  "step": 19000
580
  },
581
  {
582
- "epoch": 2.1952043228638973,
583
- "grad_norm": 8.770056410867255e-06,
584
- "learning_rate": 2.2559945964201284e-07,
585
- "loss": 0.0428,
586
  "step": 19500
587
  },
588
  {
589
- "epoch": 2.1952043228638973,
590
- "eval_loss": 0.5740877985954285,
591
- "eval_runtime": 127.9211,
592
- "eval_samples_per_second": 15.431,
593
- "eval_steps_per_second": 1.931,
594
  "step": 19500
595
  }
596
  ],
597
  "logging_steps": 500,
598
- "max_steps": 35532,
599
  "num_input_tokens_seen": 0,
600
- "num_train_epochs": 4,
601
  "save_steps": 500,
602
  "stateful_callbacks": {
603
  "TrainerControl": {
@@ -611,7 +611,7 @@
611
  "attributes": {}
612
  }
613
  },
614
- "total_flos": 7015647607921260.0,
615
  "train_batch_size": 2,
616
  "trial_name": null,
617
  "trial_params": null
 
1
  {
2
+ "best_metric": 0.4926412105560303,
3
+ "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l57-l/checkpoint-19500",
4
+ "epoch": 2.079778156996587,
5
  "eval_steps": 500,
6
  "global_step": 19500,
7
  "is_hyper_param_search": false,
 
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
+ "epoch": 0.05332764505119454,
13
+ "grad_norm": 414.2552795410156,
14
+ "learning_rate": 4.911120591581342e-07,
15
+ "loss": 0.3809,
16
  "step": 500
17
  },
18
  {
19
+ "epoch": 0.05332764505119454,
20
+ "eval_loss": 1.1109352111816406,
21
+ "eval_runtime": 63.6734,
22
+ "eval_samples_per_second": 15.501,
23
+ "eval_steps_per_second": 1.947,
24
  "step": 500
25
  },
26
  {
27
+ "epoch": 0.10665529010238908,
28
+ "grad_norm": 45.880027770996094,
29
+ "learning_rate": 4.822241183162685e-07,
30
+ "loss": 0.2649,
31
  "step": 1000
32
  },
33
  {
34
+ "epoch": 0.10665529010238908,
35
+ "eval_loss": 0.9588962197303772,
36
+ "eval_runtime": 62.6825,
37
+ "eval_samples_per_second": 15.746,
38
+ "eval_steps_per_second": 1.978,
39
  "step": 1000
40
  },
41
  {
42
+ "epoch": 0.1599829351535836,
43
+ "grad_norm": 432.9915466308594,
44
+ "learning_rate": 4.733361774744027e-07,
45
+ "loss": 0.2104,
46
  "step": 1500
47
  },
48
  {
49
+ "epoch": 0.1599829351535836,
50
+ "eval_loss": 0.90110844373703,
51
+ "eval_runtime": 62.6757,
52
+ "eval_samples_per_second": 15.748,
53
+ "eval_steps_per_second": 1.978,
54
  "step": 1500
55
  },
56
  {
57
+ "epoch": 0.21331058020477817,
58
+ "grad_norm": 0.014208819717168808,
59
+ "learning_rate": 4.6444823663253695e-07,
60
+ "loss": 0.1849,
61
  "step": 2000
62
  },
63
  {
64
+ "epoch": 0.21331058020477817,
65
+ "eval_loss": 0.8570474982261658,
66
+ "eval_runtime": 63.7296,
67
+ "eval_samples_per_second": 15.487,
68
+ "eval_steps_per_second": 1.946,
69
  "step": 2000
70
  },
71
  {
72
+ "epoch": 0.2666382252559727,
73
+ "grad_norm": 174.73324584960938,
74
+ "learning_rate": 4.5556029579067116e-07,
75
+ "loss": 0.2056,
76
  "step": 2500
77
  },
78
  {
79
+ "epoch": 0.2666382252559727,
80
+ "eval_loss": 0.7895939350128174,
81
+ "eval_runtime": 63.8387,
82
+ "eval_samples_per_second": 15.461,
83
+ "eval_steps_per_second": 1.942,
84
  "step": 2500
85
  },
86
  {
87
+ "epoch": 0.3199658703071672,
88
+ "grad_norm": 9.852073823424234e-09,
89
+ "learning_rate": 4.4667235494880547e-07,
90
+ "loss": 0.159,
91
  "step": 3000
92
  },
93
  {
94
+ "epoch": 0.3199658703071672,
95
+ "eval_loss": 0.7647623419761658,
96
+ "eval_runtime": 63.9155,
97
+ "eval_samples_per_second": 15.442,
98
+ "eval_steps_per_second": 1.94,
99
  "step": 3000
100
  },
101
  {
102
+ "epoch": 0.37329351535836175,
103
+ "grad_norm": 0.0025215104687958956,
104
+ "learning_rate": 4.377844141069397e-07,
105
+ "loss": 0.1696,
106
  "step": 3500
107
  },
108
  {
109
+ "epoch": 0.37329351535836175,
110
+ "eval_loss": 0.7638036012649536,
111
+ "eval_runtime": 63.8003,
112
+ "eval_samples_per_second": 15.47,
113
+ "eval_steps_per_second": 1.944,
114
  "step": 3500
115
  },
116
  {
117
+ "epoch": 0.42662116040955633,
118
+ "grad_norm": 2.4611830711364746,
119
+ "learning_rate": 4.2889647326507393e-07,
120
+ "loss": 0.1125,
121
  "step": 4000
122
  },
123
  {
124
+ "epoch": 0.42662116040955633,
125
+ "eval_loss": 0.7486374378204346,
126
+ "eval_runtime": 63.9847,
127
+ "eval_samples_per_second": 15.426,
128
+ "eval_steps_per_second": 1.938,
129
  "step": 4000
130
  },
131
  {
132
+ "epoch": 0.47994880546075086,
133
+ "grad_norm": 5.230295658111572,
134
+ "learning_rate": 4.2000853242320814e-07,
135
+ "loss": 0.1322,
136
  "step": 4500
137
  },
138
  {
139
+ "epoch": 0.47994880546075086,
140
+ "eval_loss": 0.7251659035682678,
141
+ "eval_runtime": 64.4782,
142
+ "eval_samples_per_second": 15.308,
143
+ "eval_steps_per_second": 1.923,
144
  "step": 4500
145
  },
146
  {
147
+ "epoch": 0.5332764505119454,
148
+ "grad_norm": 3.821002974291332e-05,
149
+ "learning_rate": 4.1112059158134245e-07,
150
+ "loss": 0.1633,
151
  "step": 5000
152
  },
153
  {
154
+ "epoch": 0.5332764505119454,
155
+ "eval_loss": 0.7035414576530457,
156
+ "eval_runtime": 62.6186,
157
+ "eval_samples_per_second": 15.762,
158
+ "eval_steps_per_second": 1.98,
159
  "step": 5000
160
  },
161
  {
162
+ "epoch": 0.58660409556314,
163
+ "grad_norm": 5.916939244343666e-06,
164
+ "learning_rate": 4.0223265073947665e-07,
165
+ "loss": 0.125,
166
  "step": 5500
167
  },
168
  {
169
+ "epoch": 0.58660409556314,
170
+ "eval_loss": 0.6792380809783936,
171
+ "eval_runtime": 65.0551,
172
+ "eval_samples_per_second": 15.172,
173
+ "eval_steps_per_second": 1.906,
174
  "step": 5500
175
  },
176
  {
177
+ "epoch": 0.6399317406143344,
178
+ "grad_norm": 92.8386459350586,
179
+ "learning_rate": 3.933447098976109e-07,
180
+ "loss": 0.1297,
181
  "step": 6000
182
  },
183
  {
184
+ "epoch": 0.6399317406143344,
185
+ "eval_loss": 0.6549482345581055,
186
+ "eval_runtime": 64.0827,
187
+ "eval_samples_per_second": 15.402,
188
+ "eval_steps_per_second": 1.935,
189
  "step": 6000
190
  },
191
  {
192
+ "epoch": 0.693259385665529,
193
+ "grad_norm": 0.00014717792510055006,
194
+ "learning_rate": 3.8445676905574517e-07,
195
+ "loss": 0.1114,
196
  "step": 6500
197
  },
198
  {
199
+ "epoch": 0.693259385665529,
200
+ "eval_loss": 0.6223254799842834,
201
+ "eval_runtime": 64.89,
202
+ "eval_samples_per_second": 15.21,
203
+ "eval_steps_per_second": 1.911,
204
  "step": 6500
205
  },
206
  {
207
+ "epoch": 0.7465870307167235,
208
+ "grad_norm": 4.789559397977428e-07,
209
+ "learning_rate": 3.755688282138794e-07,
210
+ "loss": 0.144,
211
  "step": 7000
212
  },
213
  {
214
+ "epoch": 0.7465870307167235,
215
+ "eval_loss": 0.6204637289047241,
216
+ "eval_runtime": 62.7674,
217
+ "eval_samples_per_second": 15.725,
218
+ "eval_steps_per_second": 1.976,
219
  "step": 7000
220
  },
221
  {
222
+ "epoch": 0.7999146757679181,
223
+ "grad_norm": 1.7078508138656616,
224
+ "learning_rate": 3.6668088737201363e-07,
225
+ "loss": 0.1399,
226
  "step": 7500
227
  },
228
  {
229
+ "epoch": 0.7999146757679181,
230
+ "eval_loss": 0.6074371933937073,
231
+ "eval_runtime": 63.4694,
232
+ "eval_samples_per_second": 15.551,
233
+ "eval_steps_per_second": 1.954,
234
  "step": 7500
235
  },
236
  {
237
+ "epoch": 0.8532423208191127,
238
+ "grad_norm": 870.1183471679688,
239
+ "learning_rate": 3.5779294653014783e-07,
240
+ "loss": 0.141,
241
  "step": 8000
242
  },
243
  {
244
+ "epoch": 0.8532423208191127,
245
+ "eval_loss": 0.5950366854667664,
246
+ "eval_runtime": 62.915,
247
+ "eval_samples_per_second": 15.688,
248
+ "eval_steps_per_second": 1.971,
249
  "step": 8000
250
  },
251
  {
252
+ "epoch": 0.9065699658703071,
253
+ "grad_norm": 5.781071013188921e-05,
254
+ "learning_rate": 3.4890500568828214e-07,
255
+ "loss": 0.1291,
256
  "step": 8500
257
  },
258
  {
259
+ "epoch": 0.9065699658703071,
260
+ "eval_loss": 0.5800224542617798,
261
+ "eval_runtime": 62.3336,
262
+ "eval_samples_per_second": 15.834,
263
+ "eval_steps_per_second": 1.989,
264
  "step": 8500
265
  },
266
  {
267
+ "epoch": 0.9598976109215017,
268
+ "grad_norm": 1.3669992685317993,
269
+ "learning_rate": 3.4001706484641635e-07,
270
+ "loss": 0.1246,
271
  "step": 9000
272
  },
273
  {
274
+ "epoch": 0.9598976109215017,
275
+ "eval_loss": 0.5702486038208008,
276
+ "eval_runtime": 64.055,
277
+ "eval_samples_per_second": 15.409,
278
+ "eval_steps_per_second": 1.936,
279
  "step": 9000
280
  },
281
  {
282
+ "epoch": 1.0132252559726962,
283
+ "grad_norm": 644.68115234375,
284
+ "learning_rate": 3.311291240045506e-07,
285
+ "loss": 0.1257,
286
  "step": 9500
287
  },
288
  {
289
+ "epoch": 1.0132252559726962,
290
+ "eval_loss": 0.5807780623435974,
291
+ "eval_runtime": 62.5312,
292
+ "eval_samples_per_second": 15.784,
293
+ "eval_steps_per_second": 1.983,
294
  "step": 9500
295
  },
296
  {
297
+ "epoch": 1.0665529010238908,
298
+ "grad_norm": 27.854568481445312,
299
+ "learning_rate": 3.2224118316268486e-07,
300
+ "loss": 0.0336,
301
  "step": 10000
302
  },
303
  {
304
+ "epoch": 1.0665529010238908,
305
+ "eval_loss": 0.5713164210319519,
306
+ "eval_runtime": 62.5622,
307
+ "eval_samples_per_second": 15.776,
308
+ "eval_steps_per_second": 1.982,
309
  "step": 10000
310
  },
311
  {
312
+ "epoch": 1.1198805460750854,
313
+ "grad_norm": 0.00045933027286082506,
314
+ "learning_rate": 3.133532423208191e-07,
315
+ "loss": 0.0663,
316
  "step": 10500
317
  },
318
  {
319
+ "epoch": 1.1198805460750854,
320
+ "eval_loss": 0.5601483583450317,
321
+ "eval_runtime": 63.4675,
322
+ "eval_samples_per_second": 15.551,
323
+ "eval_steps_per_second": 1.954,
324
  "step": 10500
325
  },
326
  {
327
+ "epoch": 1.17320819112628,
328
+ "grad_norm": 0.14330387115478516,
329
+ "learning_rate": 3.044653014789533e-07,
330
+ "loss": 0.078,
331
  "step": 11000
332
  },
333
  {
334
+ "epoch": 1.17320819112628,
335
+ "eval_loss": 0.5580261945724487,
336
+ "eval_runtime": 63.3927,
337
+ "eval_samples_per_second": 15.57,
338
+ "eval_steps_per_second": 1.956,
339
  "step": 11000
340
  },
341
  {
342
+ "epoch": 1.2265358361774745,
343
+ "grad_norm": 4.066005518388316e-11,
344
+ "learning_rate": 2.955773606370876e-07,
345
+ "loss": 0.1118,
346
  "step": 11500
347
  },
348
  {
349
+ "epoch": 1.2265358361774745,
350
+ "eval_loss": 0.5491370558738708,
351
+ "eval_runtime": 63.7293,
352
+ "eval_samples_per_second": 15.487,
353
+ "eval_steps_per_second": 1.946,
354
  "step": 11500
355
  },
356
  {
357
+ "epoch": 1.2798634812286689,
358
+ "grad_norm": 7.519358769059181e-05,
359
+ "learning_rate": 2.8668941979522184e-07,
360
+ "loss": 0.0534,
361
  "step": 12000
362
  },
363
  {
364
+ "epoch": 1.2798634812286689,
365
+ "eval_loss": 0.5478764772415161,
366
+ "eval_runtime": 62.7547,
367
+ "eval_samples_per_second": 15.728,
368
+ "eval_steps_per_second": 1.976,
369
  "step": 12000
370
  },
371
  {
372
+ "epoch": 1.3331911262798635,
373
+ "grad_norm": 8.400617446113756e-08,
374
+ "learning_rate": 2.778014789533561e-07,
375
+ "loss": 0.0507,
376
  "step": 12500
377
  },
378
  {
379
+ "epoch": 1.3331911262798635,
380
+ "eval_loss": 0.576600968837738,
381
+ "eval_runtime": 64.301,
382
+ "eval_samples_per_second": 15.35,
383
+ "eval_steps_per_second": 1.928,
384
  "step": 12500
385
  },
386
  {
387
+ "epoch": 1.386518771331058,
388
+ "grad_norm": 2.5385464130084356e-12,
389
+ "learning_rate": 2.689135381114903e-07,
390
+ "loss": 0.0784,
391
  "step": 13000
392
  },
393
  {
394
+ "epoch": 1.386518771331058,
395
+ "eval_loss": 0.5729931592941284,
396
+ "eval_runtime": 63.652,
397
+ "eval_samples_per_second": 15.506,
398
+ "eval_steps_per_second": 1.948,
399
  "step": 13000
400
  },
401
  {
402
+ "epoch": 1.4398464163822526,
403
+ "grad_norm": 0.08409273624420166,
404
+ "learning_rate": 2.600255972696246e-07,
405
+ "loss": 0.079,
406
  "step": 13500
407
  },
408
  {
409
+ "epoch": 1.4398464163822526,
410
+ "eval_loss": 0.5694039463996887,
411
+ "eval_runtime": 62.8247,
412
+ "eval_samples_per_second": 15.71,
413
+ "eval_steps_per_second": 1.974,
414
  "step": 13500
415
  },
416
  {
417
+ "epoch": 1.493174061433447,
418
+ "grad_norm": 3.7334253022436314e-08,
419
+ "learning_rate": 2.511376564277588e-07,
420
+ "loss": 0.0507,
421
  "step": 14000
422
  },
423
  {
424
+ "epoch": 1.493174061433447,
425
+ "eval_loss": 0.5713907480239868,
426
+ "eval_runtime": 62.3577,
427
+ "eval_samples_per_second": 15.828,
428
+ "eval_steps_per_second": 1.989,
429
  "step": 14000
430
  },
431
  {
432
+ "epoch": 1.5465017064846416,
433
+ "grad_norm": 1.5219894647598267,
434
+ "learning_rate": 2.422497155858931e-07,
435
+ "loss": 0.0774,
436
  "step": 14500
437
  },
438
  {
439
+ "epoch": 1.5465017064846416,
440
+ "eval_loss": 0.5674533843994141,
441
+ "eval_runtime": 62.3753,
442
+ "eval_samples_per_second": 15.824,
443
+ "eval_steps_per_second": 1.988,
444
  "step": 14500
445
  },
446
  {
447
+ "epoch": 1.5998293515358362,
448
+ "grad_norm": 879.1193237304688,
449
+ "learning_rate": 2.333617747440273e-07,
450
+ "loss": 0.0376,
451
  "step": 15000
452
  },
453
  {
454
+ "epoch": 1.5998293515358362,
455
+ "eval_loss": 0.561253547668457,
456
+ "eval_runtime": 62.4788,
457
+ "eval_samples_per_second": 15.797,
458
+ "eval_steps_per_second": 1.985,
459
  "step": 15000
460
  },
461
  {
462
+ "epoch": 1.6531569965870307,
463
+ "grad_norm": 0.0006221308722160757,
464
+ "learning_rate": 2.2447383390216154e-07,
465
+ "loss": 0.0929,
466
  "step": 15500
467
  },
468
  {
469
+ "epoch": 1.6531569965870307,
470
+ "eval_loss": 0.5395999550819397,
471
+ "eval_runtime": 62.5651,
472
+ "eval_samples_per_second": 15.776,
473
+ "eval_steps_per_second": 1.982,
474
  "step": 15500
475
  },
476
  {
477
+ "epoch": 1.7064846416382253,
478
+ "grad_norm": 37.85109329223633,
479
+ "learning_rate": 2.155858930602958e-07,
480
+ "loss": 0.0442,
481
  "step": 16000
482
  },
483
  {
484
+ "epoch": 1.7064846416382253,
485
+ "eval_loss": 0.5307884216308594,
486
+ "eval_runtime": 62.4681,
487
+ "eval_samples_per_second": 15.8,
488
+ "eval_steps_per_second": 1.985,
489
  "step": 16000
490
  },
491
  {
492
+ "epoch": 1.75981228668942,
493
+ "grad_norm": 9.600337885773413e-11,
494
+ "learning_rate": 2.0669795221843002e-07,
495
+ "loss": 0.0593,
496
  "step": 16500
497
  },
498
  {
499
+ "epoch": 1.75981228668942,
500
+ "eval_loss": 0.5230685472488403,
501
+ "eval_runtime": 63.8124,
502
+ "eval_samples_per_second": 15.467,
503
+ "eval_steps_per_second": 1.943,
504
  "step": 16500
505
  },
506
  {
507
+ "epoch": 1.8131399317406145,
508
+ "grad_norm": 9.067448940129619e-13,
509
+ "learning_rate": 1.9781001137656428e-07,
510
+ "loss": 0.0578,
511
  "step": 17000
512
  },
513
  {
514
+ "epoch": 1.8131399317406145,
515
+ "eval_loss": 0.5200281739234924,
516
+ "eval_runtime": 62.3767,
517
+ "eval_samples_per_second": 15.823,
518
+ "eval_steps_per_second": 1.988,
519
  "step": 17000
520
  },
521
  {
522
+ "epoch": 1.8664675767918089,
523
+ "grad_norm": 0.00843018852174282,
524
+ "learning_rate": 1.889220705346985e-07,
525
+ "loss": 0.0928,
526
  "step": 17500
527
  },
528
  {
529
+ "epoch": 1.8664675767918089,
530
+ "eval_loss": 0.509168803691864,
531
+ "eval_runtime": 62.4108,
532
+ "eval_samples_per_second": 15.815,
533
+ "eval_steps_per_second": 1.987,
534
  "step": 17500
535
  },
536
  {
537
+ "epoch": 1.9197952218430034,
538
+ "grad_norm": 4.264631314754297e-09,
539
+ "learning_rate": 1.8003412969283277e-07,
540
+ "loss": 0.0633,
541
  "step": 18000
542
  },
543
  {
544
+ "epoch": 1.9197952218430034,
545
+ "eval_loss": 0.4954444468021393,
546
+ "eval_runtime": 63.3639,
547
+ "eval_samples_per_second": 15.577,
548
+ "eval_steps_per_second": 1.957,
549
  "step": 18000
550
  },
551
  {
552
+ "epoch": 1.9731228668941978,
553
+ "grad_norm": 459.5137939453125,
554
+ "learning_rate": 1.7114618885096697e-07,
555
+ "loss": 0.067,
556
  "step": 18500
557
  },
558
  {
559
+ "epoch": 1.9731228668941978,
560
+ "eval_loss": 0.49844062328338623,
561
+ "eval_runtime": 63.6573,
562
+ "eval_samples_per_second": 15.505,
563
+ "eval_steps_per_second": 1.948,
564
  "step": 18500
565
  },
566
  {
567
+ "epoch": 2.0264505119453924,
568
+ "grad_norm": 0.27590689063072205,
569
+ "learning_rate": 1.6225824800910123e-07,
570
+ "loss": 0.0525,
571
  "step": 19000
572
  },
573
  {
574
+ "epoch": 2.0264505119453924,
575
+ "eval_loss": 0.4999145269393921,
576
+ "eval_runtime": 63.1727,
577
+ "eval_samples_per_second": 15.624,
578
+ "eval_steps_per_second": 1.963,
579
  "step": 19000
580
  },
581
  {
582
+ "epoch": 2.079778156996587,
583
+ "grad_norm": 1.349623744317796e-06,
584
+ "learning_rate": 1.533703071672355e-07,
585
+ "loss": 0.0301,
586
  "step": 19500
587
  },
588
  {
589
+ "epoch": 2.079778156996587,
590
+ "eval_loss": 0.4926412105560303,
591
+ "eval_runtime": 63.4421,
592
+ "eval_samples_per_second": 15.557,
593
+ "eval_steps_per_second": 1.955,
594
  "step": 19500
595
  }
596
  ],
597
  "logging_steps": 500,
598
+ "max_steps": 28128,
599
  "num_input_tokens_seen": 0,
600
+ "num_train_epochs": 3,
601
  "save_steps": 500,
602
  "stateful_callbacks": {
603
  "TrainerControl": {
 
611
  "attributes": {}
612
  }
613
  },
614
+ "total_flos": 7016007366272520.0,
615
  "train_batch_size": 2,
616
  "trial_name": null,
617
  "trial_params": null
checkpoint-19500/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:319be2a769ef7117b6826908976f080f9082f159f4f0ad75da1a151ad29d9bff
3
  size 4847
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:057f6bbf1ff3e34abda0863071bbf5fdcc7edb1102f04dba05c7a15087bc3396
3
  size 4847