xezpeleta commited on
Commit
8d88cf9
1 Parent(s): b382d42

Add results

Browse files
Files changed (5) hide show
  1. all_results.json +5 -5
  2. ggml-small.eu.bin +1 -1
  3. nohup.out +18 -0
  4. train_results.json +5 -5
  5. trainer_state.json +424 -424
all_results.json CHANGED
@@ -1,12 +1,12 @@
1
  {
2
- "epoch": 12.01,
3
  "eval_loss": 0.3812163174152374,
4
  "eval_runtime": 951.9575,
5
  "eval_samples_per_second": 6.924,
6
  "eval_steps_per_second": 0.433,
7
  "eval_wer": 18.775568066750374,
8
- "train_loss": 0.106446673027426,
9
- "train_runtime": 27653.1068,
10
- "train_samples_per_second": 5.786,
11
- "train_steps_per_second": 0.181
12
  }
 
1
  {
2
+ "epoch": 6.06,
3
  "eval_loss": 0.3812163174152374,
4
  "eval_runtime": 951.9575,
5
  "eval_samples_per_second": 6.924,
6
  "eval_steps_per_second": 0.433,
7
  "eval_wer": 18.775568066750374,
8
+ "train_loss": 0.15837200704216958,
9
+ "train_runtime": 29005.3522,
10
+ "train_samples_per_second": 5.516,
11
+ "train_steps_per_second": 0.172
12
  }
ggml-small.eu.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2e35b59747aa9d0976001fe99c33ff3f422c3af984ee4440b5c1392c282b6d51
3
  size 487601984
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:939aa258c716f2fa50d1b18d3a649fcaac19fbfc865df87e09303b0e4383ef73
3
  size 487601984
nohup.out CHANGED
@@ -70598,3 +70598,21 @@ The progress bars may be unreliable.
70598
 
70599
 
70600
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
70598
 
70599
 
70600
 
70601
+ To https://huggingface.co/xezpeleta/whisper-small-eu
70602
+ df538ec..b382d42 main -> main
70603
+
70604
+ 12/29/2023 22:56:29 - WARNING - huggingface_hub.repository - To https://huggingface.co/xezpeleta/whisper-small-eu
70605
+ df538ec..b382d42 main -> main
70606
+
70607
+ ***** train metrics *****
70608
+ epoch = 6.06
70609
+ train_loss = 0.1584
70610
+ train_runtime = 8:03:25.35
70611
+ train_samples_per_second = 5.516
70612
+ train_steps_per_second = 0.172
70613
+ 12/29/2023 22:56:32 - INFO - __main__ - *** Evaluate ***
70614
+ [INFO|trainer.py:2955] 2023-12-29 22:56:32,446 >> ***** Running Evaluation *****
70615
+ [INFO|trainer.py:2959] 2023-12-29 22:56:32,446 >> Num examples: Unknown
70616
+ [INFO|trainer.py:2960] 2023-12-29 22:56:32,446 >> Batch size = 16
70617
+
70618
+ [INFO|trainer_utils.py:689] 2023-12-29 22:56:36,997 >> The following columns in the evaluation set don't have a corresponding argument in `WhisperForConditionalGeneration.forward` and have been ignored: input_length. If input_length are not expected by `WhisperForConditionalGeneration.forward`, you can safely ignore this message.
train_results.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "epoch": 12.01,
3
- "train_loss": 0.106446673027426,
4
- "train_runtime": 27653.1068,
5
- "train_samples_per_second": 5.786,
6
- "train_steps_per_second": 0.181
7
  }
 
1
  {
2
+ "epoch": 6.06,
3
+ "train_loss": 0.15837200704216958,
4
+ "train_runtime": 29005.3522,
5
+ "train_samples_per_second": 5.516,
6
+ "train_steps_per_second": 0.172
7
  }
trainer_state.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "best_metric": 18.775568066750374,
3
- "best_model_checkpoint": "./checkpoint-4000",
4
- "epoch": 12.0112,
5
  "global_step": 5000,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
@@ -10,1261 +10,1261 @@
10
  {
11
  "epoch": 0.01,
12
  "learning_rate": 4.4e-07,
13
- "loss": 1.6639,
14
  "step": 25
15
  },
16
  {
17
  "epoch": 0.01,
18
  "learning_rate": 9.400000000000001e-07,
19
- "loss": 1.4728,
20
  "step": 50
21
  },
22
  {
23
  "epoch": 0.01,
24
  "learning_rate": 1.44e-06,
25
- "loss": 1.1563,
26
  "step": 75
27
  },
28
  {
29
  "epoch": 0.02,
30
  "learning_rate": 1.94e-06,
31
- "loss": 0.9368,
32
  "step": 100
33
  },
34
  {
35
  "epoch": 0.03,
36
  "learning_rate": 2.4400000000000004e-06,
37
- "loss": 0.8889,
38
  "step": 125
39
  },
40
  {
41
  "epoch": 0.03,
42
  "learning_rate": 2.9400000000000002e-06,
43
- "loss": 0.8114,
44
  "step": 150
45
  },
46
  {
47
  "epoch": 0.04,
48
  "learning_rate": 3.44e-06,
49
- "loss": 0.6924,
50
  "step": 175
51
  },
52
  {
53
  "epoch": 0.04,
54
  "learning_rate": 3.94e-06,
55
- "loss": 0.661,
56
  "step": 200
57
  },
58
  {
59
  "epoch": 0.04,
60
  "learning_rate": 4.440000000000001e-06,
61
- "loss": 0.6427,
62
  "step": 225
63
  },
64
  {
65
  "epoch": 0.05,
66
  "learning_rate": 4.94e-06,
67
- "loss": 0.5217,
68
  "step": 250
69
  },
70
  {
71
  "epoch": 0.06,
72
  "learning_rate": 5.4400000000000004e-06,
73
- "loss": 0.4991,
74
  "step": 275
75
  },
76
  {
77
  "epoch": 0.06,
78
  "learning_rate": 5.94e-06,
79
- "loss": 0.4768,
80
  "step": 300
81
  },
82
  {
83
  "epoch": 0.07,
84
  "learning_rate": 6.440000000000001e-06,
85
- "loss": 0.4632,
86
  "step": 325
87
  },
88
  {
89
  "epoch": 0.07,
90
  "learning_rate": 6.9400000000000005e-06,
91
- "loss": 0.4289,
92
  "step": 350
93
  },
94
  {
95
  "epoch": 0.07,
96
  "learning_rate": 7.440000000000001e-06,
97
- "loss": 0.404,
98
  "step": 375
99
  },
100
  {
101
  "epoch": 0.08,
102
  "learning_rate": 7.94e-06,
103
- "loss": 0.3833,
104
  "step": 400
105
  },
106
  {
107
- "epoch": 1.0,
108
  "learning_rate": 8.44e-06,
109
- "loss": 0.4201,
110
  "step": 425
111
  },
112
  {
113
- "epoch": 1.01,
114
  "learning_rate": 8.94e-06,
115
- "loss": 0.3989,
116
  "step": 450
117
  },
118
  {
119
- "epoch": 1.01,
120
  "learning_rate": 9.440000000000001e-06,
121
- "loss": 0.3453,
122
  "step": 475
123
  },
124
  {
125
- "epoch": 1.02,
126
  "learning_rate": 9.940000000000001e-06,
127
- "loss": 0.3559,
128
  "step": 500
129
  },
130
  {
131
- "epoch": 1.02,
132
  "learning_rate": 9.951111111111111e-06,
133
- "loss": 0.3429,
134
  "step": 525
135
  },
136
  {
137
- "epoch": 1.03,
138
  "learning_rate": 9.895555555555557e-06,
139
- "loss": 0.3701,
140
  "step": 550
141
  },
142
  {
143
- "epoch": 1.03,
144
  "learning_rate": 9.84e-06,
145
- "loss": 0.3084,
146
  "step": 575
147
  },
148
  {
149
- "epoch": 1.04,
150
  "learning_rate": 9.784444444444445e-06,
151
- "loss": 0.2956,
152
  "step": 600
153
  },
154
  {
155
- "epoch": 1.04,
156
  "learning_rate": 9.72888888888889e-06,
157
- "loss": 0.2923,
158
  "step": 625
159
  },
160
  {
161
- "epoch": 1.05,
162
  "learning_rate": 9.673333333333334e-06,
163
- "loss": 0.2786,
164
  "step": 650
165
  },
166
  {
167
- "epoch": 1.05,
168
  "learning_rate": 9.617777777777778e-06,
169
- "loss": 0.2427,
170
  "step": 675
171
  },
172
  {
173
- "epoch": 1.06,
174
  "learning_rate": 9.562222222222223e-06,
175
- "loss": 0.2387,
176
  "step": 700
177
  },
178
  {
179
- "epoch": 1.06,
180
  "learning_rate": 9.506666666666667e-06,
181
- "loss": 0.2371,
182
  "step": 725
183
  },
184
  {
185
- "epoch": 1.07,
186
  "learning_rate": 9.451111111111112e-06,
187
- "loss": 0.2114,
188
  "step": 750
189
  },
190
  {
191
- "epoch": 1.07,
192
  "learning_rate": 9.395555555555556e-06,
193
- "loss": 0.1894,
194
  "step": 775
195
  },
196
  {
197
- "epoch": 1.08,
198
  "learning_rate": 9.340000000000002e-06,
199
- "loss": 0.1867,
200
  "step": 800
201
  },
202
  {
203
- "epoch": 2.0,
204
  "learning_rate": 9.284444444444444e-06,
205
- "loss": 0.1927,
206
  "step": 825
207
  },
208
  {
209
- "epoch": 2.01,
210
  "learning_rate": 9.22888888888889e-06,
211
- "loss": 0.1951,
212
  "step": 850
213
  },
214
  {
215
- "epoch": 2.01,
216
  "learning_rate": 9.173333333333334e-06,
217
- "loss": 0.1775,
218
  "step": 875
219
  },
220
  {
221
- "epoch": 2.02,
222
  "learning_rate": 9.117777777777778e-06,
223
- "loss": 0.1672,
224
  "step": 900
225
  },
226
  {
227
- "epoch": 2.02,
228
  "learning_rate": 9.062222222222224e-06,
229
- "loss": 0.1591,
230
  "step": 925
231
  },
232
  {
233
- "epoch": 2.03,
234
  "learning_rate": 9.006666666666666e-06,
235
- "loss": 0.1703,
236
  "step": 950
237
  },
238
  {
239
- "epoch": 2.03,
240
  "learning_rate": 8.951111111111112e-06,
241
- "loss": 0.1636,
242
  "step": 975
243
  },
244
  {
245
- "epoch": 2.04,
246
  "learning_rate": 8.895555555555556e-06,
247
- "loss": 0.1413,
248
  "step": 1000
249
  },
250
  {
251
- "epoch": 2.04,
252
- "eval_loss": 0.31775403022766113,
253
- "eval_runtime": 963.5028,
254
- "eval_samples_per_second": 6.841,
255
- "eval_steps_per_second": 0.428,
256
- "eval_wer": 22.013852322896838,
257
  "step": 1000
258
  },
259
  {
260
- "epoch": 2.04,
261
  "learning_rate": 8.84e-06,
262
- "loss": 0.1371,
263
  "step": 1025
264
  },
265
  {
266
- "epoch": 2.05,
267
  "learning_rate": 8.784444444444446e-06,
268
- "loss": 0.1494,
269
  "step": 1050
270
  },
271
  {
272
- "epoch": 2.05,
273
  "learning_rate": 8.72888888888889e-06,
274
- "loss": 0.1196,
275
  "step": 1075
276
  },
277
  {
278
- "epoch": 2.06,
279
  "learning_rate": 8.673333333333334e-06,
280
- "loss": 0.1152,
281
  "step": 1100
282
  },
283
  {
284
- "epoch": 2.06,
285
  "learning_rate": 8.617777777777778e-06,
286
- "loss": 0.1158,
287
  "step": 1125
288
  },
289
  {
290
- "epoch": 2.07,
291
  "learning_rate": 8.562222222222224e-06,
292
- "loss": 0.1054,
293
  "step": 1150
294
  },
295
  {
296
- "epoch": 2.07,
297
  "learning_rate": 8.506666666666668e-06,
298
- "loss": 0.0912,
299
  "step": 1175
300
  },
301
  {
302
- "epoch": 2.08,
303
  "learning_rate": 8.451111111111112e-06,
304
- "loss": 0.0915,
305
  "step": 1200
306
  },
307
  {
308
- "epoch": 2.08,
309
  "learning_rate": 8.395555555555557e-06,
310
- "loss": 0.0901,
311
  "step": 1225
312
  },
313
  {
314
- "epoch": 3.0,
315
  "learning_rate": 8.34e-06,
316
- "loss": 0.0953,
317
  "step": 1250
318
  },
319
  {
320
- "epoch": 3.01,
321
  "learning_rate": 8.284444444444446e-06,
322
- "loss": 0.0923,
323
  "step": 1275
324
  },
325
  {
326
- "epoch": 3.01,
327
  "learning_rate": 8.22888888888889e-06,
328
- "loss": 0.0749,
329
  "step": 1300
330
  },
331
  {
332
- "epoch": 3.02,
333
  "learning_rate": 8.173333333333334e-06,
334
- "loss": 0.0796,
335
  "step": 1325
336
  },
337
  {
338
- "epoch": 3.02,
339
  "learning_rate": 8.11777777777778e-06,
340
- "loss": 0.0786,
341
  "step": 1350
342
  },
343
  {
344
- "epoch": 3.03,
345
  "learning_rate": 8.062222222222222e-06,
346
- "loss": 0.0819,
347
  "step": 1375
348
  },
349
  {
350
- "epoch": 3.03,
351
  "learning_rate": 8.006666666666667e-06,
352
- "loss": 0.0747,
353
  "step": 1400
354
  },
355
  {
356
- "epoch": 3.04,
357
  "learning_rate": 7.951111111111111e-06,
358
- "loss": 0.0646,
359
  "step": 1425
360
  },
361
  {
362
- "epoch": 3.04,
363
  "learning_rate": 7.895555555555557e-06,
364
- "loss": 0.0707,
365
  "step": 1450
366
  },
367
  {
368
- "epoch": 3.05,
369
  "learning_rate": 7.840000000000001e-06,
370
- "loss": 0.0639,
371
  "step": 1475
372
  },
373
  {
374
- "epoch": 3.05,
375
  "learning_rate": 7.784444444444445e-06,
376
- "loss": 0.0544,
377
  "step": 1500
378
  },
379
  {
380
- "epoch": 3.06,
381
  "learning_rate": 7.72888888888889e-06,
382
- "loss": 0.0531,
383
  "step": 1525
384
  },
385
  {
386
- "epoch": 3.06,
387
  "learning_rate": 7.673333333333333e-06,
388
- "loss": 0.0518,
389
  "step": 1550
390
  },
391
  {
392
- "epoch": 3.07,
393
  "learning_rate": 7.617777777777778e-06,
394
- "loss": 0.0461,
395
  "step": 1575
396
  },
397
  {
398
- "epoch": 3.07,
399
  "learning_rate": 7.562222222222223e-06,
400
- "loss": 0.0424,
401
  "step": 1600
402
  },
403
  {
404
- "epoch": 3.08,
405
  "learning_rate": 7.506666666666668e-06,
406
- "loss": 0.0392,
407
  "step": 1625
408
  },
409
  {
410
- "epoch": 4.0,
411
  "learning_rate": 7.451111111111111e-06,
412
- "loss": 0.0376,
413
  "step": 1650
414
  },
415
  {
416
- "epoch": 4.01,
417
  "learning_rate": 7.395555555555556e-06,
418
- "loss": 0.0447,
419
  "step": 1675
420
  },
421
  {
422
- "epoch": 4.01,
423
  "learning_rate": 7.340000000000001e-06,
424
- "loss": 0.0362,
425
  "step": 1700
426
  },
427
  {
428
- "epoch": 4.02,
429
  "learning_rate": 7.284444444444445e-06,
430
- "loss": 0.0368,
431
  "step": 1725
432
  },
433
  {
434
- "epoch": 4.02,
435
  "learning_rate": 7.22888888888889e-06,
436
- "loss": 0.0367,
437
  "step": 1750
438
  },
439
  {
440
- "epoch": 4.03,
441
  "learning_rate": 7.173333333333335e-06,
442
- "loss": 0.0391,
443
  "step": 1775
444
  },
445
  {
446
- "epoch": 4.03,
447
  "learning_rate": 7.117777777777778e-06,
448
- "loss": 0.0331,
449
  "step": 1800
450
  },
451
  {
452
- "epoch": 4.04,
453
  "learning_rate": 7.062222222222223e-06,
454
- "loss": 0.028,
455
  "step": 1825
456
  },
457
  {
458
- "epoch": 4.04,
459
  "learning_rate": 7.006666666666667e-06,
460
- "loss": 0.0315,
461
  "step": 1850
462
  },
463
  {
464
- "epoch": 4.05,
465
  "learning_rate": 6.951111111111112e-06,
466
- "loss": 0.0308,
467
  "step": 1875
468
  },
469
  {
470
- "epoch": 4.05,
471
  "learning_rate": 6.8955555555555565e-06,
472
- "loss": 0.0248,
473
  "step": 1900
474
  },
475
  {
476
- "epoch": 4.06,
477
  "learning_rate": 6.8400000000000014e-06,
478
- "loss": 0.023,
479
  "step": 1925
480
  },
481
  {
482
- "epoch": 4.06,
483
  "learning_rate": 6.784444444444445e-06,
484
- "loss": 0.0233,
485
  "step": 1950
486
  },
487
  {
488
- "epoch": 4.07,
489
  "learning_rate": 6.7288888888888895e-06,
490
- "loss": 0.0236,
491
  "step": 1975
492
  },
493
  {
494
- "epoch": 4.07,
495
  "learning_rate": 6.6733333333333335e-06,
496
- "loss": 0.0181,
497
  "step": 2000
498
  },
499
  {
500
- "epoch": 4.07,
501
- "eval_loss": 0.33755382895469666,
502
- "eval_runtime": 952.5181,
503
- "eval_samples_per_second": 6.92,
504
- "eval_steps_per_second": 0.433,
505
- "eval_wer": 20.286362347604197,
506
  "step": 2000
507
  },
508
  {
509
- "epoch": 4.08,
510
  "learning_rate": 6.617777777777778e-06,
511
- "loss": 0.0176,
512
  "step": 2025
513
  },
514
  {
515
- "epoch": 4.08,
516
  "learning_rate": 6.562222222222223e-06,
517
- "loss": 0.0177,
518
  "step": 2050
519
  },
520
  {
521
- "epoch": 5.0,
522
  "learning_rate": 6.5066666666666665e-06,
523
- "loss": 0.0198,
524
  "step": 2075
525
  },
526
  {
527
- "epoch": 5.01,
528
  "learning_rate": 6.451111111111111e-06,
529
- "loss": 0.0188,
530
  "step": 2100
531
  },
532
  {
533
- "epoch": 5.01,
534
  "learning_rate": 6.395555555555556e-06,
535
- "loss": 0.0153,
536
  "step": 2125
537
  },
538
  {
539
- "epoch": 5.02,
540
  "learning_rate": 6.34e-06,
541
- "loss": 0.0157,
542
  "step": 2150
543
  },
544
  {
545
- "epoch": 5.02,
546
  "learning_rate": 6.284444444444445e-06,
547
- "loss": 0.0196,
548
  "step": 2175
549
  },
550
  {
551
- "epoch": 5.03,
552
  "learning_rate": 6.22888888888889e-06,
553
- "loss": 0.0153,
554
  "step": 2200
555
  },
556
  {
557
- "epoch": 5.03,
558
  "learning_rate": 6.173333333333333e-06,
559
- "loss": 0.0132,
560
  "step": 2225
561
  },
562
  {
563
- "epoch": 5.04,
564
  "learning_rate": 6.117777777777778e-06,
565
- "loss": 0.0144,
566
  "step": 2250
567
  },
568
  {
569
- "epoch": 5.04,
570
  "learning_rate": 6.062222222222223e-06,
571
- "loss": 0.0145,
572
  "step": 2275
573
  },
574
  {
575
- "epoch": 5.05,
576
  "learning_rate": 6.006666666666667e-06,
577
- "loss": 0.0131,
578
  "step": 2300
579
  },
580
  {
581
- "epoch": 5.05,
582
  "learning_rate": 5.951111111111112e-06,
583
- "loss": 0.0107,
584
  "step": 2325
585
  },
586
  {
587
- "epoch": 5.06,
588
  "learning_rate": 5.895555555555557e-06,
589
- "loss": 0.0122,
590
  "step": 2350
591
  },
592
  {
593
- "epoch": 5.06,
594
  "learning_rate": 5.84e-06,
595
- "loss": 0.0121,
596
  "step": 2375
597
  },
598
  {
599
- "epoch": 5.07,
600
  "learning_rate": 5.784444444444445e-06,
601
- "loss": 0.0103,
602
  "step": 2400
603
  },
604
  {
605
- "epoch": 5.07,
606
  "learning_rate": 5.72888888888889e-06,
607
- "loss": 0.0089,
608
  "step": 2425
609
  },
610
  {
611
- "epoch": 5.08,
612
  "learning_rate": 5.673333333333334e-06,
613
- "loss": 0.0087,
614
  "step": 2450
615
  },
616
  {
617
- "epoch": 6.0,
618
  "learning_rate": 5.617777777777779e-06,
619
- "loss": 0.0087,
620
  "step": 2475
621
  },
622
  {
623
- "epoch": 6.01,
624
  "learning_rate": 5.562222222222222e-06,
625
- "loss": 0.0106,
626
  "step": 2500
627
  },
628
  {
629
- "epoch": 6.01,
630
  "learning_rate": 5.506666666666667e-06,
631
- "loss": 0.0094,
632
  "step": 2525
633
  },
634
  {
635
- "epoch": 6.02,
636
  "learning_rate": 5.451111111111112e-06,
637
- "loss": 0.0082,
638
  "step": 2550
639
  },
640
  {
641
- "epoch": 6.02,
642
  "learning_rate": 5.3955555555555565e-06,
643
- "loss": 0.0079,
644
  "step": 2575
645
  },
646
  {
647
- "epoch": 6.03,
648
  "learning_rate": 5.3400000000000005e-06,
649
- "loss": 0.0084,
650
  "step": 2600
651
  },
652
  {
653
- "epoch": 6.03,
654
  "learning_rate": 5.2844444444444454e-06,
655
- "loss": 0.0076,
656
  "step": 2625
657
  },
658
  {
659
- "epoch": 6.04,
660
  "learning_rate": 5.228888888888889e-06,
661
- "loss": 0.0069,
662
  "step": 2650
663
  },
664
  {
665
- "epoch": 6.04,
666
  "learning_rate": 5.1733333333333335e-06,
667
- "loss": 0.0067,
668
  "step": 2675
669
  },
670
  {
671
- "epoch": 6.05,
672
  "learning_rate": 5.117777777777778e-06,
673
- "loss": 0.0073,
674
  "step": 2700
675
  },
676
  {
677
- "epoch": 6.05,
678
  "learning_rate": 5.062222222222222e-06,
679
- "loss": 0.0062,
680
  "step": 2725
681
  },
682
  {
683
- "epoch": 6.06,
684
  "learning_rate": 5.006666666666667e-06,
685
- "loss": 0.0057,
686
  "step": 2750
687
  },
688
  {
689
- "epoch": 6.06,
690
  "learning_rate": 4.951111111111111e-06,
691
- "loss": 0.0064,
692
  "step": 2775
693
  },
694
  {
695
- "epoch": 6.07,
696
  "learning_rate": 4.895555555555556e-06,
697
- "loss": 0.0053,
698
  "step": 2800
699
  },
700
  {
701
- "epoch": 6.07,
702
  "learning_rate": 4.84e-06,
703
- "loss": 0.0047,
704
  "step": 2825
705
  },
706
  {
707
- "epoch": 6.08,
708
  "learning_rate": 4.784444444444445e-06,
709
- "loss": 0.005,
710
  "step": 2850
711
  },
712
  {
713
- "epoch": 6.08,
714
  "learning_rate": 4.728888888888889e-06,
715
- "loss": 0.0047,
716
  "step": 2875
717
  },
718
  {
719
- "epoch": 7.0,
720
  "learning_rate": 4.673333333333333e-06,
721
- "loss": 0.0052,
722
  "step": 2900
723
  },
724
  {
725
- "epoch": 7.01,
726
  "learning_rate": 4.617777777777778e-06,
727
- "loss": 0.005,
728
  "step": 2925
729
  },
730
  {
731
- "epoch": 7.01,
732
  "learning_rate": 4.562222222222222e-06,
733
- "loss": 0.0041,
734
  "step": 2950
735
  },
736
  {
737
- "epoch": 7.02,
738
  "learning_rate": 4.506666666666667e-06,
739
- "loss": 0.0045,
740
  "step": 2975
741
  },
742
  {
743
- "epoch": 7.02,
744
  "learning_rate": 4.451111111111112e-06,
745
- "loss": 0.0044,
746
  "step": 3000
747
  },
748
  {
749
- "epoch": 7.02,
750
- "eval_loss": 0.3602650463581085,
751
- "eval_runtime": 960.4855,
752
- "eval_samples_per_second": 6.862,
753
- "eval_steps_per_second": 0.429,
754
- "eval_wer": 18.876827737048888,
755
  "step": 3000
756
  },
757
  {
758
- "epoch": 7.03,
759
  "learning_rate": 4.395555555555556e-06,
760
- "loss": 0.0043,
761
  "step": 3025
762
  },
763
  {
764
- "epoch": 7.03,
765
  "learning_rate": 4.34e-06,
766
- "loss": 0.0041,
767
  "step": 3050
768
  },
769
  {
770
- "epoch": 7.04,
771
  "learning_rate": 4.284444444444445e-06,
772
- "loss": 0.0035,
773
  "step": 3075
774
  },
775
  {
776
- "epoch": 7.04,
777
  "learning_rate": 4.228888888888889e-06,
778
- "loss": 0.0044,
779
  "step": 3100
780
  },
781
  {
782
- "epoch": 7.05,
783
  "learning_rate": 4.173333333333334e-06,
784
- "loss": 0.0039,
785
  "step": 3125
786
  },
787
  {
788
- "epoch": 7.05,
789
  "learning_rate": 4.117777777777779e-06,
790
- "loss": 0.0031,
791
  "step": 3150
792
  },
793
  {
794
- "epoch": 7.06,
795
  "learning_rate": 4.062222222222223e-06,
796
- "loss": 0.0032,
797
  "step": 3175
798
  },
799
  {
800
- "epoch": 7.06,
801
  "learning_rate": 4.006666666666667e-06,
802
- "loss": 0.0034,
803
  "step": 3200
804
  },
805
  {
806
- "epoch": 7.07,
807
  "learning_rate": 3.951111111111112e-06,
808
- "loss": 0.0028,
809
  "step": 3225
810
  },
811
  {
812
- "epoch": 7.07,
813
  "learning_rate": 3.895555555555556e-06,
814
- "loss": 0.0032,
815
  "step": 3250
816
  },
817
  {
818
- "epoch": 7.08,
819
  "learning_rate": 3.8400000000000005e-06,
820
- "loss": 0.003,
821
  "step": 3275
822
  },
823
  {
824
- "epoch": 8.0,
825
  "learning_rate": 3.784444444444445e-06,
826
- "loss": 0.0034,
827
  "step": 3300
828
  },
829
  {
830
- "epoch": 8.01,
831
  "learning_rate": 3.728888888888889e-06,
832
- "loss": 0.003,
833
  "step": 3325
834
  },
835
  {
836
- "epoch": 8.01,
837
  "learning_rate": 3.673333333333334e-06,
838
- "loss": 0.0027,
839
  "step": 3350
840
  },
841
  {
842
- "epoch": 8.02,
843
  "learning_rate": 3.617777777777778e-06,
844
- "loss": 0.0026,
845
  "step": 3375
846
  },
847
  {
848
- "epoch": 8.02,
849
  "learning_rate": 3.5622222222222224e-06,
850
- "loss": 0.0028,
851
  "step": 3400
852
  },
853
  {
854
- "epoch": 8.03,
855
  "learning_rate": 3.5066666666666673e-06,
856
- "loss": 0.0029,
857
  "step": 3425
858
  },
859
  {
860
- "epoch": 8.03,
861
  "learning_rate": 3.4511111111111113e-06,
862
- "loss": 0.0027,
863
  "step": 3450
864
  },
865
  {
866
- "epoch": 8.04,
867
  "learning_rate": 3.3955555555555558e-06,
868
- "loss": 0.0022,
869
  "step": 3475
870
  },
871
  {
872
- "epoch": 8.04,
873
  "learning_rate": 3.3400000000000006e-06,
874
- "loss": 0.0024,
875
  "step": 3500
876
  },
877
  {
878
- "epoch": 8.05,
879
  "learning_rate": 3.2844444444444447e-06,
880
- "loss": 0.0028,
881
  "step": 3525
882
  },
883
  {
884
- "epoch": 8.05,
885
  "learning_rate": 3.228888888888889e-06,
886
- "loss": 0.0025,
887
  "step": 3550
888
  },
889
  {
890
- "epoch": 8.06,
891
  "learning_rate": 3.173333333333334e-06,
892
- "loss": 0.0022,
893
  "step": 3575
894
  },
895
  {
896
- "epoch": 8.06,
897
  "learning_rate": 3.117777777777778e-06,
898
- "loss": 0.0021,
899
  "step": 3600
900
  },
901
  {
902
- "epoch": 8.07,
903
  "learning_rate": 3.0622222222222225e-06,
904
- "loss": 0.002,
905
  "step": 3625
906
  },
907
  {
908
- "epoch": 8.07,
909
  "learning_rate": 3.0066666666666674e-06,
910
- "loss": 0.0019,
911
  "step": 3650
912
  },
913
  {
914
- "epoch": 8.08,
915
  "learning_rate": 2.9511111111111114e-06,
916
- "loss": 0.0019,
917
  "step": 3675
918
  },
919
  {
920
- "epoch": 8.08,
921
  "learning_rate": 2.895555555555556e-06,
922
- "loss": 0.0019,
923
  "step": 3700
924
  },
925
  {
926
- "epoch": 9.0,
927
  "learning_rate": 2.84e-06,
928
- "loss": 0.0021,
929
  "step": 3725
930
  },
931
  {
932
- "epoch": 9.01,
933
  "learning_rate": 2.784444444444445e-06,
934
- "loss": 0.0021,
935
  "step": 3750
936
  },
937
  {
938
- "epoch": 9.01,
939
  "learning_rate": 2.7288888888888893e-06,
940
- "loss": 0.0018,
941
  "step": 3775
942
  },
943
  {
944
- "epoch": 9.02,
945
  "learning_rate": 2.6733333333333333e-06,
946
- "loss": 0.0019,
947
  "step": 3800
948
  },
949
  {
950
- "epoch": 9.02,
951
  "learning_rate": 2.617777777777778e-06,
952
- "loss": 0.0018,
953
  "step": 3825
954
  },
955
  {
956
- "epoch": 9.03,
957
  "learning_rate": 2.5622222222222226e-06,
958
- "loss": 0.0021,
959
  "step": 3850
960
  },
961
  {
962
- "epoch": 9.03,
963
  "learning_rate": 2.5066666666666667e-06,
964
- "loss": 0.0017,
965
  "step": 3875
966
  },
967
  {
968
- "epoch": 9.04,
969
  "learning_rate": 2.451111111111111e-06,
970
- "loss": 0.0018,
971
  "step": 3900
972
  },
973
  {
974
- "epoch": 9.04,
975
  "learning_rate": 2.3955555555555556e-06,
976
- "loss": 0.0017,
977
  "step": 3925
978
  },
979
  {
980
- "epoch": 9.05,
981
  "learning_rate": 2.3400000000000005e-06,
982
- "loss": 0.002,
983
  "step": 3950
984
  },
985
  {
986
- "epoch": 9.05,
987
  "learning_rate": 2.2844444444444445e-06,
988
- "loss": 0.0017,
989
  "step": 3975
990
  },
991
  {
992
- "epoch": 9.06,
993
  "learning_rate": 2.228888888888889e-06,
994
- "loss": 0.0016,
995
  "step": 4000
996
  },
997
  {
998
- "epoch": 9.06,
999
- "eval_loss": 0.3812163174152374,
1000
- "eval_runtime": 968.0397,
1001
- "eval_samples_per_second": 6.809,
1002
- "eval_steps_per_second": 0.426,
1003
- "eval_wer": 18.775568066750374,
1004
  "step": 4000
1005
  },
1006
  {
1007
- "epoch": 9.06,
1008
  "learning_rate": 2.1733333333333334e-06,
1009
- "loss": 0.0016,
1010
  "step": 4025
1011
  },
1012
  {
1013
- "epoch": 9.07,
1014
  "learning_rate": 2.117777777777778e-06,
1015
- "loss": 0.0015,
1016
  "step": 4050
1017
  },
1018
  {
1019
- "epoch": 9.07,
1020
  "learning_rate": 2.0622222222222223e-06,
1021
- "loss": 0.0015,
1022
  "step": 4075
1023
  },
1024
  {
1025
- "epoch": 9.08,
1026
  "learning_rate": 2.006666666666667e-06,
1027
- "loss": 0.0015,
1028
  "step": 4100
1029
  },
1030
  {
1031
- "epoch": 10.0,
1032
  "learning_rate": 1.9511111111111113e-06,
1033
- "loss": 0.0015,
1034
  "step": 4125
1035
  },
1036
  {
1037
- "epoch": 10.01,
1038
  "learning_rate": 1.8955555555555557e-06,
1039
- "loss": 0.0016,
1040
  "step": 4150
1041
  },
1042
  {
1043
- "epoch": 10.01,
1044
  "learning_rate": 1.8400000000000002e-06,
1045
- "loss": 0.0015,
1046
  "step": 4175
1047
  },
1048
  {
1049
- "epoch": 10.02,
1050
  "learning_rate": 1.7844444444444444e-06,
1051
- "loss": 0.0015,
1052
  "step": 4200
1053
  },
1054
  {
1055
- "epoch": 10.02,
1056
  "learning_rate": 1.728888888888889e-06,
1057
- "loss": 0.0014,
1058
  "step": 4225
1059
  },
1060
  {
1061
- "epoch": 10.03,
1062
  "learning_rate": 1.6733333333333335e-06,
1063
- "loss": 0.0016,
1064
  "step": 4250
1065
  },
1066
  {
1067
- "epoch": 10.03,
1068
  "learning_rate": 1.6177777777777778e-06,
1069
- "loss": 0.0015,
1070
  "step": 4275
1071
  },
1072
  {
1073
- "epoch": 10.04,
1074
  "learning_rate": 1.5622222222222225e-06,
1075
- "loss": 0.0015,
1076
  "step": 4300
1077
  },
1078
  {
1079
- "epoch": 10.04,
1080
  "learning_rate": 1.506666666666667e-06,
1081
- "loss": 0.0016,
1082
  "step": 4325
1083
  },
1084
  {
1085
- "epoch": 10.05,
1086
  "learning_rate": 1.4511111111111112e-06,
1087
- "loss": 0.0015,
1088
  "step": 4350
1089
  },
1090
  {
1091
- "epoch": 10.05,
1092
  "learning_rate": 1.3955555555555556e-06,
1093
- "loss": 0.0014,
1094
  "step": 4375
1095
  },
1096
  {
1097
- "epoch": 10.06,
1098
  "learning_rate": 1.34e-06,
1099
- "loss": 0.0014,
1100
  "step": 4400
1101
  },
1102
  {
1103
- "epoch": 10.06,
1104
  "learning_rate": 1.2844444444444445e-06,
1105
- "loss": 0.0013,
1106
  "step": 4425
1107
  },
1108
  {
1109
- "epoch": 10.07,
1110
  "learning_rate": 1.228888888888889e-06,
1111
- "loss": 0.0013,
1112
  "step": 4450
1113
  },
1114
  {
1115
- "epoch": 10.07,
1116
  "learning_rate": 1.1733333333333335e-06,
1117
- "loss": 0.0013,
1118
  "step": 4475
1119
  },
1120
  {
1121
- "epoch": 10.08,
1122
  "learning_rate": 1.117777777777778e-06,
1123
- "loss": 0.0013,
1124
  "step": 4500
1125
  },
1126
  {
1127
- "epoch": 10.08,
1128
  "learning_rate": 1.0622222222222222e-06,
1129
- "loss": 0.0013,
1130
  "step": 4525
1131
  },
1132
  {
1133
- "epoch": 11.0,
1134
  "learning_rate": 1.0066666666666668e-06,
1135
- "loss": 0.0014,
1136
  "step": 4550
1137
  },
1138
  {
1139
- "epoch": 11.01,
1140
  "learning_rate": 9.511111111111111e-07,
1141
- "loss": 0.0014,
1142
  "step": 4575
1143
  },
1144
  {
1145
- "epoch": 11.01,
1146
  "learning_rate": 8.955555555555557e-07,
1147
- "loss": 0.0013,
1148
  "step": 4600
1149
  },
1150
  {
1151
- "epoch": 11.02,
1152
  "learning_rate": 8.400000000000001e-07,
1153
- "loss": 0.0013,
1154
  "step": 4625
1155
  },
1156
  {
1157
- "epoch": 11.02,
1158
  "learning_rate": 7.844444444444445e-07,
1159
- "loss": 0.0014,
1160
  "step": 4650
1161
  },
1162
  {
1163
- "epoch": 11.03,
1164
  "learning_rate": 7.28888888888889e-07,
1165
- "loss": 0.0015,
1166
  "step": 4675
1167
  },
1168
  {
1169
- "epoch": 11.03,
1170
  "learning_rate": 6.733333333333334e-07,
1171
- "loss": 0.0013,
1172
  "step": 4700
1173
  },
1174
  {
1175
- "epoch": 11.04,
1176
  "learning_rate": 6.177777777777778e-07,
1177
- "loss": 0.0013,
1178
  "step": 4725
1179
  },
1180
  {
1181
- "epoch": 11.04,
1182
  "learning_rate": 5.622222222222223e-07,
1183
- "loss": 0.0013,
1184
  "step": 4750
1185
  },
1186
  {
1187
- "epoch": 11.05,
1188
  "learning_rate": 5.066666666666667e-07,
1189
- "loss": 0.0013,
1190
  "step": 4775
1191
  },
1192
  {
1193
- "epoch": 11.05,
1194
  "learning_rate": 4.511111111111111e-07,
1195
- "loss": 0.0013,
1196
  "step": 4800
1197
  },
1198
  {
1199
- "epoch": 11.06,
1200
  "learning_rate": 3.9555555555555557e-07,
1201
- "loss": 0.0013,
1202
  "step": 4825
1203
  },
1204
  {
1205
- "epoch": 11.06,
1206
  "learning_rate": 3.4000000000000003e-07,
1207
- "loss": 0.0013,
1208
  "step": 4850
1209
  },
1210
  {
1211
- "epoch": 11.07,
1212
  "learning_rate": 2.844444444444445e-07,
1213
- "loss": 0.0013,
1214
  "step": 4875
1215
  },
1216
  {
1217
- "epoch": 11.07,
1218
  "learning_rate": 2.2888888888888892e-07,
1219
- "loss": 0.0012,
1220
  "step": 4900
1221
  },
1222
  {
1223
- "epoch": 11.08,
1224
  "learning_rate": 1.7333333333333335e-07,
1225
- "loss": 0.0012,
1226
  "step": 4925
1227
  },
1228
  {
1229
- "epoch": 12.0,
1230
  "learning_rate": 1.1777777777777778e-07,
1231
- "loss": 0.0013,
1232
  "step": 4950
1233
  },
1234
  {
1235
- "epoch": 12.01,
1236
  "learning_rate": 6.222222222222223e-08,
1237
- "loss": 0.0013,
1238
  "step": 4975
1239
  },
1240
  {
1241
- "epoch": 12.01,
1242
  "learning_rate": 6.666666666666667e-09,
1243
- "loss": 0.0012,
1244
  "step": 5000
1245
  },
1246
  {
1247
- "epoch": 12.01,
1248
- "eval_loss": 0.3913772702217102,
1249
- "eval_runtime": 973.2149,
1250
- "eval_samples_per_second": 6.772,
1251
- "eval_steps_per_second": 0.423,
1252
- "eval_wer": 18.83024828871157,
1253
  "step": 5000
1254
  },
1255
  {
1256
- "epoch": 12.01,
1257
  "step": 5000,
1258
- "total_flos": 4.616673800159232e+19,
1259
- "train_loss": 0.106446673027426,
1260
- "train_runtime": 27653.1068,
1261
- "train_samples_per_second": 5.786,
1262
- "train_steps_per_second": 0.181
1263
  }
1264
  ],
1265
  "max_steps": 5000,
1266
  "num_train_epochs": 9223372036854775807,
1267
- "total_flos": 4.616673800159232e+19,
1268
  "trial_name": null,
1269
  "trial_params": null
1270
  }
 
1
  {
2
+ "best_metric": 12.012786552211754,
3
+ "best_model_checkpoint": "./checkpoint-5000",
4
+ "epoch": 6.0604,
5
  "global_step": 5000,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
 
10
  {
11
  "epoch": 0.01,
12
  "learning_rate": 4.4e-07,
13
+ "loss": 1.6183,
14
  "step": 25
15
  },
16
  {
17
  "epoch": 0.01,
18
  "learning_rate": 9.400000000000001e-07,
19
+ "loss": 1.4856,
20
  "step": 50
21
  },
22
  {
23
  "epoch": 0.01,
24
  "learning_rate": 1.44e-06,
25
+ "loss": 1.1898,
26
  "step": 75
27
  },
28
  {
29
  "epoch": 0.02,
30
  "learning_rate": 1.94e-06,
31
+ "loss": 0.9354,
32
  "step": 100
33
  },
34
  {
35
  "epoch": 0.03,
36
  "learning_rate": 2.4400000000000004e-06,
37
+ "loss": 0.8172,
38
  "step": 125
39
  },
40
  {
41
  "epoch": 0.03,
42
  "learning_rate": 2.9400000000000002e-06,
43
+ "loss": 0.7386,
44
  "step": 150
45
  },
46
  {
47
  "epoch": 0.04,
48
  "learning_rate": 3.44e-06,
49
+ "loss": 0.6768,
50
  "step": 175
51
  },
52
  {
53
  "epoch": 0.04,
54
  "learning_rate": 3.94e-06,
55
+ "loss": 0.6069,
56
  "step": 200
57
  },
58
  {
59
  "epoch": 0.04,
60
  "learning_rate": 4.440000000000001e-06,
61
+ "loss": 0.5739,
62
  "step": 225
63
  },
64
  {
65
  "epoch": 0.05,
66
  "learning_rate": 4.94e-06,
67
+ "loss": 0.5269,
68
  "step": 250
69
  },
70
  {
71
  "epoch": 0.06,
72
  "learning_rate": 5.4400000000000004e-06,
73
+ "loss": 0.5011,
74
  "step": 275
75
  },
76
  {
77
  "epoch": 0.06,
78
  "learning_rate": 5.94e-06,
79
+ "loss": 0.4842,
80
  "step": 300
81
  },
82
  {
83
  "epoch": 0.07,
84
  "learning_rate": 6.440000000000001e-06,
85
+ "loss": 0.4698,
86
  "step": 325
87
  },
88
  {
89
  "epoch": 0.07,
90
  "learning_rate": 6.9400000000000005e-06,
91
+ "loss": 0.446,
92
  "step": 350
93
  },
94
  {
95
  "epoch": 0.07,
96
  "learning_rate": 7.440000000000001e-06,
97
+ "loss": 0.4378,
98
  "step": 375
99
  },
100
  {
101
  "epoch": 0.08,
102
  "learning_rate": 7.94e-06,
103
+ "loss": 0.3966,
104
  "step": 400
105
  },
106
  {
107
+ "epoch": 0.09,
108
  "learning_rate": 8.44e-06,
109
+ "loss": 0.3916,
110
  "step": 425
111
  },
112
  {
113
+ "epoch": 0.09,
114
  "learning_rate": 8.94e-06,
115
+ "loss": 0.3746,
116
  "step": 450
117
  },
118
  {
119
+ "epoch": 0.1,
120
  "learning_rate": 9.440000000000001e-06,
121
+ "loss": 0.3372,
122
  "step": 475
123
  },
124
  {
125
+ "epoch": 0.1,
126
  "learning_rate": 9.940000000000001e-06,
127
+ "loss": 0.329,
128
  "step": 500
129
  },
130
  {
131
+ "epoch": 0.1,
132
  "learning_rate": 9.951111111111111e-06,
133
+ "loss": 0.3364,
134
  "step": 525
135
  },
136
  {
137
+ "epoch": 0.11,
138
  "learning_rate": 9.895555555555557e-06,
139
+ "loss": 0.3074,
140
  "step": 550
141
  },
142
  {
143
+ "epoch": 0.12,
144
  "learning_rate": 9.84e-06,
145
+ "loss": 0.3134,
146
  "step": 575
147
  },
148
  {
149
+ "epoch": 0.12,
150
  "learning_rate": 9.784444444444445e-06,
151
+ "loss": 0.348,
152
  "step": 600
153
  },
154
  {
155
+ "epoch": 0.12,
156
  "learning_rate": 9.72888888888889e-06,
157
+ "loss": 0.3037,
158
  "step": 625
159
  },
160
  {
161
+ "epoch": 0.13,
162
  "learning_rate": 9.673333333333334e-06,
163
+ "loss": 0.2768,
164
  "step": 650
165
  },
166
  {
167
+ "epoch": 0.14,
168
  "learning_rate": 9.617777777777778e-06,
169
+ "loss": 0.3341,
170
  "step": 675
171
  },
172
  {
173
+ "epoch": 0.14,
174
  "learning_rate": 9.562222222222223e-06,
175
+ "loss": 0.3203,
176
  "step": 700
177
  },
178
  {
179
+ "epoch": 0.14,
180
  "learning_rate": 9.506666666666667e-06,
181
+ "loss": 0.2681,
182
  "step": 725
183
  },
184
  {
185
+ "epoch": 0.15,
186
  "learning_rate": 9.451111111111112e-06,
187
+ "loss": 0.2906,
188
  "step": 750
189
  },
190
  {
191
+ "epoch": 0.15,
192
  "learning_rate": 9.395555555555556e-06,
193
+ "loss": 0.2918,
194
  "step": 775
195
  },
196
  {
197
+ "epoch": 1.0,
198
  "learning_rate": 9.340000000000002e-06,
199
+ "loss": 0.2697,
200
  "step": 800
201
  },
202
  {
203
+ "epoch": 1.01,
204
  "learning_rate": 9.284444444444444e-06,
205
+ "loss": 0.2868,
206
  "step": 825
207
  },
208
  {
209
+ "epoch": 1.01,
210
  "learning_rate": 9.22888888888889e-06,
211
+ "loss": 0.2654,
212
  "step": 850
213
  },
214
  {
215
+ "epoch": 1.02,
216
  "learning_rate": 9.173333333333334e-06,
217
+ "loss": 0.2549,
218
  "step": 875
219
  },
220
  {
221
+ "epoch": 1.02,
222
  "learning_rate": 9.117777777777778e-06,
223
+ "loss": 0.2508,
224
  "step": 900
225
  },
226
  {
227
+ "epoch": 1.03,
228
  "learning_rate": 9.062222222222224e-06,
229
+ "loss": 0.2285,
230
  "step": 925
231
  },
232
  {
233
+ "epoch": 1.03,
234
  "learning_rate": 9.006666666666666e-06,
235
+ "loss": 0.2362,
236
  "step": 950
237
  },
238
  {
239
+ "epoch": 1.04,
240
  "learning_rate": 8.951111111111112e-06,
241
+ "loss": 0.2239,
242
  "step": 975
243
  },
244
  {
245
+ "epoch": 1.04,
246
  "learning_rate": 8.895555555555556e-06,
247
+ "loss": 0.2009,
248
  "step": 1000
249
  },
250
  {
251
+ "epoch": 1.04,
252
+ "eval_loss": 0.24461345374584198,
253
+ "eval_runtime": 1705.6758,
254
+ "eval_samples_per_second": 7.337,
255
+ "eval_steps_per_second": 0.459,
256
+ "eval_wer": 17.688063892927563,
257
  "step": 1000
258
  },
259
  {
260
+ "epoch": 1.05,
261
  "learning_rate": 8.84e-06,
262
+ "loss": 0.196,
263
  "step": 1025
264
  },
265
  {
266
+ "epoch": 1.05,
267
  "learning_rate": 8.784444444444446e-06,
268
+ "loss": 0.2057,
269
  "step": 1050
270
  },
271
  {
272
+ "epoch": 1.06,
273
  "learning_rate": 8.72888888888889e-06,
274
+ "loss": 0.2115,
275
  "step": 1075
276
  },
277
  {
278
+ "epoch": 1.06,
279
  "learning_rate": 8.673333333333334e-06,
280
+ "loss": 0.1891,
281
  "step": 1100
282
  },
283
  {
284
+ "epoch": 1.07,
285
  "learning_rate": 8.617777777777778e-06,
286
+ "loss": 0.1985,
287
  "step": 1125
288
  },
289
  {
290
+ "epoch": 1.07,
291
  "learning_rate": 8.562222222222224e-06,
292
+ "loss": 0.184,
293
  "step": 1150
294
  },
295
  {
296
+ "epoch": 1.08,
297
  "learning_rate": 8.506666666666668e-06,
298
+ "loss": 0.1581,
299
  "step": 1175
300
  },
301
  {
302
+ "epoch": 1.08,
303
  "learning_rate": 8.451111111111112e-06,
304
+ "loss": 0.1609,
305
  "step": 1200
306
  },
307
  {
308
+ "epoch": 1.09,
309
  "learning_rate": 8.395555555555557e-06,
310
+ "loss": 0.1528,
311
  "step": 1225
312
  },
313
  {
314
+ "epoch": 1.09,
315
  "learning_rate": 8.34e-06,
316
+ "loss": 0.1387,
317
  "step": 1250
318
  },
319
  {
320
+ "epoch": 1.1,
321
  "learning_rate": 8.284444444444446e-06,
322
+ "loss": 0.1312,
323
  "step": 1275
324
  },
325
  {
326
+ "epoch": 1.1,
327
  "learning_rate": 8.22888888888889e-06,
328
+ "loss": 0.1471,
329
  "step": 1300
330
  },
331
  {
332
+ "epoch": 1.11,
333
  "learning_rate": 8.173333333333334e-06,
334
+ "loss": 0.1208,
335
  "step": 1325
336
  },
337
  {
338
+ "epoch": 1.11,
339
  "learning_rate": 8.11777777777778e-06,
340
+ "loss": 0.1387,
341
  "step": 1350
342
  },
343
  {
344
+ "epoch": 1.12,
345
  "learning_rate": 8.062222222222222e-06,
346
+ "loss": 0.1393,
347
  "step": 1375
348
  },
349
  {
350
+ "epoch": 1.12,
351
  "learning_rate": 8.006666666666667e-06,
352
+ "loss": 0.1347,
353
  "step": 1400
354
  },
355
  {
356
+ "epoch": 1.13,
357
  "learning_rate": 7.951111111111111e-06,
358
+ "loss": 0.1278,
359
  "step": 1425
360
  },
361
  {
362
+ "epoch": 1.13,
363
  "learning_rate": 7.895555555555557e-06,
364
+ "loss": 0.1528,
365
  "step": 1450
366
  },
367
  {
368
+ "epoch": 1.14,
369
  "learning_rate": 7.840000000000001e-06,
370
+ "loss": 0.1614,
371
  "step": 1475
372
  },
373
  {
374
+ "epoch": 1.14,
375
  "learning_rate": 7.784444444444445e-06,
376
+ "loss": 0.1255,
377
  "step": 1500
378
  },
379
  {
380
+ "epoch": 1.15,
381
  "learning_rate": 7.72888888888889e-06,
382
+ "loss": 0.138,
383
  "step": 1525
384
  },
385
  {
386
+ "epoch": 1.15,
387
  "learning_rate": 7.673333333333333e-06,
388
+ "loss": 0.1341,
389
  "step": 1550
390
  },
391
  {
392
+ "epoch": 2.0,
393
  "learning_rate": 7.617777777777778e-06,
394
+ "loss": 0.1223,
395
  "step": 1575
396
  },
397
  {
398
+ "epoch": 2.01,
399
  "learning_rate": 7.562222222222223e-06,
400
+ "loss": 0.1272,
401
  "step": 1600
402
  },
403
  {
404
+ "epoch": 2.01,
405
  "learning_rate": 7.506666666666668e-06,
406
+ "loss": 0.1324,
407
  "step": 1625
408
  },
409
  {
410
+ "epoch": 2.02,
411
  "learning_rate": 7.451111111111111e-06,
412
+ "loss": 0.1277,
413
  "step": 1650
414
  },
415
  {
416
+ "epoch": 2.02,
417
  "learning_rate": 7.395555555555556e-06,
418
+ "loss": 0.1184,
419
  "step": 1675
420
  },
421
  {
422
+ "epoch": 2.03,
423
  "learning_rate": 7.340000000000001e-06,
424
+ "loss": 0.1117,
425
  "step": 1700
426
  },
427
  {
428
+ "epoch": 2.03,
429
  "learning_rate": 7.284444444444445e-06,
430
+ "loss": 0.1092,
431
  "step": 1725
432
  },
433
  {
434
+ "epoch": 2.04,
435
  "learning_rate": 7.22888888888889e-06,
436
+ "loss": 0.1027,
437
  "step": 1750
438
  },
439
  {
440
+ "epoch": 2.04,
441
  "learning_rate": 7.173333333333335e-06,
442
+ "loss": 0.1018,
443
  "step": 1775
444
  },
445
  {
446
+ "epoch": 2.05,
447
  "learning_rate": 7.117777777777778e-06,
448
+ "loss": 0.0909,
449
  "step": 1800
450
  },
451
  {
452
+ "epoch": 2.05,
453
  "learning_rate": 7.062222222222223e-06,
454
+ "loss": 0.104,
455
  "step": 1825
456
  },
457
  {
458
+ "epoch": 2.06,
459
  "learning_rate": 7.006666666666667e-06,
460
+ "loss": 0.1002,
461
  "step": 1850
462
  },
463
  {
464
+ "epoch": 2.06,
465
  "learning_rate": 6.951111111111112e-06,
466
+ "loss": 0.0964,
467
  "step": 1875
468
  },
469
  {
470
+ "epoch": 2.07,
471
  "learning_rate": 6.8955555555555565e-06,
472
+ "loss": 0.0874,
473
  "step": 1900
474
  },
475
  {
476
+ "epoch": 2.07,
477
  "learning_rate": 6.8400000000000014e-06,
478
+ "loss": 0.0885,
479
  "step": 1925
480
  },
481
  {
482
+ "epoch": 2.08,
483
  "learning_rate": 6.784444444444445e-06,
484
+ "loss": 0.084,
485
  "step": 1950
486
  },
487
  {
488
+ "epoch": 2.08,
489
  "learning_rate": 6.7288888888888895e-06,
490
+ "loss": 0.0766,
491
  "step": 1975
492
  },
493
  {
494
+ "epoch": 2.09,
495
  "learning_rate": 6.6733333333333335e-06,
496
+ "loss": 0.0759,
497
  "step": 2000
498
  },
499
  {
500
+ "epoch": 2.09,
501
+ "eval_loss": 0.21021738648414612,
502
+ "eval_runtime": 1689.1729,
503
+ "eval_samples_per_second": 7.409,
504
+ "eval_steps_per_second": 0.464,
505
+ "eval_wer": 14.258399888466212,
506
  "step": 2000
507
  },
508
  {
509
+ "epoch": 2.09,
510
  "learning_rate": 6.617777777777778e-06,
511
+ "loss": 0.0705,
512
  "step": 2025
513
  },
514
  {
515
+ "epoch": 2.1,
516
  "learning_rate": 6.562222222222223e-06,
517
+ "loss": 0.0634,
518
  "step": 2050
519
  },
520
  {
521
+ "epoch": 2.1,
522
  "learning_rate": 6.5066666666666665e-06,
523
+ "loss": 0.0671,
524
  "step": 2075
525
  },
526
  {
527
+ "epoch": 2.11,
528
  "learning_rate": 6.451111111111111e-06,
529
+ "loss": 0.0658,
530
  "step": 2100
531
  },
532
  {
533
+ "epoch": 2.11,
534
  "learning_rate": 6.395555555555556e-06,
535
+ "loss": 0.0601,
536
  "step": 2125
537
  },
538
  {
539
+ "epoch": 2.12,
540
  "learning_rate": 6.34e-06,
541
+ "loss": 0.0701,
542
  "step": 2150
543
  },
544
  {
545
+ "epoch": 2.12,
546
  "learning_rate": 6.284444444444445e-06,
547
+ "loss": 0.0642,
548
  "step": 2175
549
  },
550
  {
551
+ "epoch": 2.13,
552
  "learning_rate": 6.22888888888889e-06,
553
+ "loss": 0.0612,
554
  "step": 2200
555
  },
556
  {
557
+ "epoch": 2.13,
558
  "learning_rate": 6.173333333333333e-06,
559
+ "loss": 0.0571,
560
  "step": 2225
561
  },
562
  {
563
+ "epoch": 2.14,
564
  "learning_rate": 6.117777777777778e-06,
565
+ "loss": 0.1102,
566
  "step": 2250
567
  },
568
  {
569
+ "epoch": 2.14,
570
  "learning_rate": 6.062222222222223e-06,
571
+ "loss": 0.0744,
572
  "step": 2275
573
  },
574
  {
575
+ "epoch": 2.15,
576
  "learning_rate": 6.006666666666667e-06,
577
+ "loss": 0.0638,
578
  "step": 2300
579
  },
580
  {
581
+ "epoch": 2.15,
582
  "learning_rate": 5.951111111111112e-06,
583
+ "loss": 0.0638,
584
  "step": 2325
585
  },
586
  {
587
+ "epoch": 3.0,
588
  "learning_rate": 5.895555555555557e-06,
589
+ "loss": 0.0601,
590
  "step": 2350
591
  },
592
  {
593
+ "epoch": 3.01,
594
  "learning_rate": 5.84e-06,
595
+ "loss": 0.056,
596
  "step": 2375
597
  },
598
  {
599
+ "epoch": 3.01,
600
  "learning_rate": 5.784444444444445e-06,
601
+ "loss": 0.062,
602
  "step": 2400
603
  },
604
  {
605
+ "epoch": 3.02,
606
  "learning_rate": 5.72888888888889e-06,
607
+ "loss": 0.0645,
608
  "step": 2425
609
  },
610
  {
611
+ "epoch": 3.02,
612
  "learning_rate": 5.673333333333334e-06,
613
+ "loss": 0.0595,
614
  "step": 2450
615
  },
616
  {
617
+ "epoch": 3.03,
618
  "learning_rate": 5.617777777777779e-06,
619
+ "loss": 0.0543,
620
  "step": 2475
621
  },
622
  {
623
+ "epoch": 3.03,
624
  "learning_rate": 5.562222222222222e-06,
625
+ "loss": 0.0525,
626
  "step": 2500
627
  },
628
  {
629
+ "epoch": 3.04,
630
  "learning_rate": 5.506666666666667e-06,
631
+ "loss": 0.0494,
632
  "step": 2525
633
  },
634
  {
635
+ "epoch": 3.04,
636
  "learning_rate": 5.451111111111112e-06,
637
+ "loss": 0.0473,
638
  "step": 2550
639
  },
640
  {
641
+ "epoch": 3.05,
642
  "learning_rate": 5.3955555555555565e-06,
643
+ "loss": 0.0483,
644
  "step": 2575
645
  },
646
  {
647
+ "epoch": 3.05,
648
  "learning_rate": 5.3400000000000005e-06,
649
+ "loss": 0.0467,
650
  "step": 2600
651
  },
652
  {
653
+ "epoch": 3.06,
654
  "learning_rate": 5.2844444444444454e-06,
655
+ "loss": 0.0503,
656
  "step": 2625
657
  },
658
  {
659
+ "epoch": 3.06,
660
  "learning_rate": 5.228888888888889e-06,
661
+ "loss": 0.0428,
662
  "step": 2650
663
  },
664
  {
665
+ "epoch": 3.07,
666
  "learning_rate": 5.1733333333333335e-06,
667
+ "loss": 0.0418,
668
  "step": 2675
669
  },
670
  {
671
+ "epoch": 3.07,
672
  "learning_rate": 5.117777777777778e-06,
673
+ "loss": 0.0424,
674
  "step": 2700
675
  },
676
  {
677
+ "epoch": 3.08,
678
  "learning_rate": 5.062222222222222e-06,
679
+ "loss": 0.0406,
680
  "step": 2725
681
  },
682
  {
683
+ "epoch": 3.08,
684
  "learning_rate": 5.006666666666667e-06,
685
+ "loss": 0.0327,
686
  "step": 2750
687
  },
688
  {
689
+ "epoch": 3.09,
690
  "learning_rate": 4.951111111111111e-06,
691
+ "loss": 0.0371,
692
  "step": 2775
693
  },
694
  {
695
+ "epoch": 3.09,
696
  "learning_rate": 4.895555555555556e-06,
697
+ "loss": 0.0313,
698
  "step": 2800
699
  },
700
  {
701
+ "epoch": 3.1,
702
  "learning_rate": 4.84e-06,
703
+ "loss": 0.0295,
704
  "step": 2825
705
  },
706
  {
707
+ "epoch": 3.1,
708
  "learning_rate": 4.784444444444445e-06,
709
+ "loss": 0.0284,
710
  "step": 2850
711
  },
712
  {
713
+ "epoch": 3.11,
714
  "learning_rate": 4.728888888888889e-06,
715
+ "loss": 0.0322,
716
  "step": 2875
717
  },
718
  {
719
+ "epoch": 3.11,
720
  "learning_rate": 4.673333333333333e-06,
721
+ "loss": 0.0307,
722
  "step": 2900
723
  },
724
  {
725
+ "epoch": 3.12,
726
  "learning_rate": 4.617777777777778e-06,
727
+ "loss": 0.0316,
728
  "step": 2925
729
  },
730
  {
731
+ "epoch": 3.12,
732
  "learning_rate": 4.562222222222222e-06,
733
+ "loss": 0.0286,
734
  "step": 2950
735
  },
736
  {
737
+ "epoch": 3.13,
738
  "learning_rate": 4.506666666666667e-06,
739
+ "loss": 0.0289,
740
  "step": 2975
741
  },
742
  {
743
+ "epoch": 3.13,
744
  "learning_rate": 4.451111111111112e-06,
745
+ "loss": 0.0264,
746
  "step": 3000
747
  },
748
  {
749
+ "epoch": 3.13,
750
+ "eval_loss": 0.2200043797492981,
751
+ "eval_runtime": 1693.8446,
752
+ "eval_samples_per_second": 7.389,
753
+ "eval_steps_per_second": 0.462,
754
+ "eval_wer": 13.689776733254993,
755
  "step": 3000
756
  },
757
  {
758
+ "epoch": 3.14,
759
  "learning_rate": 4.395555555555556e-06,
760
+ "loss": 0.0609,
761
  "step": 3025
762
  },
763
  {
764
+ "epoch": 3.14,
765
  "learning_rate": 4.34e-06,
766
+ "loss": 0.0455,
767
  "step": 3050
768
  },
769
  {
770
+ "epoch": 3.15,
771
  "learning_rate": 4.284444444444445e-06,
772
+ "loss": 0.0323,
773
  "step": 3075
774
  },
775
  {
776
+ "epoch": 3.15,
777
  "learning_rate": 4.228888888888889e-06,
778
+ "loss": 0.0298,
779
  "step": 3100
780
  },
781
  {
782
+ "epoch": 3.16,
783
  "learning_rate": 4.173333333333334e-06,
784
+ "loss": 0.0265,
785
  "step": 3125
786
  },
787
  {
788
+ "epoch": 4.0,
789
  "learning_rate": 4.117777777777779e-06,
790
+ "loss": 0.1155,
791
  "step": 3150
792
  },
793
  {
794
+ "epoch": 4.01,
795
  "learning_rate": 4.062222222222223e-06,
796
+ "loss": 0.1149,
797
  "step": 3175
798
  },
799
  {
800
+ "epoch": 4.01,
801
  "learning_rate": 4.006666666666667e-06,
802
+ "loss": 0.1179,
803
  "step": 3200
804
  },
805
  {
806
+ "epoch": 4.02,
807
  "learning_rate": 3.951111111111112e-06,
808
+ "loss": 0.1228,
809
  "step": 3225
810
  },
811
  {
812
+ "epoch": 4.02,
813
  "learning_rate": 3.895555555555556e-06,
814
+ "loss": 0.1806,
815
  "step": 3250
816
  },
817
  {
818
+ "epoch": 4.03,
819
  "learning_rate": 3.8400000000000005e-06,
820
+ "loss": 0.2425,
821
  "step": 3275
822
  },
823
  {
824
+ "epoch": 4.03,
825
  "learning_rate": 3.784444444444445e-06,
826
+ "loss": 0.1122,
827
  "step": 3300
828
  },
829
  {
830
+ "epoch": 4.04,
831
  "learning_rate": 3.728888888888889e-06,
832
+ "loss": 0.0986,
833
  "step": 3325
834
  },
835
  {
836
+ "epoch": 4.04,
837
  "learning_rate": 3.673333333333334e-06,
838
+ "loss": 0.0942,
839
  "step": 3350
840
  },
841
  {
842
+ "epoch": 4.05,
843
  "learning_rate": 3.617777777777778e-06,
844
+ "loss": 0.1099,
845
  "step": 3375
846
  },
847
  {
848
+ "epoch": 4.05,
849
  "learning_rate": 3.5622222222222224e-06,
850
+ "loss": 0.1332,
851
  "step": 3400
852
  },
853
  {
854
+ "epoch": 4.06,
855
  "learning_rate": 3.5066666666666673e-06,
856
+ "loss": 0.1368,
857
  "step": 3425
858
  },
859
  {
860
+ "epoch": 4.06,
861
  "learning_rate": 3.4511111111111113e-06,
862
+ "loss": 0.164,
863
  "step": 3450
864
  },
865
  {
866
+ "epoch": 4.07,
867
  "learning_rate": 3.3955555555555558e-06,
868
+ "loss": 0.174,
869
  "step": 3475
870
  },
871
  {
872
+ "epoch": 4.07,
873
  "learning_rate": 3.3400000000000006e-06,
874
+ "loss": 0.1675,
875
  "step": 3500
876
  },
877
  {
878
+ "epoch": 4.08,
879
  "learning_rate": 3.2844444444444447e-06,
880
+ "loss": 0.0996,
881
  "step": 3525
882
  },
883
  {
884
+ "epoch": 4.08,
885
  "learning_rate": 3.228888888888889e-06,
886
+ "loss": 0.0795,
887
  "step": 3550
888
  },
889
  {
890
+ "epoch": 4.09,
891
  "learning_rate": 3.173333333333334e-06,
892
+ "loss": 0.0683,
893
  "step": 3575
894
  },
895
  {
896
+ "epoch": 4.09,
897
  "learning_rate": 3.117777777777778e-06,
898
+ "loss": 0.0635,
899
  "step": 3600
900
  },
901
  {
902
+ "epoch": 4.1,
903
  "learning_rate": 3.0622222222222225e-06,
904
+ "loss": 0.0751,
905
  "step": 3625
906
  },
907
  {
908
+ "epoch": 4.1,
909
  "learning_rate": 3.0066666666666674e-06,
910
+ "loss": 0.0712,
911
  "step": 3650
912
  },
913
  {
914
+ "epoch": 4.11,
915
  "learning_rate": 2.9511111111111114e-06,
916
+ "loss": 0.0703,
917
  "step": 3675
918
  },
919
  {
920
+ "epoch": 4.11,
921
  "learning_rate": 2.895555555555556e-06,
922
+ "loss": 0.1152,
923
  "step": 3700
924
  },
925
  {
926
+ "epoch": 4.12,
927
  "learning_rate": 2.84e-06,
928
+ "loss": 0.1039,
929
  "step": 3725
930
  },
931
  {
932
+ "epoch": 4.12,
933
  "learning_rate": 2.784444444444445e-06,
934
+ "loss": 0.0863,
935
  "step": 3750
936
  },
937
  {
938
+ "epoch": 4.13,
939
  "learning_rate": 2.7288888888888893e-06,
940
+ "loss": 0.0882,
941
  "step": 3775
942
  },
943
  {
944
+ "epoch": 4.13,
945
  "learning_rate": 2.6733333333333333e-06,
946
+ "loss": 0.0677,
947
  "step": 3800
948
  },
949
  {
950
+ "epoch": 4.14,
951
  "learning_rate": 2.617777777777778e-06,
952
+ "loss": 0.0511,
953
  "step": 3825
954
  },
955
  {
956
+ "epoch": 4.14,
957
  "learning_rate": 2.5622222222222226e-06,
958
+ "loss": 0.0283,
959
  "step": 3850
960
  },
961
  {
962
+ "epoch": 4.15,
963
  "learning_rate": 2.5066666666666667e-06,
964
+ "loss": 0.0246,
965
  "step": 3875
966
  },
967
  {
968
+ "epoch": 4.15,
969
  "learning_rate": 2.451111111111111e-06,
970
+ "loss": 0.0231,
971
  "step": 3900
972
  },
973
  {
974
+ "epoch": 5.0,
975
  "learning_rate": 2.3955555555555556e-06,
976
+ "loss": 0.0442,
977
  "step": 3925
978
  },
979
  {
980
+ "epoch": 5.01,
981
  "learning_rate": 2.3400000000000005e-06,
982
+ "loss": 0.063,
983
  "step": 3950
984
  },
985
  {
986
+ "epoch": 5.01,
987
  "learning_rate": 2.2844444444444445e-06,
988
+ "loss": 0.0548,
989
  "step": 3975
990
  },
991
  {
992
+ "epoch": 5.02,
993
  "learning_rate": 2.228888888888889e-06,
994
+ "loss": 0.0633,
995
  "step": 4000
996
  },
997
  {
998
+ "epoch": 5.02,
999
+ "eval_loss": 0.1954876184463501,
1000
+ "eval_runtime": 1710.419,
1001
+ "eval_samples_per_second": 7.317,
1002
+ "eval_steps_per_second": 0.458,
1003
+ "eval_wer": 12.553526260232228,
1004
  "step": 4000
1005
  },
1006
  {
1007
+ "epoch": 5.02,
1008
  "learning_rate": 2.1733333333333334e-06,
1009
+ "loss": 0.0801,
1010
  "step": 4025
1011
  },
1012
  {
1013
+ "epoch": 5.03,
1014
  "learning_rate": 2.117777777777778e-06,
1015
+ "loss": 0.1296,
1016
  "step": 4050
1017
  },
1018
  {
1019
+ "epoch": 5.03,
1020
  "learning_rate": 2.0622222222222223e-06,
1021
+ "loss": 0.0944,
1022
  "step": 4075
1023
  },
1024
  {
1025
+ "epoch": 5.04,
1026
  "learning_rate": 2.006666666666667e-06,
1027
+ "loss": 0.0541,
1028
  "step": 4100
1029
  },
1030
  {
1031
+ "epoch": 5.04,
1032
  "learning_rate": 1.9511111111111113e-06,
1033
+ "loss": 0.0511,
1034
  "step": 4125
1035
  },
1036
  {
1037
+ "epoch": 5.05,
1038
  "learning_rate": 1.8955555555555557e-06,
1039
+ "loss": 0.0524,
1040
  "step": 4150
1041
  },
1042
  {
1043
+ "epoch": 5.05,
1044
  "learning_rate": 1.8400000000000002e-06,
1045
+ "loss": 0.0656,
1046
  "step": 4175
1047
  },
1048
  {
1049
+ "epoch": 5.06,
1050
  "learning_rate": 1.7844444444444444e-06,
1051
+ "loss": 0.0675,
1052
  "step": 4200
1053
  },
1054
  {
1055
+ "epoch": 5.06,
1056
  "learning_rate": 1.728888888888889e-06,
1057
+ "loss": 0.0916,
1058
  "step": 4225
1059
  },
1060
  {
1061
+ "epoch": 5.07,
1062
  "learning_rate": 1.6733333333333335e-06,
1063
+ "loss": 0.0977,
1064
  "step": 4250
1065
  },
1066
  {
1067
+ "epoch": 5.07,
1068
  "learning_rate": 1.6177777777777778e-06,
1069
+ "loss": 0.108,
1070
  "step": 4275
1071
  },
1072
  {
1073
+ "epoch": 5.08,
1074
  "learning_rate": 1.5622222222222225e-06,
1075
+ "loss": 0.0672,
1076
  "step": 4300
1077
  },
1078
  {
1079
+ "epoch": 5.08,
1080
  "learning_rate": 1.506666666666667e-06,
1081
+ "loss": 0.0444,
1082
  "step": 4325
1083
  },
1084
  {
1085
+ "epoch": 5.09,
1086
  "learning_rate": 1.4511111111111112e-06,
1087
+ "loss": 0.0371,
1088
  "step": 4350
1089
  },
1090
  {
1091
+ "epoch": 5.09,
1092
  "learning_rate": 1.3955555555555556e-06,
1093
+ "loss": 0.0368,
1094
  "step": 4375
1095
  },
1096
  {
1097
+ "epoch": 5.1,
1098
  "learning_rate": 1.34e-06,
1099
+ "loss": 0.038,
1100
  "step": 4400
1101
  },
1102
  {
1103
+ "epoch": 5.1,
1104
  "learning_rate": 1.2844444444444445e-06,
1105
+ "loss": 0.0366,
1106
  "step": 4425
1107
  },
1108
  {
1109
+ "epoch": 5.11,
1110
  "learning_rate": 1.228888888888889e-06,
1111
+ "loss": 0.041,
1112
  "step": 4450
1113
  },
1114
  {
1115
+ "epoch": 5.11,
1116
  "learning_rate": 1.1733333333333335e-06,
1117
+ "loss": 0.0613,
1118
  "step": 4475
1119
  },
1120
  {
1121
+ "epoch": 5.12,
1122
  "learning_rate": 1.117777777777778e-06,
1123
+ "loss": 0.06,
1124
  "step": 4500
1125
  },
1126
  {
1127
+ "epoch": 5.12,
1128
  "learning_rate": 1.0622222222222222e-06,
1129
+ "loss": 0.0519,
1130
  "step": 4525
1131
  },
1132
  {
1133
+ "epoch": 5.13,
1134
  "learning_rate": 1.0066666666666668e-06,
1135
+ "loss": 0.0475,
1136
  "step": 4550
1137
  },
1138
  {
1139
+ "epoch": 5.13,
1140
  "learning_rate": 9.511111111111111e-07,
1141
+ "loss": 0.0328,
1142
  "step": 4575
1143
  },
1144
  {
1145
+ "epoch": 5.14,
1146
  "learning_rate": 8.955555555555557e-07,
1147
+ "loss": 0.046,
1148
  "step": 4600
1149
  },
1150
  {
1151
+ "epoch": 5.14,
1152
  "learning_rate": 8.400000000000001e-07,
1153
+ "loss": 0.0184,
1154
  "step": 4625
1155
  },
1156
  {
1157
+ "epoch": 5.15,
1158
  "learning_rate": 7.844444444444445e-07,
1159
+ "loss": 0.014,
1160
  "step": 4650
1161
  },
1162
  {
1163
+ "epoch": 5.15,
1164
  "learning_rate": 7.28888888888889e-07,
1165
+ "loss": 0.0138,
1166
  "step": 4675
1167
  },
1168
  {
1169
+ "epoch": 6.0,
1170
  "learning_rate": 6.733333333333334e-07,
1171
+ "loss": 0.0116,
1172
  "step": 4700
1173
  },
1174
  {
1175
+ "epoch": 6.01,
1176
  "learning_rate": 6.177777777777778e-07,
1177
+ "loss": 0.014,
1178
  "step": 4725
1179
  },
1180
  {
1181
+ "epoch": 6.01,
1182
  "learning_rate": 5.622222222222223e-07,
1183
+ "loss": 0.0156,
1184
  "step": 4750
1185
  },
1186
  {
1187
+ "epoch": 6.02,
1188
  "learning_rate": 5.066666666666667e-07,
1189
+ "loss": 0.0136,
1190
  "step": 4775
1191
  },
1192
  {
1193
+ "epoch": 6.02,
1194
  "learning_rate": 4.511111111111111e-07,
1195
+ "loss": 0.0144,
1196
  "step": 4800
1197
  },
1198
  {
1199
+ "epoch": 6.03,
1200
  "learning_rate": 3.9555555555555557e-07,
1201
+ "loss": 0.0155,
1202
  "step": 4825
1203
  },
1204
  {
1205
+ "epoch": 6.03,
1206
  "learning_rate": 3.4000000000000003e-07,
1207
+ "loss": 0.0187,
1208
  "step": 4850
1209
  },
1210
  {
1211
+ "epoch": 6.04,
1212
  "learning_rate": 2.844444444444445e-07,
1213
+ "loss": 0.0189,
1214
  "step": 4875
1215
  },
1216
  {
1217
+ "epoch": 6.04,
1218
  "learning_rate": 2.2888888888888892e-07,
1219
+ "loss": 0.0187,
1220
  "step": 4900
1221
  },
1222
  {
1223
+ "epoch": 6.05,
1224
  "learning_rate": 1.7333333333333335e-07,
1225
+ "loss": 0.0194,
1226
  "step": 4925
1227
  },
1228
  {
1229
+ "epoch": 6.05,
1230
  "learning_rate": 1.1777777777777778e-07,
1231
+ "loss": 0.019,
1232
  "step": 4950
1233
  },
1234
  {
1235
+ "epoch": 6.06,
1236
  "learning_rate": 6.222222222222223e-08,
1237
+ "loss": 0.0183,
1238
  "step": 4975
1239
  },
1240
  {
1241
+ "epoch": 6.06,
1242
  "learning_rate": 6.666666666666667e-09,
1243
+ "loss": 0.0199,
1244
  "step": 5000
1245
  },
1246
  {
1247
+ "epoch": 6.06,
1248
+ "eval_loss": 0.19964517652988434,
1249
+ "eval_runtime": 1705.0556,
1250
+ "eval_samples_per_second": 7.34,
1251
+ "eval_steps_per_second": 0.459,
1252
+ "eval_wer": 12.012786552211754,
1253
  "step": 5000
1254
  },
1255
  {
1256
+ "epoch": 6.06,
1257
  "step": 5000,
1258
+ "total_flos": 4.612864472875008e+19,
1259
+ "train_loss": 0.15837200704216958,
1260
+ "train_runtime": 29005.3522,
1261
+ "train_samples_per_second": 5.516,
1262
+ "train_steps_per_second": 0.172
1263
  }
1264
  ],
1265
  "max_steps": 5000,
1266
  "num_train_epochs": 9223372036854775807,
1267
+ "total_flos": 4.612864472875008e+19,
1268
  "trial_name": null,
1269
  "trial_params": null
1270
  }