TheRains commited on
Commit
dfa4f88
1 Parent(s): 05761ef

End of training

Browse files
all_results.json CHANGED
@@ -1,12 +1,12 @@
1
  {
2
- "epoch": 11.88,
3
- "eval_loss": 0.00039999079308472574,
4
- "eval_runtime": 3540.3367,
5
- "eval_samples_per_second": 1.424,
6
- "eval_steps_per_second": 0.238,
7
- "eval_wer": 0.3947650718729886,
8
- "train_loss": 0.13066921601230277,
9
- "train_runtime": 56461.7194,
10
- "train_samples_per_second": 1.063,
11
- "train_steps_per_second": 0.089
12
  }
 
1
  {
2
+ "epoch": 7.26,
3
+ "eval_loss": 0.456723153591156,
4
+ "eval_runtime": 2157.1928,
5
+ "eval_samples_per_second": 1.679,
6
+ "eval_steps_per_second": 0.28,
7
+ "eval_wer": 17.593742811134117,
8
+ "train_loss": 0.17717905388567598,
9
+ "train_runtime": 47932.271,
10
+ "train_samples_per_second": 1.252,
11
+ "train_steps_per_second": 0.104
12
  }
eval_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
- "epoch": 11.88,
3
- "eval_loss": 0.00039999079308472574,
4
- "eval_runtime": 3540.3367,
5
- "eval_samples_per_second": 1.424,
6
- "eval_steps_per_second": 0.238,
7
- "eval_wer": 0.3947650718729886
8
  }
 
1
  {
2
+ "epoch": 7.26,
3
+ "eval_loss": 0.456723153591156,
4
+ "eval_runtime": 2157.1928,
5
+ "eval_samples_per_second": 1.679,
6
+ "eval_steps_per_second": 0.28,
7
+ "eval_wer": 17.593742811134117
8
  }
runs/Aug11_09-04-47_dsai-server/events.out.tfevents.1691769882.dsai-server ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9ab57bfc0c2f62fb1160d1e6f5d083779cc95ae7274c0fa5ce31e1025d3be00
3
+ size 40
train_results.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "epoch": 11.88,
3
- "train_loss": 0.13066921601230277,
4
- "train_runtime": 56461.7194,
5
- "train_samples_per_second": 1.063,
6
- "train_steps_per_second": 0.089
7
  }
 
1
  {
2
+ "epoch": 7.26,
3
+ "train_loss": 0.17717905388567598,
4
+ "train_runtime": 47932.271,
5
+ "train_samples_per_second": 1.252,
6
+ "train_steps_per_second": 0.104
7
  }
trainer_state.json CHANGED
@@ -1,1270 +1,1270 @@
1
  {
2
- "best_metric": 0.3947650718729886,
3
  "best_model_checkpoint": "./output/small/cv9-special-batch12-lr4-small/checkpoint-5000",
4
- "epoch": 11.876484560570072,
5
  "global_step": 5000,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
- "epoch": 0.06,
12
  "learning_rate": 4.4e-06,
13
- "loss": 1.2864,
14
  "step": 25
15
  },
16
  {
17
- "epoch": 0.12,
18
  "learning_rate": 9.4e-06,
19
- "loss": 0.6622,
20
  "step": 50
21
  },
22
  {
23
- "epoch": 0.18,
24
  "learning_rate": 1.44e-05,
25
- "loss": 0.4779,
26
  "step": 75
27
  },
28
  {
29
- "epoch": 0.24,
30
  "learning_rate": 1.94e-05,
31
- "loss": 0.4205,
32
  "step": 100
33
  },
34
  {
35
- "epoch": 0.3,
36
  "learning_rate": 2.44e-05,
37
- "loss": 0.3893,
38
  "step": 125
39
  },
40
  {
41
- "epoch": 0.36,
42
  "learning_rate": 2.94e-05,
43
- "loss": 0.4268,
44
  "step": 150
45
  },
46
  {
47
- "epoch": 0.42,
48
  "learning_rate": 3.4399999999999996e-05,
49
- "loss": 0.4469,
50
  "step": 175
51
  },
52
  {
53
- "epoch": 0.48,
54
  "learning_rate": 3.94e-05,
55
- "loss": 0.4307,
56
  "step": 200
57
  },
58
  {
59
- "epoch": 0.53,
60
  "learning_rate": 4.44e-05,
61
- "loss": 0.4571,
62
  "step": 225
63
  },
64
  {
65
- "epoch": 0.59,
66
  "learning_rate": 4.94e-05,
67
- "loss": 0.4735,
68
  "step": 250
69
  },
70
  {
71
- "epoch": 0.65,
72
  "learning_rate": 5.440000000000001e-05,
73
- "loss": 0.5237,
74
  "step": 275
75
  },
76
  {
77
- "epoch": 0.71,
78
  "learning_rate": 5.94e-05,
79
- "loss": 0.5129,
80
  "step": 300
81
  },
82
  {
83
- "epoch": 0.77,
84
  "learning_rate": 6.440000000000001e-05,
85
- "loss": 0.5336,
86
  "step": 325
87
  },
88
  {
89
- "epoch": 0.83,
90
  "learning_rate": 6.939999999999999e-05,
91
- "loss": 0.536,
92
  "step": 350
93
  },
94
  {
95
- "epoch": 0.89,
96
- "learning_rate": 7.42e-05,
97
- "loss": 0.5525,
98
  "step": 375
99
  },
100
  {
101
- "epoch": 0.95,
102
- "learning_rate": 7.920000000000001e-05,
103
- "loss": 0.5774,
104
  "step": 400
105
  },
106
  {
107
- "epoch": 1.01,
108
- "learning_rate": 8.42e-05,
109
- "loss": 0.5515,
110
  "step": 425
111
  },
112
  {
113
- "epoch": 1.07,
114
- "learning_rate": 8.92e-05,
115
- "loss": 0.3206,
116
  "step": 450
117
  },
118
  {
119
- "epoch": 1.13,
120
- "learning_rate": 9.42e-05,
121
- "loss": 0.3541,
122
  "step": 475
123
  },
124
  {
125
- "epoch": 1.19,
126
- "learning_rate": 9.92e-05,
127
- "loss": 0.3557,
128
  "step": 500
129
  },
130
  {
131
- "epoch": 1.25,
132
- "learning_rate": 9.953333333333333e-05,
133
- "loss": 0.3749,
134
  "step": 525
135
  },
136
  {
137
- "epoch": 1.31,
138
- "learning_rate": 9.897777777777778e-05,
139
- "loss": 0.4073,
140
  "step": 550
141
  },
142
  {
143
- "epoch": 1.37,
144
- "learning_rate": 9.842222222222223e-05,
145
- "loss": 0.4469,
146
  "step": 575
147
  },
148
  {
149
- "epoch": 1.43,
150
- "learning_rate": 9.786666666666667e-05,
151
- "loss": 0.4036,
152
  "step": 600
153
  },
154
  {
155
- "epoch": 1.48,
156
- "learning_rate": 9.731111111111112e-05,
157
- "loss": 0.3851,
158
  "step": 625
159
  },
160
  {
161
- "epoch": 1.54,
162
- "learning_rate": 9.675555555555556e-05,
163
- "loss": 0.41,
164
  "step": 650
165
  },
166
  {
167
- "epoch": 1.6,
168
- "learning_rate": 9.620000000000001e-05,
169
- "loss": 0.4313,
170
  "step": 675
171
  },
172
  {
173
- "epoch": 1.66,
174
- "learning_rate": 9.564444444444444e-05,
175
- "loss": 0.3967,
176
  "step": 700
177
  },
178
  {
179
- "epoch": 1.72,
180
- "learning_rate": 9.50888888888889e-05,
181
- "loss": 0.4012,
182
  "step": 725
183
  },
184
  {
185
- "epoch": 1.78,
186
- "learning_rate": 9.453333333333335e-05,
187
- "loss": 0.4157,
188
  "step": 750
189
  },
190
  {
191
- "epoch": 1.84,
192
- "learning_rate": 9.397777777777778e-05,
193
- "loss": 0.442,
194
  "step": 775
195
  },
196
  {
197
- "epoch": 1.9,
198
- "learning_rate": 9.342222222222222e-05,
199
- "loss": 0.4496,
200
  "step": 800
201
  },
202
  {
203
- "epoch": 1.96,
204
- "learning_rate": 9.286666666666667e-05,
205
- "loss": 0.4323,
206
  "step": 825
207
  },
208
  {
209
- "epoch": 2.02,
210
- "learning_rate": 9.231111111111112e-05,
211
- "loss": 0.3541,
212
  "step": 850
213
  },
214
  {
215
- "epoch": 2.08,
216
- "learning_rate": 9.175555555555556e-05,
217
- "loss": 0.1938,
218
  "step": 875
219
  },
220
  {
221
- "epoch": 2.14,
222
- "learning_rate": 9.120000000000001e-05,
223
- "loss": 0.198,
224
  "step": 900
225
  },
226
  {
227
- "epoch": 2.2,
228
- "learning_rate": 9.064444444444446e-05,
229
- "loss": 0.1856,
230
  "step": 925
231
  },
232
  {
233
- "epoch": 2.26,
234
- "learning_rate": 9.00888888888889e-05,
235
- "loss": 0.2461,
236
  "step": 950
237
  },
238
  {
239
- "epoch": 2.32,
240
- "learning_rate": 8.953333333333333e-05,
241
- "loss": 0.2167,
242
  "step": 975
243
  },
244
  {
245
- "epoch": 2.38,
246
- "learning_rate": 8.897777777777778e-05,
247
- "loss": 0.2264,
248
  "step": 1000
249
  },
250
  {
251
- "epoch": 2.38,
252
- "eval_loss": 0.1835644692182541,
253
- "eval_runtime": 3841.5239,
254
- "eval_samples_per_second": 1.313,
255
- "eval_steps_per_second": 0.219,
256
- "eval_wer": 14.998927268826431,
257
  "step": 1000
258
  },
259
  {
260
- "epoch": 2.43,
261
- "learning_rate": 8.842222222222223e-05,
262
- "loss": 0.2213,
263
  "step": 1025
264
  },
265
  {
266
- "epoch": 2.49,
267
- "learning_rate": 8.786666666666667e-05,
268
- "loss": 0.2369,
269
  "step": 1050
270
  },
271
  {
272
- "epoch": 2.55,
273
- "learning_rate": 8.731111111111111e-05,
274
- "loss": 0.2331,
275
  "step": 1075
276
  },
277
  {
278
- "epoch": 2.61,
279
- "learning_rate": 8.675555555555556e-05,
280
- "loss": 0.242,
281
  "step": 1100
282
  },
283
  {
284
- "epoch": 2.67,
285
- "learning_rate": 8.620000000000001e-05,
286
- "loss": 0.2202,
287
  "step": 1125
288
  },
289
  {
290
- "epoch": 2.73,
291
- "learning_rate": 8.564444444444445e-05,
292
- "loss": 0.2433,
293
  "step": 1150
294
  },
295
  {
296
- "epoch": 2.79,
297
- "learning_rate": 8.50888888888889e-05,
298
- "loss": 0.2322,
299
  "step": 1175
300
  },
301
  {
302
- "epoch": 2.85,
303
- "learning_rate": 8.453333333333335e-05,
304
- "loss": 0.2293,
305
  "step": 1200
306
  },
307
  {
308
- "epoch": 2.91,
309
- "learning_rate": 8.397777777777778e-05,
310
- "loss": 0.2536,
311
  "step": 1225
312
  },
313
  {
314
- "epoch": 2.97,
315
- "learning_rate": 8.342222222222222e-05,
316
- "loss": 0.2033,
317
  "step": 1250
318
  },
319
  {
320
- "epoch": 3.03,
321
- "learning_rate": 8.286666666666667e-05,
322
- "loss": 0.1874,
323
  "step": 1275
324
  },
325
  {
326
- "epoch": 3.09,
327
- "learning_rate": 8.231111111111112e-05,
328
- "loss": 0.1299,
329
  "step": 1300
330
  },
331
  {
332
- "epoch": 3.15,
333
- "learning_rate": 8.175555555555556e-05,
334
- "loss": 0.121,
335
  "step": 1325
336
  },
337
  {
338
- "epoch": 3.21,
339
- "learning_rate": 8.120000000000001e-05,
340
- "loss": 0.1583,
341
  "step": 1350
342
  },
343
  {
344
- "epoch": 3.27,
345
- "learning_rate": 8.064444444444445e-05,
346
- "loss": 0.1572,
347
  "step": 1375
348
  },
349
  {
350
- "epoch": 3.33,
351
- "learning_rate": 8.00888888888889e-05,
352
- "loss": 0.143,
353
  "step": 1400
354
  },
355
  {
356
- "epoch": 3.38,
357
- "learning_rate": 7.953333333333333e-05,
358
- "loss": 0.1455,
359
  "step": 1425
360
  },
361
  {
362
- "epoch": 3.44,
363
- "learning_rate": 7.897777777777778e-05,
364
- "loss": 0.1275,
365
  "step": 1450
366
  },
367
  {
368
- "epoch": 3.5,
369
- "learning_rate": 7.842222222222223e-05,
370
- "loss": 0.1438,
371
  "step": 1475
372
  },
373
  {
374
- "epoch": 3.56,
375
- "learning_rate": 7.786666666666667e-05,
376
- "loss": 0.1506,
377
  "step": 1500
378
  },
379
  {
380
- "epoch": 3.62,
381
- "learning_rate": 7.731111111111111e-05,
382
- "loss": 0.1527,
383
  "step": 1525
384
  },
385
  {
386
- "epoch": 3.68,
387
- "learning_rate": 7.675555555555556e-05,
388
- "loss": 0.1321,
389
  "step": 1550
390
  },
391
  {
392
- "epoch": 3.74,
393
- "learning_rate": 7.620000000000001e-05,
394
- "loss": 0.1424,
395
  "step": 1575
396
  },
397
  {
398
- "epoch": 3.8,
399
- "learning_rate": 7.564444444444445e-05,
400
- "loss": 0.151,
401
  "step": 1600
402
  },
403
  {
404
- "epoch": 3.86,
405
- "learning_rate": 7.50888888888889e-05,
406
- "loss": 0.1436,
407
  "step": 1625
408
  },
409
  {
410
- "epoch": 3.92,
411
- "learning_rate": 7.453333333333333e-05,
412
- "loss": 0.1442,
413
  "step": 1650
414
  },
415
  {
416
- "epoch": 3.98,
417
- "learning_rate": 7.397777777777778e-05,
418
- "loss": 0.1365,
419
  "step": 1675
420
  },
421
  {
422
- "epoch": 4.04,
423
- "learning_rate": 7.342222222222222e-05,
424
- "loss": 0.1192,
425
  "step": 1700
426
  },
427
  {
428
- "epoch": 4.1,
429
- "learning_rate": 7.286666666666667e-05,
430
- "loss": 0.0831,
431
  "step": 1725
432
  },
433
  {
434
- "epoch": 4.16,
435
- "learning_rate": 7.231111111111112e-05,
436
- "loss": 0.0916,
437
  "step": 1750
438
  },
439
  {
440
- "epoch": 4.22,
441
- "learning_rate": 7.175555555555556e-05,
442
- "loss": 0.1008,
443
  "step": 1775
444
  },
445
  {
446
- "epoch": 4.28,
447
- "learning_rate": 7.12e-05,
448
- "loss": 0.0885,
449
  "step": 1800
450
  },
451
  {
452
- "epoch": 4.33,
453
- "learning_rate": 7.064444444444445e-05,
454
- "loss": 0.1035,
455
  "step": 1825
456
  },
457
  {
458
- "epoch": 4.39,
459
- "learning_rate": 7.00888888888889e-05,
460
- "loss": 0.0881,
461
  "step": 1850
462
  },
463
  {
464
- "epoch": 4.45,
465
- "learning_rate": 6.953333333333333e-05,
466
- "loss": 0.0956,
467
  "step": 1875
468
  },
469
  {
470
- "epoch": 4.51,
471
- "learning_rate": 6.897777777777778e-05,
472
- "loss": 0.0928,
473
  "step": 1900
474
  },
475
  {
476
- "epoch": 4.57,
477
- "learning_rate": 6.842222222222222e-05,
478
- "loss": 0.0819,
479
  "step": 1925
480
  },
481
  {
482
- "epoch": 4.63,
483
- "learning_rate": 6.786666666666667e-05,
484
- "loss": 0.0943,
485
  "step": 1950
486
  },
487
  {
488
- "epoch": 4.69,
489
- "learning_rate": 6.731111111111111e-05,
490
- "loss": 0.0999,
491
  "step": 1975
492
  },
493
  {
494
- "epoch": 4.75,
495
- "learning_rate": 6.675555555555556e-05,
496
- "loss": 0.0873,
497
  "step": 2000
498
  },
499
  {
500
- "epoch": 4.75,
501
- "eval_loss": 0.0633193850517273,
502
- "eval_runtime": 3766.183,
503
- "eval_samples_per_second": 1.339,
504
- "eval_steps_per_second": 0.223,
505
- "eval_wer": 5.086891225059,
506
  "step": 2000
507
  },
508
  {
509
- "epoch": 4.81,
510
- "learning_rate": 6.620000000000001e-05,
511
- "loss": 0.0905,
512
  "step": 2025
513
  },
514
  {
515
- "epoch": 4.87,
516
- "learning_rate": 6.564444444444445e-05,
517
- "loss": 0.0869,
518
  "step": 2050
519
  },
520
  {
521
- "epoch": 4.93,
522
- "learning_rate": 6.508888888888888e-05,
523
- "loss": 0.1147,
524
  "step": 2075
525
  },
526
  {
527
- "epoch": 4.99,
528
- "learning_rate": 6.453333333333333e-05,
529
- "loss": 0.0957,
530
  "step": 2100
531
  },
532
  {
533
- "epoch": 5.05,
534
- "learning_rate": 6.397777777777778e-05,
535
- "loss": 0.0705,
536
  "step": 2125
537
  },
538
  {
539
- "epoch": 5.11,
540
- "learning_rate": 6.342222222222222e-05,
541
- "loss": 0.0548,
542
  "step": 2150
543
  },
544
  {
545
- "epoch": 5.17,
546
- "learning_rate": 6.286666666666667e-05,
547
- "loss": 0.0564,
548
  "step": 2175
549
  },
550
  {
551
- "epoch": 5.23,
552
- "learning_rate": 6.231111111111111e-05,
553
- "loss": 0.0648,
554
  "step": 2200
555
  },
556
  {
557
- "epoch": 5.29,
558
- "learning_rate": 6.175555555555556e-05,
559
- "loss": 0.0606,
560
  "step": 2225
561
  },
562
  {
563
- "epoch": 5.34,
564
- "learning_rate": 6.12e-05,
565
- "loss": 0.0521,
566
  "step": 2250
567
  },
568
  {
569
- "epoch": 5.4,
570
- "learning_rate": 6.064444444444445e-05,
571
- "loss": 0.0618,
572
  "step": 2275
573
  },
574
  {
575
- "epoch": 5.46,
576
- "learning_rate": 6.00888888888889e-05,
577
- "loss": 0.051,
578
  "step": 2300
579
  },
580
  {
581
- "epoch": 5.52,
582
- "learning_rate": 5.953333333333334e-05,
583
- "loss": 0.0607,
584
  "step": 2325
585
  },
586
  {
587
- "epoch": 5.58,
588
- "learning_rate": 5.897777777777778e-05,
589
- "loss": 0.0649,
590
  "step": 2350
591
  },
592
  {
593
- "epoch": 5.64,
594
- "learning_rate": 5.842222222222222e-05,
595
- "loss": 0.0601,
596
  "step": 2375
597
  },
598
  {
599
- "epoch": 5.7,
600
- "learning_rate": 5.7866666666666666e-05,
601
- "loss": 0.0679,
602
  "step": 2400
603
  },
604
  {
605
- "epoch": 5.76,
606
- "learning_rate": 5.7311111111111117e-05,
607
- "loss": 0.0571,
608
  "step": 2425
609
  },
610
  {
611
- "epoch": 5.82,
612
- "learning_rate": 5.675555555555556e-05,
613
- "loss": 0.0628,
614
  "step": 2450
615
  },
616
  {
617
- "epoch": 5.88,
618
- "learning_rate": 5.620000000000001e-05,
619
- "loss": 0.0643,
620
  "step": 2475
621
  },
622
  {
623
- "epoch": 5.94,
624
- "learning_rate": 5.564444444444444e-05,
625
- "loss": 0.0519,
626
  "step": 2500
627
  },
628
  {
629
- "epoch": 6.0,
630
- "learning_rate": 5.508888888888889e-05,
631
- "loss": 0.0535,
632
  "step": 2525
633
  },
634
  {
635
- "epoch": 6.06,
636
- "learning_rate": 5.4533333333333335e-05,
637
- "loss": 0.0367,
638
  "step": 2550
639
  },
640
  {
641
- "epoch": 6.12,
642
- "learning_rate": 5.3977777777777786e-05,
643
- "loss": 0.0286,
644
  "step": 2575
645
  },
646
  {
647
- "epoch": 6.18,
648
- "learning_rate": 5.342222222222223e-05,
649
- "loss": 0.0392,
650
  "step": 2600
651
  },
652
  {
653
- "epoch": 6.24,
654
- "learning_rate": 5.2866666666666666e-05,
655
- "loss": 0.0435,
656
  "step": 2625
657
  },
658
  {
659
- "epoch": 6.29,
660
- "learning_rate": 5.231111111111111e-05,
661
- "loss": 0.0448,
662
  "step": 2650
663
  },
664
  {
665
- "epoch": 6.35,
666
- "learning_rate": 5.1755555555555554e-05,
667
- "loss": 0.0368,
668
  "step": 2675
669
  },
670
  {
671
- "epoch": 6.41,
672
- "learning_rate": 5.1200000000000004e-05,
673
- "loss": 0.0413,
674
  "step": 2700
675
  },
676
  {
677
- "epoch": 6.47,
678
- "learning_rate": 5.064444444444445e-05,
679
- "loss": 0.0446,
680
  "step": 2725
681
  },
682
  {
683
- "epoch": 6.53,
684
- "learning_rate": 5.00888888888889e-05,
685
- "loss": 0.0434,
686
  "step": 2750
687
  },
688
  {
689
- "epoch": 6.59,
690
- "learning_rate": 4.9533333333333336e-05,
691
- "loss": 0.042,
692
  "step": 2775
693
  },
694
  {
695
- "epoch": 6.65,
696
- "learning_rate": 4.897777777777778e-05,
697
- "loss": 0.032,
698
  "step": 2800
699
  },
700
  {
701
- "epoch": 6.71,
702
- "learning_rate": 4.842222222222222e-05,
703
- "loss": 0.0425,
704
  "step": 2825
705
  },
706
  {
707
- "epoch": 6.77,
708
- "learning_rate": 4.7866666666666674e-05,
709
- "loss": 0.0437,
710
  "step": 2850
711
  },
712
  {
713
- "epoch": 6.83,
714
- "learning_rate": 4.731111111111111e-05,
715
- "loss": 0.033,
716
  "step": 2875
717
  },
718
  {
719
- "epoch": 6.89,
720
- "learning_rate": 4.675555555555556e-05,
721
- "loss": 0.0401,
722
  "step": 2900
723
  },
724
  {
725
- "epoch": 6.95,
726
- "learning_rate": 4.6200000000000005e-05,
727
- "loss": 0.0306,
728
  "step": 2925
729
  },
730
  {
731
- "epoch": 7.01,
732
- "learning_rate": 4.564444444444444e-05,
733
- "loss": 0.0302,
734
  "step": 2950
735
  },
736
  {
737
- "epoch": 7.07,
738
- "learning_rate": 4.508888888888889e-05,
739
- "loss": 0.0234,
740
  "step": 2975
741
  },
742
  {
743
- "epoch": 7.13,
744
- "learning_rate": 4.4533333333333336e-05,
745
- "loss": 0.0236,
746
  "step": 3000
747
  },
748
  {
749
- "epoch": 7.13,
750
- "eval_loss": 0.0261550173163414,
751
- "eval_runtime": 3718.4566,
752
- "eval_samples_per_second": 1.356,
753
- "eval_steps_per_second": 0.226,
754
- "eval_wer": 1.808624758635486,
755
  "step": 3000
756
  },
757
  {
758
- "epoch": 7.19,
759
- "learning_rate": 4.397777777777778e-05,
760
- "loss": 0.0321,
761
  "step": 3025
762
  },
763
  {
764
- "epoch": 7.24,
765
- "learning_rate": 4.3422222222222224e-05,
766
- "loss": 0.0212,
767
  "step": 3050
768
  },
769
  {
770
- "epoch": 7.3,
771
- "learning_rate": 4.286666666666667e-05,
772
- "loss": 0.0312,
773
  "step": 3075
774
  },
775
  {
776
- "epoch": 7.36,
777
- "learning_rate": 4.231111111111111e-05,
778
- "loss": 0.0311,
779
  "step": 3100
780
  },
781
  {
782
- "epoch": 7.42,
783
- "learning_rate": 4.175555555555556e-05,
784
- "loss": 0.0165,
785
  "step": 3125
786
  },
787
  {
788
- "epoch": 7.48,
789
- "learning_rate": 4.12e-05,
790
- "loss": 0.0224,
791
  "step": 3150
792
  },
793
  {
794
- "epoch": 7.54,
795
- "learning_rate": 4.064444444444445e-05,
796
- "loss": 0.0244,
797
  "step": 3175
798
  },
799
  {
800
- "epoch": 7.6,
801
- "learning_rate": 4.008888888888889e-05,
802
- "loss": 0.015,
803
  "step": 3200
804
  },
805
  {
806
- "epoch": 7.66,
807
- "learning_rate": 3.9533333333333337e-05,
808
- "loss": 0.0224,
809
  "step": 3225
810
  },
811
  {
812
- "epoch": 7.72,
813
- "learning_rate": 3.897777777777778e-05,
814
- "loss": 0.0188,
815
  "step": 3250
816
  },
817
  {
818
- "epoch": 7.78,
819
- "learning_rate": 3.8422222222222224e-05,
820
- "loss": 0.0259,
821
  "step": 3275
822
  },
823
  {
824
- "epoch": 7.84,
825
- "learning_rate": 3.786666666666667e-05,
826
- "loss": 0.022,
827
  "step": 3300
828
  },
829
  {
830
- "epoch": 7.9,
831
- "learning_rate": 3.731111111111111e-05,
832
- "loss": 0.0217,
833
  "step": 3325
834
  },
835
  {
836
- "epoch": 7.96,
837
- "learning_rate": 3.675555555555556e-05,
838
- "loss": 0.0262,
839
  "step": 3350
840
  },
841
  {
842
- "epoch": 8.02,
843
- "learning_rate": 3.62e-05,
844
- "loss": 0.0222,
845
  "step": 3375
846
  },
847
  {
848
- "epoch": 8.08,
849
- "learning_rate": 3.564444444444445e-05,
850
- "loss": 0.0161,
851
  "step": 3400
852
  },
853
  {
854
- "epoch": 8.14,
855
- "learning_rate": 3.5088888888888886e-05,
856
- "loss": 0.0188,
857
  "step": 3425
858
  },
859
  {
860
- "epoch": 8.19,
861
- "learning_rate": 3.453333333333334e-05,
862
- "loss": 0.0141,
863
  "step": 3450
864
  },
865
  {
866
- "epoch": 8.25,
867
- "learning_rate": 3.397777777777778e-05,
868
- "loss": 0.0104,
869
  "step": 3475
870
  },
871
  {
872
- "epoch": 8.31,
873
- "learning_rate": 3.3422222222222224e-05,
874
- "loss": 0.012,
875
  "step": 3500
876
  },
877
  {
878
- "epoch": 8.37,
879
- "learning_rate": 3.286666666666667e-05,
880
- "loss": 0.0144,
881
  "step": 3525
882
  },
883
  {
884
- "epoch": 8.43,
885
- "learning_rate": 3.231111111111111e-05,
886
- "loss": 0.0054,
887
  "step": 3550
888
  },
889
  {
890
- "epoch": 8.49,
891
- "learning_rate": 3.1755555555555556e-05,
892
- "loss": 0.0177,
893
  "step": 3575
894
  },
895
  {
896
- "epoch": 8.55,
897
- "learning_rate": 3.12e-05,
898
- "loss": 0.0105,
899
  "step": 3600
900
  },
901
  {
902
- "epoch": 8.61,
903
- "learning_rate": 3.064444444444445e-05,
904
- "loss": 0.0093,
905
  "step": 3625
906
  },
907
  {
908
- "epoch": 8.67,
909
- "learning_rate": 3.008888888888889e-05,
910
- "loss": 0.0162,
911
  "step": 3650
912
  },
913
  {
914
- "epoch": 8.73,
915
- "learning_rate": 2.9533333333333334e-05,
916
- "loss": 0.0144,
917
  "step": 3675
918
  },
919
  {
920
- "epoch": 8.79,
921
- "learning_rate": 2.897777777777778e-05,
922
- "loss": 0.0045,
923
  "step": 3700
924
  },
925
  {
926
- "epoch": 8.85,
927
- "learning_rate": 2.842222222222222e-05,
928
- "loss": 0.0119,
929
  "step": 3725
930
  },
931
  {
932
- "epoch": 8.91,
933
- "learning_rate": 2.786666666666667e-05,
934
- "loss": 0.0107,
935
  "step": 3750
936
  },
937
  {
938
- "epoch": 8.97,
939
- "learning_rate": 2.7311111111111116e-05,
940
- "loss": 0.0108,
941
  "step": 3775
942
  },
943
  {
944
- "epoch": 9.03,
945
- "learning_rate": 2.6755555555555556e-05,
946
- "loss": 0.0079,
947
  "step": 3800
948
  },
949
  {
950
- "epoch": 9.09,
951
- "learning_rate": 2.6200000000000003e-05,
952
- "loss": 0.0079,
953
  "step": 3825
954
  },
955
  {
956
- "epoch": 9.14,
957
- "learning_rate": 2.5644444444444444e-05,
958
- "loss": 0.0056,
959
  "step": 3850
960
  },
961
  {
962
- "epoch": 9.2,
963
- "learning_rate": 2.508888888888889e-05,
964
- "loss": 0.0051,
965
  "step": 3875
966
  },
967
  {
968
- "epoch": 9.26,
969
- "learning_rate": 2.4533333333333334e-05,
970
- "loss": 0.0065,
971
  "step": 3900
972
  },
973
  {
974
- "epoch": 9.32,
975
- "learning_rate": 2.3977777777777778e-05,
976
- "loss": 0.0046,
977
  "step": 3925
978
  },
979
  {
980
- "epoch": 9.38,
981
- "learning_rate": 2.3422222222222222e-05,
982
- "loss": 0.0062,
983
  "step": 3950
984
  },
985
  {
986
- "epoch": 9.44,
987
- "learning_rate": 2.2866666666666666e-05,
988
- "loss": 0.0059,
989
  "step": 3975
990
  },
991
  {
992
- "epoch": 9.5,
993
- "learning_rate": 2.2311111111111113e-05,
994
- "loss": 0.0068,
995
  "step": 4000
996
  },
997
  {
998
- "epoch": 9.5,
999
- "eval_loss": 0.006168792489916086,
1000
- "eval_runtime": 3744.22,
1001
- "eval_samples_per_second": 1.347,
1002
- "eval_steps_per_second": 0.225,
1003
- "eval_wer": 0.6221840806693842,
1004
  "step": 4000
1005
  },
1006
  {
1007
- "epoch": 9.56,
1008
- "learning_rate": 2.1755555555555557e-05,
1009
- "loss": 0.0071,
1010
  "step": 4025
1011
  },
1012
  {
1013
- "epoch": 9.62,
1014
- "learning_rate": 2.12e-05,
1015
- "loss": 0.0096,
1016
  "step": 4050
1017
  },
1018
  {
1019
- "epoch": 9.68,
1020
- "learning_rate": 2.0644444444444447e-05,
1021
- "loss": 0.0059,
1022
  "step": 4075
1023
  },
1024
  {
1025
- "epoch": 9.74,
1026
- "learning_rate": 2.008888888888889e-05,
1027
- "loss": 0.0052,
1028
  "step": 4100
1029
  },
1030
  {
1031
- "epoch": 9.8,
1032
- "learning_rate": 1.9533333333333335e-05,
1033
- "loss": 0.0057,
1034
  "step": 4125
1035
  },
1036
  {
1037
- "epoch": 9.86,
1038
- "learning_rate": 1.897777777777778e-05,
1039
- "loss": 0.0065,
1040
  "step": 4150
1041
  },
1042
  {
1043
- "epoch": 9.92,
1044
- "learning_rate": 1.8422222222222222e-05,
1045
- "loss": 0.004,
1046
  "step": 4175
1047
  },
1048
  {
1049
- "epoch": 9.98,
1050
- "learning_rate": 1.7866666666666666e-05,
1051
- "loss": 0.0099,
1052
  "step": 4200
1053
  },
1054
  {
1055
- "epoch": 10.04,
1056
- "learning_rate": 1.731111111111111e-05,
1057
- "loss": 0.0038,
1058
  "step": 4225
1059
  },
1060
  {
1061
- "epoch": 10.1,
1062
- "learning_rate": 1.6755555555555557e-05,
1063
- "loss": 0.0035,
1064
  "step": 4250
1065
  },
1066
  {
1067
- "epoch": 10.15,
1068
- "learning_rate": 1.62e-05,
1069
- "loss": 0.0066,
1070
  "step": 4275
1071
  },
1072
  {
1073
- "epoch": 10.21,
1074
- "learning_rate": 1.5644444444444444e-05,
1075
- "loss": 0.0037,
1076
  "step": 4300
1077
  },
1078
  {
1079
- "epoch": 10.27,
1080
- "learning_rate": 1.5088888888888888e-05,
1081
- "loss": 0.0041,
1082
  "step": 4325
1083
  },
1084
  {
1085
- "epoch": 10.33,
1086
- "learning_rate": 1.4533333333333335e-05,
1087
- "loss": 0.0027,
1088
  "step": 4350
1089
  },
1090
  {
1091
- "epoch": 10.39,
1092
- "learning_rate": 1.3977777777777779e-05,
1093
- "loss": 0.0014,
1094
  "step": 4375
1095
  },
1096
  {
1097
- "epoch": 10.45,
1098
- "learning_rate": 1.3422222222222223e-05,
1099
- "loss": 0.0027,
1100
  "step": 4400
1101
  },
1102
  {
1103
- "epoch": 10.51,
1104
- "learning_rate": 1.2866666666666668e-05,
1105
- "loss": 0.0036,
1106
  "step": 4425
1107
  },
1108
  {
1109
- "epoch": 10.57,
1110
- "learning_rate": 1.2311111111111112e-05,
1111
- "loss": 0.0017,
1112
  "step": 4450
1113
  },
1114
  {
1115
- "epoch": 10.63,
1116
- "learning_rate": 1.1755555555555556e-05,
1117
- "loss": 0.0039,
1118
  "step": 4475
1119
  },
1120
  {
1121
- "epoch": 10.69,
1122
- "learning_rate": 1.1200000000000001e-05,
1123
- "loss": 0.0008,
1124
  "step": 4500
1125
  },
1126
  {
1127
- "epoch": 10.75,
1128
- "learning_rate": 1.0644444444444445e-05,
1129
- "loss": 0.002,
1130
  "step": 4525
1131
  },
1132
  {
1133
- "epoch": 10.81,
1134
- "learning_rate": 1.0088888888888889e-05,
1135
- "loss": 0.0025,
1136
  "step": 4550
1137
  },
1138
  {
1139
- "epoch": 10.87,
1140
- "learning_rate": 9.533333333333334e-06,
1141
- "loss": 0.0011,
1142
  "step": 4575
1143
  },
1144
  {
1145
- "epoch": 10.93,
1146
- "learning_rate": 8.977777777777778e-06,
1147
- "loss": 0.0022,
1148
  "step": 4600
1149
  },
1150
  {
1151
- "epoch": 10.99,
1152
- "learning_rate": 8.422222222222223e-06,
1153
- "loss": 0.0007,
1154
  "step": 4625
1155
  },
1156
  {
1157
- "epoch": 11.05,
1158
- "learning_rate": 7.866666666666667e-06,
1159
- "loss": 0.0003,
1160
  "step": 4650
1161
  },
1162
  {
1163
- "epoch": 11.1,
1164
- "learning_rate": 7.311111111111112e-06,
1165
- "loss": 0.0003,
1166
  "step": 4675
1167
  },
1168
  {
1169
- "epoch": 11.16,
1170
- "learning_rate": 6.755555555555555e-06,
1171
- "loss": 0.0007,
1172
  "step": 4700
1173
  },
1174
  {
1175
- "epoch": 11.22,
1176
- "learning_rate": 6.2e-06,
1177
- "loss": 0.0003,
1178
  "step": 4725
1179
  },
1180
  {
1181
- "epoch": 11.28,
1182
- "learning_rate": 5.6444444444444445e-06,
1183
- "loss": 0.0003,
1184
  "step": 4750
1185
  },
1186
  {
1187
- "epoch": 11.34,
1188
- "learning_rate": 5.088888888888889e-06,
1189
- "loss": 0.0006,
1190
  "step": 4775
1191
  },
1192
  {
1193
- "epoch": 11.4,
1194
- "learning_rate": 4.533333333333334e-06,
1195
- "loss": 0.0003,
1196
  "step": 4800
1197
  },
1198
  {
1199
- "epoch": 11.46,
1200
- "learning_rate": 3.977777777777778e-06,
1201
- "loss": 0.0003,
1202
  "step": 4825
1203
  },
1204
  {
1205
- "epoch": 11.52,
1206
- "learning_rate": 3.4222222222222224e-06,
1207
- "loss": 0.0003,
1208
  "step": 4850
1209
  },
1210
  {
1211
- "epoch": 11.58,
1212
- "learning_rate": 2.8666666666666666e-06,
1213
- "loss": 0.0016,
1214
  "step": 4875
1215
  },
1216
  {
1217
- "epoch": 11.64,
1218
- "learning_rate": 2.311111111111111e-06,
1219
- "loss": 0.0003,
1220
  "step": 4900
1221
  },
1222
  {
1223
- "epoch": 11.7,
1224
- "learning_rate": 1.7555555555555558e-06,
1225
- "loss": 0.0003,
1226
  "step": 4925
1227
  },
1228
  {
1229
- "epoch": 11.76,
1230
- "learning_rate": 1.2000000000000002e-06,
1231
- "loss": 0.0003,
1232
  "step": 4950
1233
  },
1234
  {
1235
- "epoch": 11.82,
1236
- "learning_rate": 6.444444444444444e-07,
1237
- "loss": 0.0027,
1238
  "step": 4975
1239
  },
1240
  {
1241
- "epoch": 11.88,
1242
- "learning_rate": 8.88888888888889e-08,
1243
- "loss": 0.0002,
1244
  "step": 5000
1245
  },
1246
  {
1247
- "epoch": 11.88,
1248
- "eval_loss": 0.00039999079308472574,
1249
- "eval_runtime": 3766.4449,
1250
- "eval_samples_per_second": 1.339,
1251
- "eval_steps_per_second": 0.223,
1252
- "eval_wer": 0.3947650718729886,
1253
  "step": 5000
1254
  },
1255
  {
1256
- "epoch": 11.88,
1257
  "step": 5000,
1258
- "total_flos": 1.728655406456832e+19,
1259
- "train_loss": 0.13066921601230277,
1260
- "train_runtime": 56461.7194,
1261
- "train_samples_per_second": 1.063,
1262
- "train_steps_per_second": 0.089
1263
  }
1264
  ],
1265
  "max_steps": 5000,
1266
- "num_train_epochs": 12,
1267
- "total_flos": 1.728655406456832e+19,
1268
  "trial_name": null,
1269
  "trial_params": null
1270
  }
 
1
  {
2
+ "best_metric": 17.593742811134117,
3
  "best_model_checkpoint": "./output/small/cv9-special-batch12-lr4-small/checkpoint-5000",
4
+ "epoch": 7.256894049346879,
5
  "global_step": 5000,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
+ "epoch": 0.04,
12
  "learning_rate": 4.4e-06,
13
+ "loss": 1.191,
14
  "step": 25
15
  },
16
  {
17
+ "epoch": 0.07,
18
  "learning_rate": 9.4e-06,
19
+ "loss": 0.6036,
20
  "step": 50
21
  },
22
  {
23
+ "epoch": 0.11,
24
  "learning_rate": 1.44e-05,
25
+ "loss": 0.4394,
26
  "step": 75
27
  },
28
  {
29
+ "epoch": 0.15,
30
  "learning_rate": 1.94e-05,
31
+ "loss": 0.42,
32
  "step": 100
33
  },
34
  {
35
+ "epoch": 0.18,
36
  "learning_rate": 2.44e-05,
37
+ "loss": 0.4484,
38
  "step": 125
39
  },
40
  {
41
+ "epoch": 0.22,
42
  "learning_rate": 2.94e-05,
43
+ "loss": 0.3933,
44
  "step": 150
45
  },
46
  {
47
+ "epoch": 0.25,
48
  "learning_rate": 3.4399999999999996e-05,
49
+ "loss": 0.4136,
50
  "step": 175
51
  },
52
  {
53
+ "epoch": 0.29,
54
  "learning_rate": 3.94e-05,
55
+ "loss": 0.4686,
56
  "step": 200
57
  },
58
  {
59
+ "epoch": 0.33,
60
  "learning_rate": 4.44e-05,
61
+ "loss": 0.4438,
62
  "step": 225
63
  },
64
  {
65
+ "epoch": 0.36,
66
  "learning_rate": 4.94e-05,
67
+ "loss": 0.5118,
68
  "step": 250
69
  },
70
  {
71
+ "epoch": 0.4,
72
  "learning_rate": 5.440000000000001e-05,
73
+ "loss": 0.4983,
74
  "step": 275
75
  },
76
  {
77
+ "epoch": 0.44,
78
  "learning_rate": 5.94e-05,
79
+ "loss": 0.4995,
80
  "step": 300
81
  },
82
  {
83
+ "epoch": 0.47,
84
  "learning_rate": 6.440000000000001e-05,
85
+ "loss": 0.4988,
86
  "step": 325
87
  },
88
  {
89
+ "epoch": 0.51,
90
  "learning_rate": 6.939999999999999e-05,
91
+ "loss": 0.5419,
92
  "step": 350
93
  },
94
  {
95
+ "epoch": 0.54,
96
+ "learning_rate": 7.44e-05,
97
+ "loss": 0.5401,
98
  "step": 375
99
  },
100
  {
101
+ "epoch": 0.58,
102
+ "learning_rate": 7.94e-05,
103
+ "loss": 0.5586,
104
  "step": 400
105
  },
106
  {
107
+ "epoch": 0.62,
108
+ "learning_rate": 8.44e-05,
109
+ "loss": 0.5791,
110
  "step": 425
111
  },
112
  {
113
+ "epoch": 0.65,
114
+ "learning_rate": 8.94e-05,
115
+ "loss": 0.6103,
116
  "step": 450
117
  },
118
  {
119
+ "epoch": 0.69,
120
+ "learning_rate": 9.44e-05,
121
+ "loss": 0.6625,
122
  "step": 475
123
  },
124
  {
125
+ "epoch": 0.73,
126
+ "learning_rate": 9.94e-05,
127
+ "loss": 0.6622,
128
  "step": 500
129
  },
130
  {
131
+ "epoch": 0.76,
132
+ "learning_rate": 9.95111111111111e-05,
133
+ "loss": 0.6632,
134
  "step": 525
135
  },
136
  {
137
+ "epoch": 0.8,
138
+ "learning_rate": 9.895555555555556e-05,
139
+ "loss": 0.657,
140
  "step": 550
141
  },
142
  {
143
+ "epoch": 0.83,
144
+ "learning_rate": 9.84e-05,
145
+ "loss": 0.6526,
146
  "step": 575
147
  },
148
  {
149
+ "epoch": 0.87,
150
+ "learning_rate": 9.784444444444444e-05,
151
+ "loss": 0.6615,
152
  "step": 600
153
  },
154
  {
155
+ "epoch": 0.91,
156
+ "learning_rate": 9.72888888888889e-05,
157
+ "loss": 0.6762,
158
  "step": 625
159
  },
160
  {
161
+ "epoch": 0.94,
162
+ "learning_rate": 9.673333333333334e-05,
163
+ "loss": 0.6791,
164
  "step": 650
165
  },
166
  {
167
+ "epoch": 0.98,
168
+ "learning_rate": 9.617777777777778e-05,
169
+ "loss": 0.6629,
170
  "step": 675
171
  },
172
  {
173
+ "epoch": 1.02,
174
+ "learning_rate": 9.562222222222222e-05,
175
+ "loss": 0.5772,
176
  "step": 700
177
  },
178
  {
179
+ "epoch": 1.05,
180
+ "learning_rate": 9.506666666666667e-05,
181
+ "loss": 0.3396,
182
  "step": 725
183
  },
184
  {
185
+ "epoch": 1.09,
186
+ "learning_rate": 9.451111111111112e-05,
187
+ "loss": 0.3575,
188
  "step": 750
189
  },
190
  {
191
+ "epoch": 1.12,
192
+ "learning_rate": 9.395555555555557e-05,
193
+ "loss": 0.3685,
194
  "step": 775
195
  },
196
  {
197
+ "epoch": 1.16,
198
+ "learning_rate": 9.340000000000001e-05,
199
+ "loss": 0.3711,
200
  "step": 800
201
  },
202
  {
203
+ "epoch": 1.2,
204
+ "learning_rate": 9.284444444444444e-05,
205
+ "loss": 0.3574,
206
  "step": 825
207
  },
208
  {
209
+ "epoch": 1.23,
210
+ "learning_rate": 9.22888888888889e-05,
211
+ "loss": 0.3793,
212
  "step": 850
213
  },
214
  {
215
+ "epoch": 1.27,
216
+ "learning_rate": 9.173333333333333e-05,
217
+ "loss": 0.375,
218
  "step": 875
219
  },
220
  {
221
+ "epoch": 1.31,
222
+ "learning_rate": 9.117777777777778e-05,
223
+ "loss": 0.4094,
224
  "step": 900
225
  },
226
  {
227
+ "epoch": 1.34,
228
+ "learning_rate": 9.062222222222223e-05,
229
+ "loss": 0.3846,
230
  "step": 925
231
  },
232
  {
233
+ "epoch": 1.38,
234
+ "learning_rate": 9.006666666666667e-05,
235
+ "loss": 0.3782,
236
  "step": 950
237
  },
238
  {
239
+ "epoch": 1.42,
240
+ "learning_rate": 8.95111111111111e-05,
241
+ "loss": 0.3584,
242
  "step": 975
243
  },
244
  {
245
+ "epoch": 1.45,
246
+ "learning_rate": 8.895555555555556e-05,
247
+ "loss": 0.3743,
248
  "step": 1000
249
  },
250
  {
251
+ "epoch": 1.45,
252
+ "eval_loss": 0.5497691631317139,
253
+ "eval_runtime": 2341.8245,
254
+ "eval_samples_per_second": 1.547,
255
+ "eval_steps_per_second": 0.258,
256
+ "eval_wer": 28.27237175063262,
257
  "step": 1000
258
  },
259
  {
260
+ "epoch": 1.49,
261
+ "learning_rate": 8.840000000000001e-05,
262
+ "loss": 0.3457,
263
  "step": 1025
264
  },
265
  {
266
+ "epoch": 1.52,
267
+ "learning_rate": 8.784444444444446e-05,
268
+ "loss": 0.3747,
269
  "step": 1050
270
  },
271
  {
272
+ "epoch": 1.56,
273
+ "learning_rate": 8.72888888888889e-05,
274
+ "loss": 0.3429,
275
  "step": 1075
276
  },
277
  {
278
+ "epoch": 1.6,
279
+ "learning_rate": 8.673333333333333e-05,
280
+ "loss": 0.3562,
281
  "step": 1100
282
  },
283
  {
284
+ "epoch": 1.63,
285
+ "learning_rate": 8.617777777777778e-05,
286
+ "loss": 0.3805,
287
  "step": 1125
288
  },
289
  {
290
+ "epoch": 1.67,
291
+ "learning_rate": 8.562222222222222e-05,
292
+ "loss": 0.3905,
293
  "step": 1150
294
  },
295
  {
296
+ "epoch": 1.71,
297
+ "learning_rate": 8.506666666666667e-05,
298
+ "loss": 0.3792,
299
  "step": 1175
300
  },
301
  {
302
+ "epoch": 1.74,
303
+ "learning_rate": 8.451111111111112e-05,
304
+ "loss": 0.3614,
305
  "step": 1200
306
  },
307
  {
308
+ "epoch": 1.78,
309
+ "learning_rate": 8.395555555555557e-05,
310
+ "loss": 0.3787,
311
  "step": 1225
312
  },
313
  {
314
+ "epoch": 1.81,
315
+ "learning_rate": 8.34e-05,
316
+ "loss": 0.3381,
317
  "step": 1250
318
  },
319
  {
320
+ "epoch": 1.85,
321
+ "learning_rate": 8.284444444444444e-05,
322
+ "loss": 0.3715,
323
  "step": 1275
324
  },
325
  {
326
+ "epoch": 1.89,
327
+ "learning_rate": 8.22888888888889e-05,
328
+ "loss": 0.3818,
329
  "step": 1300
330
  },
331
  {
332
+ "epoch": 1.92,
333
+ "learning_rate": 8.173333333333335e-05,
334
+ "loss": 0.3498,
335
  "step": 1325
336
  },
337
  {
338
+ "epoch": 1.96,
339
+ "learning_rate": 8.117777777777778e-05,
340
+ "loss": 0.3778,
341
  "step": 1350
342
  },
343
  {
344
+ "epoch": 2.0,
345
+ "learning_rate": 8.062222222222222e-05,
346
+ "loss": 0.3512,
347
  "step": 1375
348
  },
349
  {
350
+ "epoch": 2.03,
351
+ "learning_rate": 8.006666666666667e-05,
352
+ "loss": 0.1781,
353
  "step": 1400
354
  },
355
  {
356
+ "epoch": 2.07,
357
+ "learning_rate": 7.951111111111111e-05,
358
+ "loss": 0.1771,
359
  "step": 1425
360
  },
361
  {
362
+ "epoch": 2.1,
363
+ "learning_rate": 7.895555555555556e-05,
364
+ "loss": 0.1765,
365
  "step": 1450
366
  },
367
  {
368
+ "epoch": 2.14,
369
+ "learning_rate": 7.840000000000001e-05,
370
+ "loss": 0.1813,
371
  "step": 1475
372
  },
373
  {
374
+ "epoch": 2.18,
375
+ "learning_rate": 7.784444444444446e-05,
376
+ "loss": 0.1741,
377
  "step": 1500
378
  },
379
  {
380
+ "epoch": 2.21,
381
+ "learning_rate": 7.728888888888888e-05,
382
+ "loss": 0.1679,
383
  "step": 1525
384
  },
385
  {
386
+ "epoch": 2.25,
387
+ "learning_rate": 7.673333333333333e-05,
388
+ "loss": 0.1941,
389
  "step": 1550
390
  },
391
  {
392
+ "epoch": 2.29,
393
+ "learning_rate": 7.617777777777778e-05,
394
+ "loss": 0.1667,
395
  "step": 1575
396
  },
397
  {
398
+ "epoch": 2.32,
399
+ "learning_rate": 7.562222222222223e-05,
400
+ "loss": 0.1783,
401
  "step": 1600
402
  },
403
  {
404
+ "epoch": 2.36,
405
+ "learning_rate": 7.506666666666667e-05,
406
+ "loss": 0.1625,
407
  "step": 1625
408
  },
409
  {
410
+ "epoch": 2.39,
411
+ "learning_rate": 7.451111111111111e-05,
412
+ "loss": 0.1849,
413
  "step": 1650
414
  },
415
  {
416
+ "epoch": 2.43,
417
+ "learning_rate": 7.395555555555556e-05,
418
+ "loss": 0.1872,
419
  "step": 1675
420
  },
421
  {
422
+ "epoch": 2.47,
423
+ "learning_rate": 7.340000000000001e-05,
424
+ "loss": 0.1949,
425
  "step": 1700
426
  },
427
  {
428
+ "epoch": 2.5,
429
+ "learning_rate": 7.284444444444445e-05,
430
+ "loss": 0.1759,
431
  "step": 1725
432
  },
433
  {
434
+ "epoch": 2.54,
435
+ "learning_rate": 7.22888888888889e-05,
436
+ "loss": 0.182,
437
  "step": 1750
438
  },
439
  {
440
+ "epoch": 2.58,
441
+ "learning_rate": 7.173333333333335e-05,
442
+ "loss": 0.1907,
443
  "step": 1775
444
  },
445
  {
446
+ "epoch": 2.61,
447
+ "learning_rate": 7.117777777777777e-05,
448
+ "loss": 0.1982,
449
  "step": 1800
450
  },
451
  {
452
+ "epoch": 2.65,
453
+ "learning_rate": 7.062222222222222e-05,
454
+ "loss": 0.1707,
455
  "step": 1825
456
  },
457
  {
458
+ "epoch": 2.69,
459
+ "learning_rate": 7.006666666666667e-05,
460
+ "loss": 0.1818,
461
  "step": 1850
462
  },
463
  {
464
+ "epoch": 2.72,
465
+ "learning_rate": 6.951111111111112e-05,
466
+ "loss": 0.1791,
467
  "step": 1875
468
  },
469
  {
470
+ "epoch": 2.76,
471
+ "learning_rate": 6.895555555555556e-05,
472
+ "loss": 0.1866,
473
  "step": 1900
474
  },
475
  {
476
+ "epoch": 2.79,
477
+ "learning_rate": 6.840000000000001e-05,
478
+ "loss": 0.1761,
479
  "step": 1925
480
  },
481
  {
482
+ "epoch": 2.83,
483
+ "learning_rate": 6.784444444444445e-05,
484
+ "loss": 0.1766,
485
  "step": 1950
486
  },
487
  {
488
+ "epoch": 2.87,
489
+ "learning_rate": 6.72888888888889e-05,
490
+ "loss": 0.1599,
491
  "step": 1975
492
  },
493
  {
494
+ "epoch": 2.9,
495
+ "learning_rate": 6.673333333333333e-05,
496
+ "loss": 0.1633,
497
  "step": 2000
498
  },
499
  {
500
+ "epoch": 2.9,
501
+ "eval_loss": 0.5009704828262329,
502
+ "eval_runtime": 2334.1804,
503
+ "eval_samples_per_second": 1.552,
504
+ "eval_steps_per_second": 0.259,
505
+ "eval_wer": 25.152979066022546,
506
  "step": 2000
507
  },
508
  {
509
+ "epoch": 2.94,
510
+ "learning_rate": 6.617777777777778e-05,
511
+ "loss": 0.1863,
512
  "step": 2025
513
  },
514
  {
515
+ "epoch": 2.98,
516
+ "learning_rate": 6.562222222222223e-05,
517
+ "loss": 0.182,
518
  "step": 2050
519
  },
520
  {
521
+ "epoch": 3.01,
522
+ "learning_rate": 6.506666666666666e-05,
523
+ "loss": 0.149,
524
  "step": 2075
525
  },
526
  {
527
+ "epoch": 3.05,
528
+ "learning_rate": 6.451111111111111e-05,
529
+ "loss": 0.0853,
530
  "step": 2100
531
  },
532
  {
533
+ "epoch": 3.08,
534
+ "learning_rate": 6.395555555555556e-05,
535
+ "loss": 0.0913,
536
  "step": 2125
537
  },
538
  {
539
+ "epoch": 3.12,
540
+ "learning_rate": 6.340000000000001e-05,
541
+ "loss": 0.0969,
542
  "step": 2150
543
  },
544
  {
545
+ "epoch": 3.16,
546
+ "learning_rate": 6.284444444444445e-05,
547
+ "loss": 0.0838,
548
  "step": 2175
549
  },
550
  {
551
+ "epoch": 3.19,
552
+ "learning_rate": 6.22888888888889e-05,
553
+ "loss": 0.0936,
554
  "step": 2200
555
  },
556
  {
557
+ "epoch": 3.23,
558
+ "learning_rate": 6.173333333333333e-05,
559
+ "loss": 0.0939,
560
  "step": 2225
561
  },
562
  {
563
+ "epoch": 3.27,
564
+ "learning_rate": 6.117777777777778e-05,
565
+ "loss": 0.0937,
566
  "step": 2250
567
  },
568
  {
569
+ "epoch": 3.3,
570
+ "learning_rate": 6.062222222222222e-05,
571
+ "loss": 0.0939,
572
  "step": 2275
573
  },
574
  {
575
+ "epoch": 3.34,
576
+ "learning_rate": 6.006666666666667e-05,
577
+ "loss": 0.0963,
578
  "step": 2300
579
  },
580
  {
581
+ "epoch": 3.37,
582
+ "learning_rate": 5.9511111111111116e-05,
583
+ "loss": 0.0912,
584
  "step": 2325
585
  },
586
  {
587
+ "epoch": 3.41,
588
+ "learning_rate": 5.8955555555555566e-05,
589
+ "loss": 0.1011,
590
  "step": 2350
591
  },
592
  {
593
+ "epoch": 3.45,
594
+ "learning_rate": 5.8399999999999997e-05,
595
+ "loss": 0.1081,
596
  "step": 2375
597
  },
598
  {
599
+ "epoch": 3.48,
600
+ "learning_rate": 5.784444444444445e-05,
601
+ "loss": 0.0984,
602
  "step": 2400
603
  },
604
  {
605
+ "epoch": 3.52,
606
+ "learning_rate": 5.728888888888889e-05,
607
+ "loss": 0.0938,
608
  "step": 2425
609
  },
610
  {
611
+ "epoch": 3.56,
612
+ "learning_rate": 5.673333333333334e-05,
613
+ "loss": 0.0957,
614
  "step": 2450
615
  },
616
  {
617
+ "epoch": 3.59,
618
+ "learning_rate": 5.6177777777777785e-05,
619
+ "loss": 0.0971,
620
  "step": 2475
621
  },
622
  {
623
+ "epoch": 3.63,
624
+ "learning_rate": 5.562222222222222e-05,
625
+ "loss": 0.0863,
626
  "step": 2500
627
  },
628
  {
629
+ "epoch": 3.66,
630
+ "learning_rate": 5.5066666666666666e-05,
631
+ "loss": 0.0897,
632
  "step": 2525
633
  },
634
  {
635
+ "epoch": 3.7,
636
+ "learning_rate": 5.451111111111111e-05,
637
+ "loss": 0.0954,
638
  "step": 2550
639
  },
640
  {
641
+ "epoch": 3.74,
642
+ "learning_rate": 5.395555555555556e-05,
643
+ "loss": 0.0792,
644
  "step": 2575
645
  },
646
  {
647
+ "epoch": 3.77,
648
+ "learning_rate": 5.3400000000000004e-05,
649
+ "loss": 0.0961,
650
  "step": 2600
651
  },
652
  {
653
+ "epoch": 3.81,
654
+ "learning_rate": 5.2844444444444454e-05,
655
+ "loss": 0.0887,
656
  "step": 2625
657
  },
658
  {
659
+ "epoch": 3.85,
660
+ "learning_rate": 5.2288888888888884e-05,
661
+ "loss": 0.0887,
662
  "step": 2650
663
  },
664
  {
665
+ "epoch": 3.88,
666
+ "learning_rate": 5.1733333333333335e-05,
667
+ "loss": 0.0843,
668
  "step": 2675
669
  },
670
  {
671
+ "epoch": 3.92,
672
+ "learning_rate": 5.117777777777778e-05,
673
+ "loss": 0.0798,
674
  "step": 2700
675
  },
676
  {
677
+ "epoch": 3.96,
678
+ "learning_rate": 5.062222222222223e-05,
679
+ "loss": 0.0869,
680
  "step": 2725
681
  },
682
  {
683
+ "epoch": 3.99,
684
+ "learning_rate": 5.006666666666667e-05,
685
+ "loss": 0.085,
686
  "step": 2750
687
  },
688
  {
689
+ "epoch": 4.03,
690
+ "learning_rate": 4.951111111111112e-05,
691
+ "loss": 0.0477,
692
  "step": 2775
693
  },
694
  {
695
+ "epoch": 4.06,
696
+ "learning_rate": 4.895555555555556e-05,
697
+ "loss": 0.0518,
698
  "step": 2800
699
  },
700
  {
701
+ "epoch": 4.1,
702
+ "learning_rate": 4.8400000000000004e-05,
703
+ "loss": 0.0501,
704
  "step": 2825
705
  },
706
  {
707
+ "epoch": 4.14,
708
+ "learning_rate": 4.784444444444445e-05,
709
+ "loss": 0.0394,
710
  "step": 2850
711
  },
712
  {
713
+ "epoch": 4.17,
714
+ "learning_rate": 4.728888888888889e-05,
715
+ "loss": 0.046,
716
  "step": 2875
717
  },
718
  {
719
+ "epoch": 4.21,
720
+ "learning_rate": 4.6733333333333335e-05,
721
+ "loss": 0.0405,
722
  "step": 2900
723
  },
724
  {
725
+ "epoch": 4.25,
726
+ "learning_rate": 4.617777777777778e-05,
727
+ "loss": 0.0525,
728
  "step": 2925
729
  },
730
  {
731
+ "epoch": 4.28,
732
+ "learning_rate": 4.562222222222222e-05,
733
+ "loss": 0.0443,
734
  "step": 2950
735
  },
736
  {
737
+ "epoch": 4.32,
738
+ "learning_rate": 4.5066666666666667e-05,
739
+ "loss": 0.0444,
740
  "step": 2975
741
  },
742
  {
743
+ "epoch": 4.35,
744
+ "learning_rate": 4.451111111111112e-05,
745
+ "loss": 0.0505,
746
  "step": 3000
747
  },
748
  {
749
+ "epoch": 4.35,
750
+ "eval_loss": 0.5048635005950928,
751
+ "eval_runtime": 2311.8612,
752
+ "eval_samples_per_second": 1.567,
753
+ "eval_steps_per_second": 0.261,
754
+ "eval_wer": 22.167011732229124,
755
  "step": 3000
756
  },
757
  {
758
+ "epoch": 4.39,
759
+ "learning_rate": 4.3955555555555554e-05,
760
+ "loss": 0.0371,
761
  "step": 3025
762
  },
763
  {
764
+ "epoch": 4.43,
765
+ "learning_rate": 4.3400000000000005e-05,
766
+ "loss": 0.0496,
767
  "step": 3050
768
  },
769
  {
770
+ "epoch": 4.46,
771
+ "learning_rate": 4.284444444444445e-05,
772
+ "loss": 0.043,
773
  "step": 3075
774
  },
775
  {
776
+ "epoch": 4.5,
777
+ "learning_rate": 4.228888888888889e-05,
778
+ "loss": 0.0517,
779
  "step": 3100
780
  },
781
  {
782
+ "epoch": 4.54,
783
+ "learning_rate": 4.1733333333333336e-05,
784
+ "loss": 0.0509,
785
  "step": 3125
786
  },
787
  {
788
+ "epoch": 4.57,
789
+ "learning_rate": 4.117777777777778e-05,
790
+ "loss": 0.0409,
791
  "step": 3150
792
  },
793
  {
794
+ "epoch": 4.61,
795
+ "learning_rate": 4.062222222222222e-05,
796
+ "loss": 0.0389,
797
  "step": 3175
798
  },
799
  {
800
+ "epoch": 4.64,
801
+ "learning_rate": 4.006666666666667e-05,
802
+ "loss": 0.0426,
803
  "step": 3200
804
  },
805
  {
806
+ "epoch": 4.68,
807
+ "learning_rate": 3.951111111111112e-05,
808
+ "loss": 0.0414,
809
  "step": 3225
810
  },
811
  {
812
+ "epoch": 4.72,
813
+ "learning_rate": 3.8955555555555555e-05,
814
+ "loss": 0.0457,
815
  "step": 3250
816
  },
817
  {
818
+ "epoch": 4.75,
819
+ "learning_rate": 3.8400000000000005e-05,
820
+ "loss": 0.0378,
821
  "step": 3275
822
  },
823
  {
824
+ "epoch": 4.79,
825
+ "learning_rate": 3.784444444444445e-05,
826
+ "loss": 0.0413,
827
  "step": 3300
828
  },
829
  {
830
+ "epoch": 4.83,
831
+ "learning_rate": 3.728888888888889e-05,
832
+ "loss": 0.0435,
833
  "step": 3325
834
  },
835
  {
836
+ "epoch": 4.86,
837
+ "learning_rate": 3.6733333333333336e-05,
838
+ "loss": 0.0454,
839
  "step": 3350
840
  },
841
  {
842
+ "epoch": 4.9,
843
+ "learning_rate": 3.617777777777778e-05,
844
+ "loss": 0.0369,
845
  "step": 3375
846
  },
847
  {
848
+ "epoch": 4.93,
849
+ "learning_rate": 3.5622222222222224e-05,
850
+ "loss": 0.0391,
851
  "step": 3400
852
  },
853
  {
854
+ "epoch": 4.97,
855
+ "learning_rate": 3.506666666666667e-05,
856
+ "loss": 0.0354,
857
  "step": 3425
858
  },
859
  {
860
+ "epoch": 5.01,
861
+ "learning_rate": 3.451111111111111e-05,
862
+ "loss": 0.0336,
863
  "step": 3450
864
  },
865
  {
866
+ "epoch": 5.04,
867
+ "learning_rate": 3.3955555555555555e-05,
868
+ "loss": 0.0251,
869
  "step": 3475
870
  },
871
  {
872
+ "epoch": 5.08,
873
+ "learning_rate": 3.3400000000000005e-05,
874
+ "loss": 0.0131,
875
  "step": 3500
876
  },
877
  {
878
+ "epoch": 5.12,
879
+ "learning_rate": 3.284444444444444e-05,
880
+ "loss": 0.0214,
881
  "step": 3525
882
  },
883
  {
884
+ "epoch": 5.15,
885
+ "learning_rate": 3.228888888888889e-05,
886
+ "loss": 0.0191,
887
  "step": 3550
888
  },
889
  {
890
+ "epoch": 5.19,
891
+ "learning_rate": 3.173333333333334e-05,
892
+ "loss": 0.0245,
893
  "step": 3575
894
  },
895
  {
896
+ "epoch": 5.22,
897
+ "learning_rate": 3.117777777777778e-05,
898
+ "loss": 0.0132,
899
  "step": 3600
900
  },
901
  {
902
+ "epoch": 5.26,
903
+ "learning_rate": 3.0622222222222224e-05,
904
+ "loss": 0.0213,
905
  "step": 3625
906
  },
907
  {
908
+ "epoch": 5.3,
909
+ "learning_rate": 3.006666666666667e-05,
910
+ "loss": 0.0196,
911
  "step": 3650
912
  },
913
  {
914
+ "epoch": 5.33,
915
+ "learning_rate": 2.951111111111111e-05,
916
+ "loss": 0.0222,
917
  "step": 3675
918
  },
919
  {
920
+ "epoch": 5.37,
921
+ "learning_rate": 2.895555555555556e-05,
922
+ "loss": 0.0173,
923
  "step": 3700
924
  },
925
  {
926
+ "epoch": 5.41,
927
+ "learning_rate": 2.84e-05,
928
+ "loss": 0.0194,
929
  "step": 3725
930
  },
931
  {
932
+ "epoch": 5.44,
933
+ "learning_rate": 2.7844444444444446e-05,
934
+ "loss": 0.017,
935
  "step": 3750
936
  },
937
  {
938
+ "epoch": 5.48,
939
+ "learning_rate": 2.728888888888889e-05,
940
+ "loss": 0.0177,
941
  "step": 3775
942
  },
943
  {
944
+ "epoch": 5.52,
945
+ "learning_rate": 2.6733333333333334e-05,
946
+ "loss": 0.0176,
947
  "step": 3800
948
  },
949
  {
950
+ "epoch": 5.55,
951
+ "learning_rate": 2.6177777777777777e-05,
952
+ "loss": 0.0199,
953
  "step": 3825
954
  },
955
  {
956
+ "epoch": 5.59,
957
+ "learning_rate": 2.5622222222222225e-05,
958
+ "loss": 0.0168,
959
  "step": 3850
960
  },
961
  {
962
+ "epoch": 5.62,
963
+ "learning_rate": 2.5066666666666665e-05,
964
+ "loss": 0.0268,
965
  "step": 3875
966
  },
967
  {
968
+ "epoch": 5.66,
969
+ "learning_rate": 2.4511111111111112e-05,
970
+ "loss": 0.0175,
971
  "step": 3900
972
  },
973
  {
974
+ "epoch": 5.7,
975
+ "learning_rate": 2.3955555555555556e-05,
976
+ "loss": 0.0142,
977
  "step": 3925
978
  },
979
  {
980
+ "epoch": 5.73,
981
+ "learning_rate": 2.3400000000000003e-05,
982
+ "loss": 0.0153,
983
  "step": 3950
984
  },
985
  {
986
+ "epoch": 5.77,
987
+ "learning_rate": 2.2844444444444447e-05,
988
+ "loss": 0.0138,
989
  "step": 3975
990
  },
991
  {
992
+ "epoch": 5.81,
993
+ "learning_rate": 2.228888888888889e-05,
994
+ "loss": 0.0136,
995
  "step": 4000
996
  },
997
  {
998
+ "epoch": 5.81,
999
+ "eval_loss": 0.4631120264530182,
1000
+ "eval_runtime": 2311.1233,
1001
+ "eval_samples_per_second": 1.567,
1002
+ "eval_steps_per_second": 0.261,
1003
+ "eval_wer": 18.633540372670808,
1004
  "step": 4000
1005
  },
1006
  {
1007
+ "epoch": 5.84,
1008
+ "learning_rate": 2.1733333333333334e-05,
1009
+ "loss": 0.0166,
1010
  "step": 4025
1011
  },
1012
  {
1013
+ "epoch": 5.88,
1014
+ "learning_rate": 2.117777777777778e-05,
1015
+ "loss": 0.0152,
1016
  "step": 4050
1017
  },
1018
  {
1019
+ "epoch": 5.91,
1020
+ "learning_rate": 2.062222222222222e-05,
1021
+ "loss": 0.0099,
1022
  "step": 4075
1023
  },
1024
  {
1025
+ "epoch": 5.95,
1026
+ "learning_rate": 2.0066666666666665e-05,
1027
+ "loss": 0.0158,
1028
  "step": 4100
1029
  },
1030
  {
1031
+ "epoch": 5.99,
1032
+ "learning_rate": 1.9511111111111113e-05,
1033
+ "loss": 0.0131,
1034
  "step": 4125
1035
  },
1036
  {
1037
+ "epoch": 6.02,
1038
+ "learning_rate": 1.8955555555555556e-05,
1039
+ "loss": 0.01,
1040
  "step": 4150
1041
  },
1042
  {
1043
+ "epoch": 6.06,
1044
+ "learning_rate": 1.84e-05,
1045
+ "loss": 0.005,
1046
  "step": 4175
1047
  },
1048
  {
1049
+ "epoch": 6.1,
1050
+ "learning_rate": 1.7844444444444444e-05,
1051
+ "loss": 0.0083,
1052
  "step": 4200
1053
  },
1054
  {
1055
+ "epoch": 6.13,
1056
+ "learning_rate": 1.728888888888889e-05,
1057
+ "loss": 0.0047,
1058
  "step": 4225
1059
  },
1060
  {
1061
+ "epoch": 6.17,
1062
+ "learning_rate": 1.6733333333333335e-05,
1063
+ "loss": 0.0067,
1064
  "step": 4250
1065
  },
1066
  {
1067
+ "epoch": 6.2,
1068
+ "learning_rate": 1.617777777777778e-05,
1069
+ "loss": 0.0029,
1070
  "step": 4275
1071
  },
1072
  {
1073
+ "epoch": 6.24,
1074
+ "learning_rate": 1.5622222222222225e-05,
1075
+ "loss": 0.002,
1076
  "step": 4300
1077
  },
1078
  {
1079
+ "epoch": 6.28,
1080
+ "learning_rate": 1.5066666666666668e-05,
1081
+ "loss": 0.0031,
1082
  "step": 4325
1083
  },
1084
  {
1085
+ "epoch": 6.31,
1086
+ "learning_rate": 1.4511111111111111e-05,
1087
+ "loss": 0.0036,
1088
  "step": 4350
1089
  },
1090
  {
1091
+ "epoch": 6.35,
1092
+ "learning_rate": 1.3955555555555555e-05,
1093
+ "loss": 0.0041,
1094
  "step": 4375
1095
  },
1096
  {
1097
+ "epoch": 6.39,
1098
+ "learning_rate": 1.3400000000000002e-05,
1099
+ "loss": 0.0033,
1100
  "step": 4400
1101
  },
1102
  {
1103
+ "epoch": 6.42,
1104
+ "learning_rate": 1.2844444444444446e-05,
1105
+ "loss": 0.0062,
1106
  "step": 4425
1107
  },
1108
  {
1109
+ "epoch": 6.46,
1110
+ "learning_rate": 1.228888888888889e-05,
1111
+ "loss": 0.0035,
1112
  "step": 4450
1113
  },
1114
  {
1115
+ "epoch": 6.49,
1116
+ "learning_rate": 1.1733333333333333e-05,
1117
+ "loss": 0.004,
1118
  "step": 4475
1119
  },
1120
  {
1121
+ "epoch": 6.53,
1122
+ "learning_rate": 1.1177777777777779e-05,
1123
+ "loss": 0.0048,
1124
  "step": 4500
1125
  },
1126
  {
1127
+ "epoch": 6.57,
1128
+ "learning_rate": 1.0622222222222223e-05,
1129
+ "loss": 0.0041,
1130
  "step": 4525
1131
  },
1132
  {
1133
+ "epoch": 6.6,
1134
+ "learning_rate": 1.0066666666666668e-05,
1135
+ "loss": 0.0063,
1136
  "step": 4550
1137
  },
1138
  {
1139
+ "epoch": 6.64,
1140
+ "learning_rate": 9.511111111111112e-06,
1141
+ "loss": 0.0047,
1142
  "step": 4575
1143
  },
1144
  {
1145
+ "epoch": 6.68,
1146
+ "learning_rate": 8.955555555555555e-06,
1147
+ "loss": 0.0039,
1148
  "step": 4600
1149
  },
1150
  {
1151
+ "epoch": 6.71,
1152
+ "learning_rate": 8.400000000000001e-06,
1153
+ "loss": 0.0019,
1154
  "step": 4625
1155
  },
1156
  {
1157
+ "epoch": 6.75,
1158
+ "learning_rate": 7.844444444444445e-06,
1159
+ "loss": 0.0033,
1160
  "step": 4650
1161
  },
1162
  {
1163
+ "epoch": 6.79,
1164
+ "learning_rate": 7.288888888888889e-06,
1165
+ "loss": 0.0045,
1166
  "step": 4675
1167
  },
1168
  {
1169
+ "epoch": 6.82,
1170
+ "learning_rate": 6.733333333333333e-06,
1171
+ "loss": 0.0051,
1172
  "step": 4700
1173
  },
1174
  {
1175
+ "epoch": 6.86,
1176
+ "learning_rate": 6.177777777777778e-06,
1177
+ "loss": 0.0068,
1178
  "step": 4725
1179
  },
1180
  {
1181
+ "epoch": 6.89,
1182
+ "learning_rate": 5.622222222222222e-06,
1183
+ "loss": 0.0022,
1184
  "step": 4750
1185
  },
1186
  {
1187
+ "epoch": 6.93,
1188
+ "learning_rate": 5.066666666666667e-06,
1189
+ "loss": 0.0024,
1190
  "step": 4775
1191
  },
1192
  {
1193
+ "epoch": 6.97,
1194
+ "learning_rate": 4.511111111111111e-06,
1195
+ "loss": 0.0027,
1196
  "step": 4800
1197
  },
1198
  {
1199
+ "epoch": 7.0,
1200
+ "learning_rate": 3.955555555555555e-06,
1201
+ "loss": 0.007,
1202
  "step": 4825
1203
  },
1204
  {
1205
+ "epoch": 7.04,
1206
+ "learning_rate": 3.4000000000000005e-06,
1207
+ "loss": 0.0018,
1208
  "step": 4850
1209
  },
1210
  {
1211
+ "epoch": 7.08,
1212
+ "learning_rate": 2.8444444444444446e-06,
1213
+ "loss": 0.0006,
1214
  "step": 4875
1215
  },
1216
  {
1217
+ "epoch": 7.11,
1218
+ "learning_rate": 2.2888888888888892e-06,
1219
+ "loss": 0.0007,
1220
  "step": 4900
1221
  },
1222
  {
1223
+ "epoch": 7.15,
1224
+ "learning_rate": 1.7333333333333334e-06,
1225
+ "loss": 0.0005,
1226
  "step": 4925
1227
  },
1228
  {
1229
+ "epoch": 7.18,
1230
+ "learning_rate": 1.1777777777777778e-06,
1231
+ "loss": 0.0007,
1232
  "step": 4950
1233
  },
1234
  {
1235
+ "epoch": 7.22,
1236
+ "learning_rate": 6.222222222222223e-07,
1237
+ "loss": 0.0008,
1238
  "step": 4975
1239
  },
1240
  {
1241
+ "epoch": 7.26,
1242
+ "learning_rate": 6.666666666666667e-08,
1243
+ "loss": 0.0005,
1244
  "step": 5000
1245
  },
1246
  {
1247
+ "epoch": 7.26,
1248
+ "eval_loss": 0.456723153591156,
1249
+ "eval_runtime": 2309.2911,
1250
+ "eval_samples_per_second": 1.568,
1251
+ "eval_steps_per_second": 0.262,
1252
+ "eval_wer": 17.593742811134117,
1253
  "step": 5000
1254
  },
1255
  {
1256
+ "epoch": 7.26,
1257
  "step": 5000,
1258
+ "total_flos": 1.730098333458432e+19,
1259
+ "train_loss": 0.17717905388567598,
1260
+ "train_runtime": 47932.271,
1261
+ "train_samples_per_second": 1.252,
1262
+ "train_steps_per_second": 0.104
1263
  }
1264
  ],
1265
  "max_steps": 5000,
1266
+ "num_train_epochs": 8,
1267
+ "total_flos": 1.730098333458432e+19,
1268
  "trial_name": null,
1269
  "trial_params": null
1270
  }