Upload trainer_state.json

#1
Files changed (1) hide show
  1. trainer_state.json +599 -0
trainer_state.json ADDED
@@ -0,0 +1,599 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 3.3522434244455903,
3
+ "best_model_checkpoint": "./whisper-small/checkpoint-2000",
4
+ "epoch": 6.472491909385114,
5
+ "eval_steps": 10,
6
+ "global_step": 2000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.08,
13
+ "grad_norm": 81.5032730102539,
14
+ "learning_rate": 4.2000000000000006e-07,
15
+ "loss": 2.9603,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.16,
20
+ "grad_norm": 17.11170768737793,
21
+ "learning_rate": 9.200000000000001e-07,
22
+ "loss": 2.144,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.24,
27
+ "grad_norm": 11.967795372009277,
28
+ "learning_rate": 1.42e-06,
29
+ "loss": 1.3446,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 0.32,
34
+ "grad_norm": 8.235530853271484,
35
+ "learning_rate": 1.9200000000000003e-06,
36
+ "loss": 0.7532,
37
+ "step": 100
38
+ },
39
+ {
40
+ "epoch": 0.4,
41
+ "grad_norm": 6.933337688446045,
42
+ "learning_rate": 2.42e-06,
43
+ "loss": 0.5681,
44
+ "step": 125
45
+ },
46
+ {
47
+ "epoch": 0.49,
48
+ "grad_norm": 8.942669868469238,
49
+ "learning_rate": 2.92e-06,
50
+ "loss": 0.496,
51
+ "step": 150
52
+ },
53
+ {
54
+ "epoch": 0.57,
55
+ "grad_norm": 11.463906288146973,
56
+ "learning_rate": 3.4200000000000007e-06,
57
+ "loss": 0.4111,
58
+ "step": 175
59
+ },
60
+ {
61
+ "epoch": 0.65,
62
+ "grad_norm": 7.532162666320801,
63
+ "learning_rate": 3.920000000000001e-06,
64
+ "loss": 0.3332,
65
+ "step": 200
66
+ },
67
+ {
68
+ "epoch": 0.73,
69
+ "grad_norm": 7.5595197677612305,
70
+ "learning_rate": 4.42e-06,
71
+ "loss": 0.2728,
72
+ "step": 225
73
+ },
74
+ {
75
+ "epoch": 0.81,
76
+ "grad_norm": 4.813773155212402,
77
+ "learning_rate": 4.92e-06,
78
+ "loss": 0.1087,
79
+ "step": 250
80
+ },
81
+ {
82
+ "epoch": 0.89,
83
+ "grad_norm": 3.7373056411743164,
84
+ "learning_rate": 5.420000000000001e-06,
85
+ "loss": 0.0653,
86
+ "step": 275
87
+ },
88
+ {
89
+ "epoch": 0.97,
90
+ "grad_norm": 4.375819683074951,
91
+ "learning_rate": 5.92e-06,
92
+ "loss": 0.059,
93
+ "step": 300
94
+ },
95
+ {
96
+ "epoch": 1.05,
97
+ "grad_norm": 1.661054015159607,
98
+ "learning_rate": 6.42e-06,
99
+ "loss": 0.0381,
100
+ "step": 325
101
+ },
102
+ {
103
+ "epoch": 1.13,
104
+ "grad_norm": 2.9067795276641846,
105
+ "learning_rate": 6.92e-06,
106
+ "loss": 0.0327,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 1.21,
111
+ "grad_norm": 4.853701114654541,
112
+ "learning_rate": 7.420000000000001e-06,
113
+ "loss": 0.0295,
114
+ "step": 375
115
+ },
116
+ {
117
+ "epoch": 1.29,
118
+ "grad_norm": 1.736449122428894,
119
+ "learning_rate": 7.92e-06,
120
+ "loss": 0.0311,
121
+ "step": 400
122
+ },
123
+ {
124
+ "epoch": 1.38,
125
+ "grad_norm": 3.658984899520874,
126
+ "learning_rate": 8.42e-06,
127
+ "loss": 0.0296,
128
+ "step": 425
129
+ },
130
+ {
131
+ "epoch": 1.46,
132
+ "grad_norm": 2.040172576904297,
133
+ "learning_rate": 8.920000000000001e-06,
134
+ "loss": 0.0221,
135
+ "step": 450
136
+ },
137
+ {
138
+ "epoch": 1.54,
139
+ "grad_norm": 2.4981894493103027,
140
+ "learning_rate": 9.42e-06,
141
+ "loss": 0.0251,
142
+ "step": 475
143
+ },
144
+ {
145
+ "epoch": 1.62,
146
+ "grad_norm": 2.639390230178833,
147
+ "learning_rate": 9.920000000000002e-06,
148
+ "loss": 0.0261,
149
+ "step": 500
150
+ },
151
+ {
152
+ "epoch": 1.7,
153
+ "grad_norm": 2.526102066040039,
154
+ "learning_rate": 9.940000000000001e-06,
155
+ "loss": 0.0253,
156
+ "step": 525
157
+ },
158
+ {
159
+ "epoch": 1.78,
160
+ "grad_norm": 1.3722740411758423,
161
+ "learning_rate": 9.86857142857143e-06,
162
+ "loss": 0.0203,
163
+ "step": 550
164
+ },
165
+ {
166
+ "epoch": 1.86,
167
+ "grad_norm": 2.3701043128967285,
168
+ "learning_rate": 9.797142857142858e-06,
169
+ "loss": 0.0225,
170
+ "step": 575
171
+ },
172
+ {
173
+ "epoch": 1.94,
174
+ "grad_norm": 2.8724324703216553,
175
+ "learning_rate": 9.725714285714287e-06,
176
+ "loss": 0.0201,
177
+ "step": 600
178
+ },
179
+ {
180
+ "epoch": 2.02,
181
+ "grad_norm": 1.0449714660644531,
182
+ "learning_rate": 9.654285714285716e-06,
183
+ "loss": 0.0149,
184
+ "step": 625
185
+ },
186
+ {
187
+ "epoch": 2.1,
188
+ "grad_norm": 5.038928031921387,
189
+ "learning_rate": 9.582857142857143e-06,
190
+ "loss": 0.0104,
191
+ "step": 650
192
+ },
193
+ {
194
+ "epoch": 2.18,
195
+ "grad_norm": 1.6107451915740967,
196
+ "learning_rate": 9.511428571428572e-06,
197
+ "loss": 0.0086,
198
+ "step": 675
199
+ },
200
+ {
201
+ "epoch": 2.27,
202
+ "grad_norm": 1.3224350214004517,
203
+ "learning_rate": 9.440000000000001e-06,
204
+ "loss": 0.006,
205
+ "step": 700
206
+ },
207
+ {
208
+ "epoch": 2.35,
209
+ "grad_norm": 0.14248408377170563,
210
+ "learning_rate": 9.368571428571428e-06,
211
+ "loss": 0.0059,
212
+ "step": 725
213
+ },
214
+ {
215
+ "epoch": 2.43,
216
+ "grad_norm": 2.3801496028900146,
217
+ "learning_rate": 9.297142857142857e-06,
218
+ "loss": 0.0056,
219
+ "step": 750
220
+ },
221
+ {
222
+ "epoch": 2.51,
223
+ "grad_norm": 1.156813621520996,
224
+ "learning_rate": 9.225714285714286e-06,
225
+ "loss": 0.0067,
226
+ "step": 775
227
+ },
228
+ {
229
+ "epoch": 2.59,
230
+ "grad_norm": 0.9941128492355347,
231
+ "learning_rate": 9.154285714285715e-06,
232
+ "loss": 0.0076,
233
+ "step": 800
234
+ },
235
+ {
236
+ "epoch": 2.67,
237
+ "grad_norm": 1.4023518562316895,
238
+ "learning_rate": 9.082857142857143e-06,
239
+ "loss": 0.006,
240
+ "step": 825
241
+ },
242
+ {
243
+ "epoch": 2.75,
244
+ "grad_norm": 1.1814390420913696,
245
+ "learning_rate": 9.011428571428572e-06,
246
+ "loss": 0.0053,
247
+ "step": 850
248
+ },
249
+ {
250
+ "epoch": 2.83,
251
+ "grad_norm": 1.8819189071655273,
252
+ "learning_rate": 8.94e-06,
253
+ "loss": 0.008,
254
+ "step": 875
255
+ },
256
+ {
257
+ "epoch": 2.91,
258
+ "grad_norm": 0.9754526019096375,
259
+ "learning_rate": 8.86857142857143e-06,
260
+ "loss": 0.0047,
261
+ "step": 900
262
+ },
263
+ {
264
+ "epoch": 2.99,
265
+ "grad_norm": 0.40376096963882446,
266
+ "learning_rate": 8.797142857142857e-06,
267
+ "loss": 0.0062,
268
+ "step": 925
269
+ },
270
+ {
271
+ "epoch": 3.07,
272
+ "grad_norm": 0.07103849202394485,
273
+ "learning_rate": 8.725714285714286e-06,
274
+ "loss": 0.0028,
275
+ "step": 950
276
+ },
277
+ {
278
+ "epoch": 3.16,
279
+ "grad_norm": 0.5621446967124939,
280
+ "learning_rate": 8.654285714285715e-06,
281
+ "loss": 0.0025,
282
+ "step": 975
283
+ },
284
+ {
285
+ "epoch": 3.24,
286
+ "grad_norm": 0.506003201007843,
287
+ "learning_rate": 8.582857142857144e-06,
288
+ "loss": 0.0026,
289
+ "step": 1000
290
+ },
291
+ {
292
+ "epoch": 3.24,
293
+ "eval_loss": 0.020510252565145493,
294
+ "eval_runtime": 791.6891,
295
+ "eval_samples_per_second": 1.557,
296
+ "eval_steps_per_second": 0.196,
297
+ "eval_wer": 4.486848891181021,
298
+ "step": 1000
299
+ },
300
+ {
301
+ "epoch": 3.32,
302
+ "grad_norm": 0.723780632019043,
303
+ "learning_rate": 8.511428571428571e-06,
304
+ "loss": 0.003,
305
+ "step": 1025
306
+ },
307
+ {
308
+ "epoch": 3.4,
309
+ "grad_norm": 0.02721921168267727,
310
+ "learning_rate": 8.44e-06,
311
+ "loss": 0.0028,
312
+ "step": 1050
313
+ },
314
+ {
315
+ "epoch": 3.48,
316
+ "grad_norm": 1.0563775300979614,
317
+ "learning_rate": 8.36857142857143e-06,
318
+ "loss": 0.002,
319
+ "step": 1075
320
+ },
321
+ {
322
+ "epoch": 3.56,
323
+ "grad_norm": 1.54115629196167,
324
+ "learning_rate": 8.297142857142859e-06,
325
+ "loss": 0.0027,
326
+ "step": 1100
327
+ },
328
+ {
329
+ "epoch": 3.64,
330
+ "grad_norm": 0.017975807189941406,
331
+ "learning_rate": 8.225714285714288e-06,
332
+ "loss": 0.0016,
333
+ "step": 1125
334
+ },
335
+ {
336
+ "epoch": 3.72,
337
+ "grad_norm": 0.15264229476451874,
338
+ "learning_rate": 8.154285714285715e-06,
339
+ "loss": 0.0034,
340
+ "step": 1150
341
+ },
342
+ {
343
+ "epoch": 3.8,
344
+ "grad_norm": 0.655139148235321,
345
+ "learning_rate": 8.082857142857144e-06,
346
+ "loss": 0.0025,
347
+ "step": 1175
348
+ },
349
+ {
350
+ "epoch": 3.88,
351
+ "grad_norm": 0.9618725776672363,
352
+ "learning_rate": 8.011428571428573e-06,
353
+ "loss": 0.0017,
354
+ "step": 1200
355
+ },
356
+ {
357
+ "epoch": 3.96,
358
+ "grad_norm": 3.206550121307373,
359
+ "learning_rate": 7.94e-06,
360
+ "loss": 0.0031,
361
+ "step": 1225
362
+ },
363
+ {
364
+ "epoch": 4.05,
365
+ "grad_norm": 0.41130420565605164,
366
+ "learning_rate": 7.86857142857143e-06,
367
+ "loss": 0.0012,
368
+ "step": 1250
369
+ },
370
+ {
371
+ "epoch": 4.13,
372
+ "grad_norm": 0.0929947942495346,
373
+ "learning_rate": 7.797142857142858e-06,
374
+ "loss": 0.0012,
375
+ "step": 1275
376
+ },
377
+ {
378
+ "epoch": 4.21,
379
+ "grad_norm": 0.050773363560438156,
380
+ "learning_rate": 7.725714285714286e-06,
381
+ "loss": 0.0005,
382
+ "step": 1300
383
+ },
384
+ {
385
+ "epoch": 4.29,
386
+ "grad_norm": 0.0827890932559967,
387
+ "learning_rate": 7.654285714285715e-06,
388
+ "loss": 0.0005,
389
+ "step": 1325
390
+ },
391
+ {
392
+ "epoch": 4.37,
393
+ "grad_norm": 0.02210867777466774,
394
+ "learning_rate": 7.5828571428571444e-06,
395
+ "loss": 0.0007,
396
+ "step": 1350
397
+ },
398
+ {
399
+ "epoch": 4.45,
400
+ "grad_norm": 0.014925210736691952,
401
+ "learning_rate": 7.511428571428572e-06,
402
+ "loss": 0.0012,
403
+ "step": 1375
404
+ },
405
+ {
406
+ "epoch": 4.53,
407
+ "grad_norm": 0.027211977168917656,
408
+ "learning_rate": 7.440000000000001e-06,
409
+ "loss": 0.0009,
410
+ "step": 1400
411
+ },
412
+ {
413
+ "epoch": 4.61,
414
+ "grad_norm": 1.4061143398284912,
415
+ "learning_rate": 7.36857142857143e-06,
416
+ "loss": 0.0008,
417
+ "step": 1425
418
+ },
419
+ {
420
+ "epoch": 4.69,
421
+ "grad_norm": 0.1997688263654709,
422
+ "learning_rate": 7.297142857142858e-06,
423
+ "loss": 0.0013,
424
+ "step": 1450
425
+ },
426
+ {
427
+ "epoch": 4.77,
428
+ "grad_norm": 0.011363201774656773,
429
+ "learning_rate": 7.225714285714286e-06,
430
+ "loss": 0.0005,
431
+ "step": 1475
432
+ },
433
+ {
434
+ "epoch": 4.85,
435
+ "grad_norm": 0.020546328276395798,
436
+ "learning_rate": 7.154285714285715e-06,
437
+ "loss": 0.0004,
438
+ "step": 1500
439
+ },
440
+ {
441
+ "epoch": 4.94,
442
+ "grad_norm": 0.026502510532736778,
443
+ "learning_rate": 7.082857142857143e-06,
444
+ "loss": 0.0007,
445
+ "step": 1525
446
+ },
447
+ {
448
+ "epoch": 5.02,
449
+ "grad_norm": 0.1575097292661667,
450
+ "learning_rate": 7.011428571428572e-06,
451
+ "loss": 0.0009,
452
+ "step": 1550
453
+ },
454
+ {
455
+ "epoch": 5.1,
456
+ "grad_norm": 0.89250248670578,
457
+ "learning_rate": 6.9400000000000005e-06,
458
+ "loss": 0.001,
459
+ "step": 1575
460
+ },
461
+ {
462
+ "epoch": 5.18,
463
+ "grad_norm": 0.5113543272018433,
464
+ "learning_rate": 6.868571428571429e-06,
465
+ "loss": 0.0003,
466
+ "step": 1600
467
+ },
468
+ {
469
+ "epoch": 5.26,
470
+ "grad_norm": 0.05453397333621979,
471
+ "learning_rate": 6.797142857142858e-06,
472
+ "loss": 0.0003,
473
+ "step": 1625
474
+ },
475
+ {
476
+ "epoch": 5.34,
477
+ "grad_norm": 0.01590396836400032,
478
+ "learning_rate": 6.725714285714287e-06,
479
+ "loss": 0.0002,
480
+ "step": 1650
481
+ },
482
+ {
483
+ "epoch": 5.42,
484
+ "grad_norm": 0.00851709395647049,
485
+ "learning_rate": 6.654285714285716e-06,
486
+ "loss": 0.0005,
487
+ "step": 1675
488
+ },
489
+ {
490
+ "epoch": 5.5,
491
+ "grad_norm": 0.014185091480612755,
492
+ "learning_rate": 6.582857142857143e-06,
493
+ "loss": 0.0004,
494
+ "step": 1700
495
+ },
496
+ {
497
+ "epoch": 5.58,
498
+ "grad_norm": 0.017834780737757683,
499
+ "learning_rate": 6.511428571428572e-06,
500
+ "loss": 0.0003,
501
+ "step": 1725
502
+ },
503
+ {
504
+ "epoch": 5.66,
505
+ "grad_norm": 0.017684083431959152,
506
+ "learning_rate": 6.440000000000001e-06,
507
+ "loss": 0.0003,
508
+ "step": 1750
509
+ },
510
+ {
511
+ "epoch": 5.74,
512
+ "grad_norm": 0.009825997054576874,
513
+ "learning_rate": 6.368571428571429e-06,
514
+ "loss": 0.0001,
515
+ "step": 1775
516
+ },
517
+ {
518
+ "epoch": 5.83,
519
+ "grad_norm": 0.08621440827846527,
520
+ "learning_rate": 6.297142857142857e-06,
521
+ "loss": 0.0004,
522
+ "step": 1800
523
+ },
524
+ {
525
+ "epoch": 5.91,
526
+ "grad_norm": 0.019367387518286705,
527
+ "learning_rate": 6.225714285714286e-06,
528
+ "loss": 0.0007,
529
+ "step": 1825
530
+ },
531
+ {
532
+ "epoch": 5.99,
533
+ "grad_norm": 0.04589274898171425,
534
+ "learning_rate": 6.1542857142857145e-06,
535
+ "loss": 0.0003,
536
+ "step": 1850
537
+ },
538
+ {
539
+ "epoch": 6.07,
540
+ "grad_norm": 0.010185680352151394,
541
+ "learning_rate": 6.0828571428571435e-06,
542
+ "loss": 0.0002,
543
+ "step": 1875
544
+ },
545
+ {
546
+ "epoch": 6.15,
547
+ "grad_norm": 0.004478767979890108,
548
+ "learning_rate": 6.011428571428572e-06,
549
+ "loss": 0.0002,
550
+ "step": 1900
551
+ },
552
+ {
553
+ "epoch": 6.23,
554
+ "grad_norm": 0.03858327120542526,
555
+ "learning_rate": 5.94e-06,
556
+ "loss": 0.0001,
557
+ "step": 1925
558
+ },
559
+ {
560
+ "epoch": 6.31,
561
+ "grad_norm": 0.008098349906504154,
562
+ "learning_rate": 5.868571428571429e-06,
563
+ "loss": 0.0002,
564
+ "step": 1950
565
+ },
566
+ {
567
+ "epoch": 6.39,
568
+ "grad_norm": 0.0035556757356971502,
569
+ "learning_rate": 5.797142857142858e-06,
570
+ "loss": 0.0001,
571
+ "step": 1975
572
+ },
573
+ {
574
+ "epoch": 6.47,
575
+ "grad_norm": 0.00917895045131445,
576
+ "learning_rate": 5.725714285714287e-06,
577
+ "loss": 0.0003,
578
+ "step": 2000
579
+ },
580
+ {
581
+ "epoch": 6.47,
582
+ "eval_loss": 0.01798020303249359,
583
+ "eval_runtime": 807.9427,
584
+ "eval_samples_per_second": 1.526,
585
+ "eval_steps_per_second": 0.192,
586
+ "eval_wer": 3.3522434244455903,
587
+ "step": 2000
588
+ }
589
+ ],
590
+ "logging_steps": 25,
591
+ "max_steps": 4000,
592
+ "num_input_tokens_seen": 0,
593
+ "num_train_epochs": 13,
594
+ "save_steps": 10,
595
+ "total_flos": 9.21395466141696e+18,
596
+ "train_batch_size": 16,
597
+ "trial_name": null,
598
+ "trial_params": null
599
+ }