Makhmud commited on
Commit
41d135f
1 Parent(s): d356eb7

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +1017 -0
trainer_state.json ADDED
@@ -0,0 +1,1017 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 34.928488014365186,
3
+ "best_model_checkpoint": "./whisper-small-uz/checkpoint-4000",
4
+ "epoch": 2.1569156106767324,
5
+ "eval_steps": 1000,
6
+ "global_step": 4000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 5.000000000000001e-07,
14
+ "loss": 5.8743,
15
+ "step": 25
16
+ },
17
+ {
18
+ "epoch": 0.03,
19
+ "learning_rate": 1.0000000000000002e-06,
20
+ "loss": 4.5217,
21
+ "step": 50
22
+ },
23
+ {
24
+ "epoch": 0.04,
25
+ "learning_rate": 1.5e-06,
26
+ "loss": 3.1166,
27
+ "step": 75
28
+ },
29
+ {
30
+ "epoch": 0.05,
31
+ "learning_rate": 2.0000000000000003e-06,
32
+ "loss": 2.3201,
33
+ "step": 100
34
+ },
35
+ {
36
+ "epoch": 0.07,
37
+ "learning_rate": 2.5e-06,
38
+ "loss": 2.0321,
39
+ "step": 125
40
+ },
41
+ {
42
+ "epoch": 0.08,
43
+ "learning_rate": 3e-06,
44
+ "loss": 1.8315,
45
+ "step": 150
46
+ },
47
+ {
48
+ "epoch": 0.09,
49
+ "learning_rate": 3.5e-06,
50
+ "loss": 1.6767,
51
+ "step": 175
52
+ },
53
+ {
54
+ "epoch": 0.11,
55
+ "learning_rate": 4.000000000000001e-06,
56
+ "loss": 1.5361,
57
+ "step": 200
58
+ },
59
+ {
60
+ "epoch": 0.12,
61
+ "learning_rate": 4.5e-06,
62
+ "loss": 1.4069,
63
+ "step": 225
64
+ },
65
+ {
66
+ "epoch": 0.13,
67
+ "learning_rate": 5e-06,
68
+ "loss": 1.3365,
69
+ "step": 250
70
+ },
71
+ {
72
+ "epoch": 0.15,
73
+ "learning_rate": 5.500000000000001e-06,
74
+ "loss": 1.2859,
75
+ "step": 275
76
+ },
77
+ {
78
+ "epoch": 0.16,
79
+ "learning_rate": 6e-06,
80
+ "loss": 1.2073,
81
+ "step": 300
82
+ },
83
+ {
84
+ "epoch": 0.18,
85
+ "learning_rate": 6.5000000000000004e-06,
86
+ "loss": 1.1188,
87
+ "step": 325
88
+ },
89
+ {
90
+ "epoch": 0.19,
91
+ "learning_rate": 7e-06,
92
+ "loss": 1.0234,
93
+ "step": 350
94
+ },
95
+ {
96
+ "epoch": 0.2,
97
+ "learning_rate": 7.500000000000001e-06,
98
+ "loss": 0.8149,
99
+ "step": 375
100
+ },
101
+ {
102
+ "epoch": 0.22,
103
+ "learning_rate": 8.000000000000001e-06,
104
+ "loss": 0.773,
105
+ "step": 400
106
+ },
107
+ {
108
+ "epoch": 0.23,
109
+ "learning_rate": 8.5e-06,
110
+ "loss": 0.7559,
111
+ "step": 425
112
+ },
113
+ {
114
+ "epoch": 0.24,
115
+ "learning_rate": 9e-06,
116
+ "loss": 0.7063,
117
+ "step": 450
118
+ },
119
+ {
120
+ "epoch": 0.26,
121
+ "learning_rate": 9.5e-06,
122
+ "loss": 0.7092,
123
+ "step": 475
124
+ },
125
+ {
126
+ "epoch": 0.27,
127
+ "learning_rate": 1e-05,
128
+ "loss": 0.6875,
129
+ "step": 500
130
+ },
131
+ {
132
+ "epoch": 0.28,
133
+ "learning_rate": 9.92857142857143e-06,
134
+ "loss": 0.6551,
135
+ "step": 525
136
+ },
137
+ {
138
+ "epoch": 0.3,
139
+ "learning_rate": 9.857142857142859e-06,
140
+ "loss": 0.6519,
141
+ "step": 550
142
+ },
143
+ {
144
+ "epoch": 0.31,
145
+ "learning_rate": 9.785714285714286e-06,
146
+ "loss": 0.6382,
147
+ "step": 575
148
+ },
149
+ {
150
+ "epoch": 0.32,
151
+ "learning_rate": 9.714285714285715e-06,
152
+ "loss": 0.6039,
153
+ "step": 600
154
+ },
155
+ {
156
+ "epoch": 0.34,
157
+ "learning_rate": 9.642857142857144e-06,
158
+ "loss": 0.6033,
159
+ "step": 625
160
+ },
161
+ {
162
+ "epoch": 0.35,
163
+ "learning_rate": 9.571428571428573e-06,
164
+ "loss": 0.5866,
165
+ "step": 650
166
+ },
167
+ {
168
+ "epoch": 0.36,
169
+ "learning_rate": 9.5e-06,
170
+ "loss": 0.603,
171
+ "step": 675
172
+ },
173
+ {
174
+ "epoch": 0.38,
175
+ "learning_rate": 9.42857142857143e-06,
176
+ "loss": 0.5935,
177
+ "step": 700
178
+ },
179
+ {
180
+ "epoch": 0.39,
181
+ "learning_rate": 9.357142857142859e-06,
182
+ "loss": 0.557,
183
+ "step": 725
184
+ },
185
+ {
186
+ "epoch": 0.4,
187
+ "learning_rate": 9.285714285714288e-06,
188
+ "loss": 0.5639,
189
+ "step": 750
190
+ },
191
+ {
192
+ "epoch": 0.42,
193
+ "learning_rate": 9.214285714285715e-06,
194
+ "loss": 0.5286,
195
+ "step": 775
196
+ },
197
+ {
198
+ "epoch": 0.43,
199
+ "learning_rate": 9.142857142857144e-06,
200
+ "loss": 0.538,
201
+ "step": 800
202
+ },
203
+ {
204
+ "epoch": 0.44,
205
+ "learning_rate": 9.071428571428573e-06,
206
+ "loss": 0.4961,
207
+ "step": 825
208
+ },
209
+ {
210
+ "epoch": 0.46,
211
+ "learning_rate": 9e-06,
212
+ "loss": 0.5278,
213
+ "step": 850
214
+ },
215
+ {
216
+ "epoch": 0.47,
217
+ "learning_rate": 8.92857142857143e-06,
218
+ "loss": 0.5256,
219
+ "step": 875
220
+ },
221
+ {
222
+ "epoch": 0.49,
223
+ "learning_rate": 8.857142857142858e-06,
224
+ "loss": 0.5184,
225
+ "step": 900
226
+ },
227
+ {
228
+ "epoch": 0.5,
229
+ "learning_rate": 8.785714285714286e-06,
230
+ "loss": 0.4817,
231
+ "step": 925
232
+ },
233
+ {
234
+ "epoch": 0.51,
235
+ "learning_rate": 8.714285714285715e-06,
236
+ "loss": 0.5209,
237
+ "step": 950
238
+ },
239
+ {
240
+ "epoch": 0.53,
241
+ "learning_rate": 8.642857142857144e-06,
242
+ "loss": 0.4935,
243
+ "step": 975
244
+ },
245
+ {
246
+ "epoch": 0.54,
247
+ "learning_rate": 8.571428571428571e-06,
248
+ "loss": 0.4794,
249
+ "step": 1000
250
+ },
251
+ {
252
+ "epoch": 0.54,
253
+ "eval_loss": 0.45038458704948425,
254
+ "eval_runtime": 3123.7684,
255
+ "eval_samples_per_second": 3.934,
256
+ "eval_steps_per_second": 0.492,
257
+ "eval_wer": 42.07215239147632,
258
+ "step": 1000
259
+ },
260
+ {
261
+ "epoch": 0.55,
262
+ "learning_rate": 8.5e-06,
263
+ "loss": 0.4779,
264
+ "step": 1025
265
+ },
266
+ {
267
+ "epoch": 0.57,
268
+ "learning_rate": 8.428571428571429e-06,
269
+ "loss": 0.4896,
270
+ "step": 1050
271
+ },
272
+ {
273
+ "epoch": 0.58,
274
+ "learning_rate": 8.357142857142858e-06,
275
+ "loss": 0.4915,
276
+ "step": 1075
277
+ },
278
+ {
279
+ "epoch": 0.59,
280
+ "learning_rate": 8.285714285714287e-06,
281
+ "loss": 0.4944,
282
+ "step": 1100
283
+ },
284
+ {
285
+ "epoch": 0.61,
286
+ "learning_rate": 8.214285714285714e-06,
287
+ "loss": 0.4802,
288
+ "step": 1125
289
+ },
290
+ {
291
+ "epoch": 0.62,
292
+ "learning_rate": 8.142857142857143e-06,
293
+ "loss": 0.5046,
294
+ "step": 1150
295
+ },
296
+ {
297
+ "epoch": 0.63,
298
+ "learning_rate": 8.071428571428572e-06,
299
+ "loss": 0.4581,
300
+ "step": 1175
301
+ },
302
+ {
303
+ "epoch": 0.65,
304
+ "learning_rate": 8.000000000000001e-06,
305
+ "loss": 0.4465,
306
+ "step": 1200
307
+ },
308
+ {
309
+ "epoch": 0.66,
310
+ "learning_rate": 7.928571428571429e-06,
311
+ "loss": 0.4576,
312
+ "step": 1225
313
+ },
314
+ {
315
+ "epoch": 0.67,
316
+ "learning_rate": 7.857142857142858e-06,
317
+ "loss": 0.4614,
318
+ "step": 1250
319
+ },
320
+ {
321
+ "epoch": 0.69,
322
+ "learning_rate": 7.785714285714287e-06,
323
+ "loss": 0.4498,
324
+ "step": 1275
325
+ },
326
+ {
327
+ "epoch": 0.7,
328
+ "learning_rate": 7.714285714285716e-06,
329
+ "loss": 0.4521,
330
+ "step": 1300
331
+ },
332
+ {
333
+ "epoch": 0.71,
334
+ "learning_rate": 7.642857142857143e-06,
335
+ "loss": 0.4555,
336
+ "step": 1325
337
+ },
338
+ {
339
+ "epoch": 0.73,
340
+ "learning_rate": 7.571428571428572e-06,
341
+ "loss": 0.4856,
342
+ "step": 1350
343
+ },
344
+ {
345
+ "epoch": 0.74,
346
+ "learning_rate": 7.500000000000001e-06,
347
+ "loss": 0.4377,
348
+ "step": 1375
349
+ },
350
+ {
351
+ "epoch": 0.75,
352
+ "learning_rate": 7.428571428571429e-06,
353
+ "loss": 0.4529,
354
+ "step": 1400
355
+ },
356
+ {
357
+ "epoch": 0.77,
358
+ "learning_rate": 7.357142857142858e-06,
359
+ "loss": 0.4282,
360
+ "step": 1425
361
+ },
362
+ {
363
+ "epoch": 0.78,
364
+ "learning_rate": 7.285714285714286e-06,
365
+ "loss": 0.4244,
366
+ "step": 1450
367
+ },
368
+ {
369
+ "epoch": 0.8,
370
+ "learning_rate": 7.2142857142857145e-06,
371
+ "loss": 0.4424,
372
+ "step": 1475
373
+ },
374
+ {
375
+ "epoch": 0.81,
376
+ "learning_rate": 7.1428571428571436e-06,
377
+ "loss": 0.4303,
378
+ "step": 1500
379
+ },
380
+ {
381
+ "epoch": 0.82,
382
+ "learning_rate": 7.0714285714285726e-06,
383
+ "loss": 0.411,
384
+ "step": 1525
385
+ },
386
+ {
387
+ "epoch": 0.84,
388
+ "learning_rate": 7e-06,
389
+ "loss": 0.4003,
390
+ "step": 1550
391
+ },
392
+ {
393
+ "epoch": 0.85,
394
+ "learning_rate": 6.928571428571429e-06,
395
+ "loss": 0.4187,
396
+ "step": 1575
397
+ },
398
+ {
399
+ "epoch": 0.86,
400
+ "learning_rate": 6.857142857142858e-06,
401
+ "loss": 0.4248,
402
+ "step": 1600
403
+ },
404
+ {
405
+ "epoch": 0.88,
406
+ "learning_rate": 6.785714285714287e-06,
407
+ "loss": 0.4264,
408
+ "step": 1625
409
+ },
410
+ {
411
+ "epoch": 0.89,
412
+ "learning_rate": 6.714285714285714e-06,
413
+ "loss": 0.4193,
414
+ "step": 1650
415
+ },
416
+ {
417
+ "epoch": 0.9,
418
+ "learning_rate": 6.642857142857143e-06,
419
+ "loss": 0.425,
420
+ "step": 1675
421
+ },
422
+ {
423
+ "epoch": 0.92,
424
+ "learning_rate": 6.571428571428572e-06,
425
+ "loss": 0.3975,
426
+ "step": 1700
427
+ },
428
+ {
429
+ "epoch": 0.93,
430
+ "learning_rate": 6.5000000000000004e-06,
431
+ "loss": 0.4302,
432
+ "step": 1725
433
+ },
434
+ {
435
+ "epoch": 0.94,
436
+ "learning_rate": 6.4285714285714295e-06,
437
+ "loss": 0.4033,
438
+ "step": 1750
439
+ },
440
+ {
441
+ "epoch": 0.96,
442
+ "learning_rate": 6.357142857142858e-06,
443
+ "loss": 0.3912,
444
+ "step": 1775
445
+ },
446
+ {
447
+ "epoch": 0.97,
448
+ "learning_rate": 6.285714285714286e-06,
449
+ "loss": 0.4112,
450
+ "step": 1800
451
+ },
452
+ {
453
+ "epoch": 0.98,
454
+ "learning_rate": 6.214285714285715e-06,
455
+ "loss": 0.4381,
456
+ "step": 1825
457
+ },
458
+ {
459
+ "epoch": 1.0,
460
+ "learning_rate": 6.142857142857144e-06,
461
+ "loss": 0.4085,
462
+ "step": 1850
463
+ },
464
+ {
465
+ "epoch": 1.01,
466
+ "learning_rate": 6.071428571428571e-06,
467
+ "loss": 0.3286,
468
+ "step": 1875
469
+ },
470
+ {
471
+ "epoch": 1.02,
472
+ "learning_rate": 6e-06,
473
+ "loss": 0.3104,
474
+ "step": 1900
475
+ },
476
+ {
477
+ "epoch": 1.04,
478
+ "learning_rate": 5.928571428571429e-06,
479
+ "loss": 0.3033,
480
+ "step": 1925
481
+ },
482
+ {
483
+ "epoch": 1.05,
484
+ "learning_rate": 5.857142857142858e-06,
485
+ "loss": 0.302,
486
+ "step": 1950
487
+ },
488
+ {
489
+ "epoch": 1.06,
490
+ "learning_rate": 5.785714285714286e-06,
491
+ "loss": 0.3034,
492
+ "step": 1975
493
+ },
494
+ {
495
+ "epoch": 1.08,
496
+ "learning_rate": 5.7142857142857145e-06,
497
+ "loss": 0.313,
498
+ "step": 2000
499
+ },
500
+ {
501
+ "epoch": 1.08,
502
+ "eval_loss": 0.3821106553077698,
503
+ "eval_runtime": 3103.0984,
504
+ "eval_samples_per_second": 3.961,
505
+ "eval_steps_per_second": 0.495,
506
+ "eval_wer": 38.93918655901151,
507
+ "step": 2000
508
+ },
509
+ {
510
+ "epoch": 1.09,
511
+ "learning_rate": 5.6428571428571435e-06,
512
+ "loss": 0.3135,
513
+ "step": 2025
514
+ },
515
+ {
516
+ "epoch": 1.11,
517
+ "learning_rate": 5.571428571428572e-06,
518
+ "loss": 0.3213,
519
+ "step": 2050
520
+ },
521
+ {
522
+ "epoch": 1.12,
523
+ "learning_rate": 5.500000000000001e-06,
524
+ "loss": 0.3173,
525
+ "step": 2075
526
+ },
527
+ {
528
+ "epoch": 1.13,
529
+ "learning_rate": 5.428571428571429e-06,
530
+ "loss": 0.2961,
531
+ "step": 2100
532
+ },
533
+ {
534
+ "epoch": 1.15,
535
+ "learning_rate": 5.357142857142857e-06,
536
+ "loss": 0.3127,
537
+ "step": 2125
538
+ },
539
+ {
540
+ "epoch": 1.16,
541
+ "learning_rate": 5.285714285714286e-06,
542
+ "loss": 0.3217,
543
+ "step": 2150
544
+ },
545
+ {
546
+ "epoch": 1.17,
547
+ "learning_rate": 5.214285714285715e-06,
548
+ "loss": 0.3166,
549
+ "step": 2175
550
+ },
551
+ {
552
+ "epoch": 1.19,
553
+ "learning_rate": 5.142857142857142e-06,
554
+ "loss": 0.3018,
555
+ "step": 2200
556
+ },
557
+ {
558
+ "epoch": 1.2,
559
+ "learning_rate": 5.071428571428571e-06,
560
+ "loss": 0.3085,
561
+ "step": 2225
562
+ },
563
+ {
564
+ "epoch": 1.21,
565
+ "learning_rate": 5e-06,
566
+ "loss": 0.3006,
567
+ "step": 2250
568
+ },
569
+ {
570
+ "epoch": 1.23,
571
+ "learning_rate": 4.928571428571429e-06,
572
+ "loss": 0.2992,
573
+ "step": 2275
574
+ },
575
+ {
576
+ "epoch": 1.24,
577
+ "learning_rate": 4.857142857142858e-06,
578
+ "loss": 0.3049,
579
+ "step": 2300
580
+ },
581
+ {
582
+ "epoch": 1.25,
583
+ "learning_rate": 4.785714285714287e-06,
584
+ "loss": 0.3068,
585
+ "step": 2325
586
+ },
587
+ {
588
+ "epoch": 1.27,
589
+ "learning_rate": 4.714285714285715e-06,
590
+ "loss": 0.2947,
591
+ "step": 2350
592
+ },
593
+ {
594
+ "epoch": 1.28,
595
+ "learning_rate": 4.642857142857144e-06,
596
+ "loss": 0.3066,
597
+ "step": 2375
598
+ },
599
+ {
600
+ "epoch": 1.29,
601
+ "learning_rate": 4.571428571428572e-06,
602
+ "loss": 0.3068,
603
+ "step": 2400
604
+ },
605
+ {
606
+ "epoch": 1.31,
607
+ "learning_rate": 4.5e-06,
608
+ "loss": 0.32,
609
+ "step": 2425
610
+ },
611
+ {
612
+ "epoch": 1.32,
613
+ "learning_rate": 4.428571428571429e-06,
614
+ "loss": 0.2943,
615
+ "step": 2450
616
+ },
617
+ {
618
+ "epoch": 1.33,
619
+ "learning_rate": 4.357142857142857e-06,
620
+ "loss": 0.3142,
621
+ "step": 2475
622
+ },
623
+ {
624
+ "epoch": 1.35,
625
+ "learning_rate": 4.2857142857142855e-06,
626
+ "loss": 0.3189,
627
+ "step": 2500
628
+ },
629
+ {
630
+ "epoch": 1.36,
631
+ "learning_rate": 4.2142857142857145e-06,
632
+ "loss": 0.3004,
633
+ "step": 2525
634
+ },
635
+ {
636
+ "epoch": 1.38,
637
+ "learning_rate": 4.1428571428571435e-06,
638
+ "loss": 0.308,
639
+ "step": 2550
640
+ },
641
+ {
642
+ "epoch": 1.39,
643
+ "learning_rate": 4.071428571428572e-06,
644
+ "loss": 0.3011,
645
+ "step": 2575
646
+ },
647
+ {
648
+ "epoch": 1.4,
649
+ "learning_rate": 4.000000000000001e-06,
650
+ "loss": 0.3101,
651
+ "step": 2600
652
+ },
653
+ {
654
+ "epoch": 1.42,
655
+ "learning_rate": 3.928571428571429e-06,
656
+ "loss": 0.3057,
657
+ "step": 2625
658
+ },
659
+ {
660
+ "epoch": 1.43,
661
+ "learning_rate": 3.857142857142858e-06,
662
+ "loss": 0.323,
663
+ "step": 2650
664
+ },
665
+ {
666
+ "epoch": 1.44,
667
+ "learning_rate": 3.785714285714286e-06,
668
+ "loss": 0.3076,
669
+ "step": 2675
670
+ },
671
+ {
672
+ "epoch": 1.46,
673
+ "learning_rate": 3.7142857142857146e-06,
674
+ "loss": 0.3101,
675
+ "step": 2700
676
+ },
677
+ {
678
+ "epoch": 1.47,
679
+ "learning_rate": 3.642857142857143e-06,
680
+ "loss": 0.2994,
681
+ "step": 2725
682
+ },
683
+ {
684
+ "epoch": 1.48,
685
+ "learning_rate": 3.5714285714285718e-06,
686
+ "loss": 0.3041,
687
+ "step": 2750
688
+ },
689
+ {
690
+ "epoch": 1.5,
691
+ "learning_rate": 3.5e-06,
692
+ "loss": 0.3126,
693
+ "step": 2775
694
+ },
695
+ {
696
+ "epoch": 1.51,
697
+ "learning_rate": 3.428571428571429e-06,
698
+ "loss": 0.3026,
699
+ "step": 2800
700
+ },
701
+ {
702
+ "epoch": 1.52,
703
+ "learning_rate": 3.357142857142857e-06,
704
+ "loss": 0.3005,
705
+ "step": 2825
706
+ },
707
+ {
708
+ "epoch": 1.54,
709
+ "learning_rate": 3.285714285714286e-06,
710
+ "loss": 0.2911,
711
+ "step": 2850
712
+ },
713
+ {
714
+ "epoch": 1.55,
715
+ "learning_rate": 3.2142857142857147e-06,
716
+ "loss": 0.2966,
717
+ "step": 2875
718
+ },
719
+ {
720
+ "epoch": 1.56,
721
+ "learning_rate": 3.142857142857143e-06,
722
+ "loss": 0.3015,
723
+ "step": 2900
724
+ },
725
+ {
726
+ "epoch": 1.58,
727
+ "learning_rate": 3.071428571428572e-06,
728
+ "loss": 0.2913,
729
+ "step": 2925
730
+ },
731
+ {
732
+ "epoch": 1.59,
733
+ "learning_rate": 3e-06,
734
+ "loss": 0.3016,
735
+ "step": 2950
736
+ },
737
+ {
738
+ "epoch": 1.6,
739
+ "learning_rate": 2.928571428571429e-06,
740
+ "loss": 0.2983,
741
+ "step": 2975
742
+ },
743
+ {
744
+ "epoch": 1.62,
745
+ "learning_rate": 2.8571428571428573e-06,
746
+ "loss": 0.2948,
747
+ "step": 3000
748
+ },
749
+ {
750
+ "epoch": 1.62,
751
+ "eval_loss": 0.3468478322029114,
752
+ "eval_runtime": 3105.1004,
753
+ "eval_samples_per_second": 3.958,
754
+ "eval_steps_per_second": 0.495,
755
+ "eval_wer": 35.427000012556974,
756
+ "step": 3000
757
+ },
758
+ {
759
+ "epoch": 1.63,
760
+ "learning_rate": 2.785714285714286e-06,
761
+ "loss": 0.2908,
762
+ "step": 3025
763
+ },
764
+ {
765
+ "epoch": 1.64,
766
+ "learning_rate": 2.7142857142857144e-06,
767
+ "loss": 0.2932,
768
+ "step": 3050
769
+ },
770
+ {
771
+ "epoch": 1.66,
772
+ "learning_rate": 2.642857142857143e-06,
773
+ "loss": 0.2841,
774
+ "step": 3075
775
+ },
776
+ {
777
+ "epoch": 1.67,
778
+ "learning_rate": 2.571428571428571e-06,
779
+ "loss": 0.2967,
780
+ "step": 3100
781
+ },
782
+ {
783
+ "epoch": 1.69,
784
+ "learning_rate": 2.5e-06,
785
+ "loss": 0.292,
786
+ "step": 3125
787
+ },
788
+ {
789
+ "epoch": 1.7,
790
+ "learning_rate": 2.428571428571429e-06,
791
+ "loss": 0.2984,
792
+ "step": 3150
793
+ },
794
+ {
795
+ "epoch": 1.71,
796
+ "learning_rate": 2.3571428571428574e-06,
797
+ "loss": 0.3032,
798
+ "step": 3175
799
+ },
800
+ {
801
+ "epoch": 1.73,
802
+ "learning_rate": 2.285714285714286e-06,
803
+ "loss": 0.2994,
804
+ "step": 3200
805
+ },
806
+ {
807
+ "epoch": 1.74,
808
+ "learning_rate": 2.2142857142857146e-06,
809
+ "loss": 0.3113,
810
+ "step": 3225
811
+ },
812
+ {
813
+ "epoch": 1.75,
814
+ "learning_rate": 2.1428571428571427e-06,
815
+ "loss": 0.2926,
816
+ "step": 3250
817
+ },
818
+ {
819
+ "epoch": 1.77,
820
+ "learning_rate": 2.0714285714285717e-06,
821
+ "loss": 0.2749,
822
+ "step": 3275
823
+ },
824
+ {
825
+ "epoch": 1.78,
826
+ "learning_rate": 2.0000000000000003e-06,
827
+ "loss": 0.2947,
828
+ "step": 3300
829
+ },
830
+ {
831
+ "epoch": 1.79,
832
+ "learning_rate": 1.928571428571429e-06,
833
+ "loss": 0.2796,
834
+ "step": 3325
835
+ },
836
+ {
837
+ "epoch": 1.81,
838
+ "learning_rate": 1.8571428571428573e-06,
839
+ "loss": 0.2781,
840
+ "step": 3350
841
+ },
842
+ {
843
+ "epoch": 1.82,
844
+ "learning_rate": 1.7857142857142859e-06,
845
+ "loss": 0.296,
846
+ "step": 3375
847
+ },
848
+ {
849
+ "epoch": 1.83,
850
+ "learning_rate": 1.7142857142857145e-06,
851
+ "loss": 0.2912,
852
+ "step": 3400
853
+ },
854
+ {
855
+ "epoch": 1.85,
856
+ "learning_rate": 1.642857142857143e-06,
857
+ "loss": 0.2926,
858
+ "step": 3425
859
+ },
860
+ {
861
+ "epoch": 1.86,
862
+ "learning_rate": 1.5714285714285714e-06,
863
+ "loss": 0.2783,
864
+ "step": 3450
865
+ },
866
+ {
867
+ "epoch": 1.87,
868
+ "learning_rate": 1.5e-06,
869
+ "loss": 0.286,
870
+ "step": 3475
871
+ },
872
+ {
873
+ "epoch": 1.89,
874
+ "learning_rate": 1.4285714285714286e-06,
875
+ "loss": 0.2876,
876
+ "step": 3500
877
+ },
878
+ {
879
+ "epoch": 1.9,
880
+ "learning_rate": 1.3571428571428572e-06,
881
+ "loss": 0.2921,
882
+ "step": 3525
883
+ },
884
+ {
885
+ "epoch": 1.91,
886
+ "learning_rate": 1.2857142857142856e-06,
887
+ "loss": 0.2826,
888
+ "step": 3550
889
+ },
890
+ {
891
+ "epoch": 1.93,
892
+ "learning_rate": 1.2142857142857144e-06,
893
+ "loss": 0.2786,
894
+ "step": 3575
895
+ },
896
+ {
897
+ "epoch": 1.94,
898
+ "learning_rate": 1.142857142857143e-06,
899
+ "loss": 0.2856,
900
+ "step": 3600
901
+ },
902
+ {
903
+ "epoch": 1.95,
904
+ "learning_rate": 1.0714285714285714e-06,
905
+ "loss": 0.2834,
906
+ "step": 3625
907
+ },
908
+ {
909
+ "epoch": 1.97,
910
+ "learning_rate": 1.0000000000000002e-06,
911
+ "loss": 0.2862,
912
+ "step": 3650
913
+ },
914
+ {
915
+ "epoch": 1.98,
916
+ "learning_rate": 9.285714285714287e-07,
917
+ "loss": 0.2818,
918
+ "step": 3675
919
+ },
920
+ {
921
+ "epoch": 2.0,
922
+ "learning_rate": 8.571428571428572e-07,
923
+ "loss": 0.2835,
924
+ "step": 3700
925
+ },
926
+ {
927
+ "epoch": 2.01,
928
+ "learning_rate": 7.857142857142857e-07,
929
+ "loss": 0.2625,
930
+ "step": 3725
931
+ },
932
+ {
933
+ "epoch": 2.02,
934
+ "learning_rate": 7.142857142857143e-07,
935
+ "loss": 0.2344,
936
+ "step": 3750
937
+ },
938
+ {
939
+ "epoch": 2.04,
940
+ "learning_rate": 6.428571428571428e-07,
941
+ "loss": 0.2297,
942
+ "step": 3775
943
+ },
944
+ {
945
+ "epoch": 2.05,
946
+ "learning_rate": 5.714285714285715e-07,
947
+ "loss": 0.2424,
948
+ "step": 3800
949
+ },
950
+ {
951
+ "epoch": 2.06,
952
+ "learning_rate": 5.000000000000001e-07,
953
+ "loss": 0.2417,
954
+ "step": 3825
955
+ },
956
+ {
957
+ "epoch": 2.08,
958
+ "learning_rate": 4.285714285714286e-07,
959
+ "loss": 0.2407,
960
+ "step": 3850
961
+ },
962
+ {
963
+ "epoch": 2.09,
964
+ "learning_rate": 3.5714285714285716e-07,
965
+ "loss": 0.2326,
966
+ "step": 3875
967
+ },
968
+ {
969
+ "epoch": 2.1,
970
+ "learning_rate": 2.8571428571428575e-07,
971
+ "loss": 0.2247,
972
+ "step": 3900
973
+ },
974
+ {
975
+ "epoch": 2.12,
976
+ "learning_rate": 2.142857142857143e-07,
977
+ "loss": 0.2334,
978
+ "step": 3925
979
+ },
980
+ {
981
+ "epoch": 2.13,
982
+ "learning_rate": 1.4285714285714287e-07,
983
+ "loss": 0.2205,
984
+ "step": 3950
985
+ },
986
+ {
987
+ "epoch": 2.14,
988
+ "learning_rate": 7.142857142857144e-08,
989
+ "loss": 0.2362,
990
+ "step": 3975
991
+ },
992
+ {
993
+ "epoch": 2.16,
994
+ "learning_rate": 0.0,
995
+ "loss": 0.249,
996
+ "step": 4000
997
+ },
998
+ {
999
+ "epoch": 2.16,
1000
+ "eval_loss": 0.34158244729042053,
1001
+ "eval_runtime": 3115.7686,
1002
+ "eval_samples_per_second": 3.944,
1003
+ "eval_steps_per_second": 0.493,
1004
+ "eval_wer": 34.928488014365186,
1005
+ "step": 4000
1006
+ }
1007
+ ],
1008
+ "logging_steps": 25,
1009
+ "max_steps": 4000,
1010
+ "num_input_tokens_seen": 0,
1011
+ "num_train_epochs": 3,
1012
+ "save_steps": 1000,
1013
+ "total_flos": 3.693258236215296e+19,
1014
+ "train_batch_size": 16,
1015
+ "trial_name": null,
1016
+ "trial_params": null
1017
+ }