Hugging Face's logo

voidful
/
wav2vec2-xlsr-53-espeak-librispeech-ft

Automatic Speech Recognition
Transformers
PyTorch
Safetensors
wav2vec2
Inference Endpoints
Model card Files Community
1
voidful commited on
Commit
00841b1
·
1 Parent(s): 33f58b0

update model train librispeech 960h

Browse files
Files changed (8) hide show
  1. config.json +1 -1
  2. optimizer.pt +1 -1
  3. pytorch_model.bin +1 -1
  4. rng_state.pth +2 -2
  5. scaler.pt +1 -1
  6. scheduler.pt +1 -1
  7. trainer_state.json +768 -2518
  8. training_args.bin +1 -1
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "facebook/wav2vec2-xlsr-53-espeak-cv-ft",
3
  "activation_dropout": 0.0,
4
  "adapter_kernel_size": 3,
5
  "adapter_stride": 2,
 
1
  {
2
+ "_name_or_path": "voidful/wav2vec2-xlsr-53-espeak-librispeech-ft",
3
  "activation_dropout": 0.0,
4
  "adapter_kernel_size": 3,
5
  "adapter_stride": 2,
optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:72c454d705307858a64501b4f3a2ff951c5bcc69988e28a321278d83ffc8821f
3
  size 2526970977
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dbb9f80648af2e4a75fd6bc270b240715c8ab228cdd4ca8f8ccca104ef11d12c
3
  size 2526970977
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:913b32dd39aa285c85539b7204a6c6fa0095ed7c0119a4813a63915d9f0d0987
3
  size 1263505905
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e38eadd5ca7d259c6a2eaf8f9de793fa359b07101bd0d44d7efd9b4e48aba87f
3
  size 1263505905
rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2ab2d87dbfe0578659f3eb0d2fa7d82725e633973524a5b1084e84fa711bfd9b
3
- size 16607
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4b0f6239b0e06cbe40f14fcde2063b022ff4b2b975c7fecf7a0ec7c82bcb437
3
+ size 16543
scaler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7eb683c2463b79fd5fe01c07aca28a71e50a471eca615f5459c9c813160a5d3c
3
  size 559
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dbe91f9306df798d3d2e4cffb541a8fa17d343a4e30e2c3aba8f813440b4d15d
3
  size 559
scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e99c9a35af79b7bf9789cdf16efad705b985669a66da18f1a3f77aade805a3b8
3
  size 623
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:650d0683edf363b9d9227770db847cedd29bab07c49052d274437fcc0a3bab6d
3
  size 623
trainer_state.json CHANGED
@@ -1,3266 +1,1516 @@
1
  {
2
- "best_metric": 0.0955534353852272,
3
- "best_model_checkpoint": "facebook/wav2vec2-xlsr-53-espeak-cv-ft-clean/checkpoint-4000",
4
- "epoch": 7.680945347119645,
5
- "global_step": 5200,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
- "epoch": 0.01,
12
- "learning_rate": 1.6380000000000002e-05,
13
- "loss": 1.8358,
14
  "step": 10
15
  },
16
  {
17
- "epoch": 0.03,
18
- "learning_rate": 3.978e-05,
19
- "loss": 1.5868,
20
  "step": 20
21
  },
22
  {
23
- "epoch": 0.04,
24
- "learning_rate": 6.318e-05,
25
- "loss": 1.4356,
26
  "step": 30
27
  },
28
  {
29
- "epoch": 0.06,
30
- "learning_rate": 8.658e-05,
31
- "loss": 1.2539,
32
  "step": 40
33
  },
34
  {
35
- "epoch": 0.07,
36
- "learning_rate": 0.00010997999999999999,
37
- "loss": 1.2239,
38
  "step": 50
39
  },
40
  {
41
- "epoch": 0.09,
42
- "learning_rate": 0.00013337999999999997,
43
- "loss": 0.966,
44
  "step": 60
45
  },
46
  {
47
- "epoch": 0.1,
48
- "learning_rate": 0.00015678,
49
- "loss": 0.9195,
50
  "step": 70
51
  },
52
  {
53
- "epoch": 0.12,
54
- "learning_rate": 0.00018018,
55
- "loss": 0.744,
56
  "step": 80
57
  },
58
  {
59
- "epoch": 0.13,
60
- "learning_rate": 0.00020358,
61
- "loss": 0.6638,
62
  "step": 90
63
  },
64
  {
65
- "epoch": 0.15,
66
- "learning_rate": 0.00022698,
67
- "loss": 0.6554,
68
  "step": 100
69
  },
70
  {
71
- "epoch": 0.16,
72
- "learning_rate": 0.00023395957551826258,
73
- "loss": 0.5807,
74
  "step": 110
75
  },
76
  {
77
- "epoch": 0.18,
78
- "learning_rate": 0.00023390182625863772,
79
- "loss": 0.4771,
80
  "step": 120
81
  },
82
  {
83
- "epoch": 0.19,
84
- "learning_rate": 0.00023384407699901283,
85
- "loss": 0.4634,
86
  "step": 130
87
  },
88
  {
89
- "epoch": 0.21,
90
- "learning_rate": 0.00023378632773938794,
91
- "loss": 0.5991,
92
  "step": 140
93
  },
94
  {
95
- "epoch": 0.22,
96
- "learning_rate": 0.00023372857847976308,
97
- "loss": 0.5512,
98
  "step": 150
99
  },
100
  {
101
- "epoch": 0.24,
102
- "learning_rate": 0.0002336708292201382,
103
- "loss": 0.4616,
104
  "step": 160
105
  },
106
  {
107
- "epoch": 0.25,
108
- "learning_rate": 0.0002336130799605133,
109
- "loss": 0.4028,
110
  "step": 170
111
  },
112
  {
113
- "epoch": 0.27,
114
- "learning_rate": 0.00023355533070088845,
115
- "loss": 0.386,
116
  "step": 180
117
  },
118
  {
119
- "epoch": 0.28,
120
- "learning_rate": 0.00023349758144126356,
121
- "loss": 0.3513,
122
  "step": 190
123
  },
124
  {
125
- "epoch": 0.3,
126
- "learning_rate": 0.00023343983218163868,
127
- "loss": 0.4561,
128
  "step": 200
129
  },
130
  {
131
- "epoch": 0.31,
132
- "learning_rate": 0.00023338208292201382,
133
- "loss": 0.3157,
134
  "step": 210
135
  },
136
  {
137
- "epoch": 0.32,
138
- "learning_rate": 0.00023332433366238893,
139
- "loss": 0.2983,
140
  "step": 220
141
  },
142
  {
143
- "epoch": 0.34,
144
- "learning_rate": 0.00023326658440276404,
145
- "loss": 0.3078,
146
  "step": 230
147
  },
148
  {
149
- "epoch": 0.35,
150
- "learning_rate": 0.00023320883514313918,
151
- "loss": 0.3487,
152
  "step": 240
153
  },
154
  {
155
- "epoch": 0.37,
156
- "learning_rate": 0.0002331510858835143,
157
- "loss": 0.4429,
158
  "step": 250
159
  },
160
  {
161
- "epoch": 0.38,
162
- "learning_rate": 0.0002330933366238894,
163
- "loss": 0.3194,
164
  "step": 260
165
  },
166
  {
167
- "epoch": 0.4,
168
- "learning_rate": 0.00023303558736426457,
169
- "loss": 0.3121,
170
  "step": 270
171
  },
172
  {
173
- "epoch": 0.41,
174
- "learning_rate": 0.0002329778381046397,
175
- "loss": 0.3229,
176
  "step": 280
177
  },
178
  {
179
- "epoch": 0.43,
180
- "learning_rate": 0.00023292008884501483,
181
- "loss": 0.317,
182
  "step": 290
183
  },
184
  {
185
- "epoch": 0.44,
186
- "learning_rate": 0.00023286233958538994,
187
- "loss": 0.3877,
188
  "step": 300
189
  },
190
  {
191
- "epoch": 0.46,
192
- "learning_rate": 0.00023280459032576505,
193
- "loss": 0.2582,
194
  "step": 310
195
  },
196
  {
197
- "epoch": 0.47,
198
- "learning_rate": 0.0002327468410661402,
199
- "loss": 0.2814,
200
  "step": 320
201
  },
202
  {
203
- "epoch": 0.49,
204
- "learning_rate": 0.0002326890918065153,
205
- "loss": 0.2822,
206
  "step": 330
207
  },
208
  {
209
- "epoch": 0.5,
210
- "learning_rate": 0.00023263134254689042,
211
- "loss": 0.3077,
212
  "step": 340
213
  },
214
  {
215
- "epoch": 0.52,
216
- "learning_rate": 0.00023257359328726556,
217
- "loss": 0.3531,
218
  "step": 350
219
  },
220
  {
221
- "epoch": 0.53,
222
- "learning_rate": 0.00023251584402764067,
223
- "loss": 0.272,
224
  "step": 360
225
  },
226
  {
227
- "epoch": 0.55,
228
- "learning_rate": 0.00023245809476801578,
229
- "loss": 0.3182,
230
  "step": 370
231
  },
232
  {
233
- "epoch": 0.56,
234
- "learning_rate": 0.00023240034550839092,
235
- "loss": 0.2482,
236
  "step": 380
237
  },
238
  {
239
- "epoch": 0.58,
240
- "learning_rate": 0.00023234259624876604,
241
- "loss": 0.2565,
242
  "step": 390
243
  },
244
  {
245
- "epoch": 0.59,
246
- "learning_rate": 0.00023228484698914115,
247
- "loss": 0.3474,
248
  "step": 400
249
  },
250
  {
251
- "epoch": 0.59,
252
- "eval_cer": 0.20053818094839443,
253
- "eval_loss": 0.12517954409122467,
254
- "eval_runtime": 167.3283,
255
- "eval_samples_per_second": 14.23,
256
- "eval_steps_per_second": 0.478,
257
- "eval_wer": 0.996220075598488,
258
  "step": 400
259
  },
260
  {
261
- "epoch": 0.61,
262
- "learning_rate": 0.0002322270977295163,
263
- "loss": 0.2553,
264
  "step": 410
265
  },
266
  {
267
- "epoch": 0.62,
268
- "learning_rate": 0.0002321693484698914,
269
- "loss": 0.29,
270
  "step": 420
271
  },
272
  {
273
- "epoch": 0.64,
274
- "learning_rate": 0.00023211159921026654,
275
- "loss": 0.2583,
276
  "step": 430
277
  },
278
  {
279
- "epoch": 0.65,
280
- "learning_rate": 0.00023205384995064166,
281
- "loss": 0.2701,
282
  "step": 440
283
  },
284
  {
285
- "epoch": 0.66,
286
- "learning_rate": 0.00023199610069101677,
287
- "loss": 0.3269,
288
  "step": 450
289
  },
290
  {
291
- "epoch": 0.68,
292
- "learning_rate": 0.0002319383514313919,
293
- "loss": 0.256,
294
  "step": 460
295
  },
296
  {
297
- "epoch": 0.69,
298
- "learning_rate": 0.00023188060217176702,
299
- "loss": 0.277,
300
  "step": 470
301
  },
302
  {
303
- "epoch": 0.71,
304
- "learning_rate": 0.00023182285291214214,
305
- "loss": 0.2547,
306
  "step": 480
307
  },
308
  {
309
- "epoch": 0.72,
310
- "learning_rate": 0.00023176510365251728,
311
- "loss": 0.2853,
312
  "step": 490
313
  },
314
  {
315
- "epoch": 0.74,
316
- "learning_rate": 0.0002317073543928924,
317
- "loss": 0.3382,
318
  "step": 500
319
  },
320
  {
321
- "epoch": 0.75,
322
- "learning_rate": 0.0002316496051332675,
323
- "loss": 0.2393,
324
  "step": 510
325
  },
326
  {
327
- "epoch": 0.77,
328
- "learning_rate": 0.00023159185587364264,
329
- "loss": 0.2374,
330
  "step": 520
331
  },
332
  {
333
- "epoch": 0.78,
334
- "learning_rate": 0.00023153410661401775,
335
- "loss": 0.2067,
336
  "step": 530
337
  },
338
  {
339
- "epoch": 0.8,
340
- "learning_rate": 0.00023147635735439287,
341
- "loss": 0.268,
342
  "step": 540
343
  },
344
  {
345
- "epoch": 0.81,
346
- "learning_rate": 0.000231418608094768,
347
- "loss": 0.3156,
348
  "step": 550
349
  },
350
  {
351
- "epoch": 0.83,
352
- "learning_rate": 0.00023136085883514312,
353
- "loss": 0.223,
354
  "step": 560
355
  },
356
  {
357
- "epoch": 0.84,
358
- "learning_rate": 0.00023130310957551823,
359
- "loss": 0.2374,
360
  "step": 570
361
  },
362
  {
363
- "epoch": 0.86,
364
- "learning_rate": 0.0002312453603158934,
365
- "loss": 0.2144,
366
  "step": 580
367
  },
368
  {
369
- "epoch": 0.87,
370
- "learning_rate": 0.0002311876110562685,
371
- "loss": 0.2324,
372
  "step": 590
373
  },
374
  {
375
- "epoch": 0.89,
376
- "learning_rate": 0.00023112986179664365,
377
- "loss": 0.3466,
378
  "step": 600
379
  },
380
  {
381
- "epoch": 0.9,
382
- "learning_rate": 0.00023107211253701877,
383
- "loss": 0.2559,
384
  "step": 610
385
  },
386
  {
387
- "epoch": 0.92,
388
- "learning_rate": 0.00023101436327739388,
389
- "loss": 0.2261,
390
  "step": 620
391
  },
392
  {
393
- "epoch": 0.93,
394
- "learning_rate": 0.00023095661401776902,
395
- "loss": 0.2305,
396
  "step": 630
397
  },
398
  {
399
- "epoch": 0.95,
400
- "learning_rate": 0.00023089886475814413,
401
- "loss": 0.2357,
402
  "step": 640
403
  },
404
  {
405
- "epoch": 0.96,
406
- "learning_rate": 0.00023084111549851924,
407
- "loss": 0.356,
408
  "step": 650
409
  },
410
  {
411
- "epoch": 0.97,
412
- "learning_rate": 0.00023078336623889438,
413
- "loss": 0.2131,
414
  "step": 660
415
  },
416
  {
417
- "epoch": 0.99,
418
- "learning_rate": 0.0002307256169792695,
419
- "loss": 0.2462,
420
  "step": 670
421
  },
422
  {
423
- "epoch": 1.0,
424
- "learning_rate": 0.0002306678677196446,
425
- "loss": 0.2411,
426
  "step": 680
427
  },
428
  {
429
- "epoch": 1.02,
430
- "learning_rate": 0.00023061011846001975,
431
- "loss": 0.209,
432
  "step": 690
433
  },
434
  {
435
- "epoch": 1.03,
436
- "learning_rate": 0.00023055236920039486,
437
- "loss": 0.1952,
438
  "step": 700
439
  },
440
  {
441
- "epoch": 1.05,
442
- "learning_rate": 0.00023049461994076998,
443
- "loss": 0.1918,
444
  "step": 710
445
  },
446
  {
447
- "epoch": 1.06,
448
- "learning_rate": 0.00023043687068114512,
449
- "loss": 0.2332,
450
  "step": 720
451
  },
452
  {
453
- "epoch": 1.08,
454
- "learning_rate": 0.00023037912142152023,
455
- "loss": 0.2301,
456
  "step": 730
457
  },
458
  {
459
- "epoch": 1.09,
460
- "learning_rate": 0.00023032137216189537,
461
- "loss": 0.2231,
462
  "step": 740
463
  },
464
  {
465
- "epoch": 1.11,
466
- "learning_rate": 0.00023026362290227048,
467
- "loss": 0.2176,
468
  "step": 750
469
  },
470
  {
471
- "epoch": 1.12,
472
- "learning_rate": 0.0002302058736426456,
473
- "loss": 0.1718,
474
  "step": 760
475
  },
476
  {
477
- "epoch": 1.14,
478
- "learning_rate": 0.00023014812438302074,
479
- "loss": 0.213,
480
  "step": 770
481
  },
482
  {
483
- "epoch": 1.15,
484
- "learning_rate": 0.00023009037512339585,
485
- "loss": 0.2865,
486
  "step": 780
487
  },
488
  {
489
- "epoch": 1.17,
490
- "learning_rate": 0.00023003262586377096,
491
- "loss": 0.2347,
492
  "step": 790
493
  },
494
  {
495
- "epoch": 1.18,
496
- "learning_rate": 0.0002299748766041461,
497
- "loss": 0.2116,
498
  "step": 800
499
  },
500
  {
501
- "epoch": 1.18,
502
- "eval_cer": 0.16925790827004725,
503
- "eval_loss": 0.10652817785739899,
504
- "eval_runtime": 166.4912,
505
- "eval_samples_per_second": 14.301,
506
- "eval_steps_per_second": 0.481,
507
- "eval_wer": 0.9886602267954641,
508
  "step": 800
509
  },
510
  {
511
- "epoch": 1.2,
512
- "learning_rate": 0.0002299229022704837,
513
- "loss": 0.1885,
514
  "step": 810
515
  },
516
  {
517
- "epoch": 1.21,
518
- "learning_rate": 0.00022986515301085883,
519
- "loss": 0.1934,
520
  "step": 820
521
  },
522
  {
523
- "epoch": 1.23,
524
- "learning_rate": 0.00022980740375123395,
525
- "loss": 0.305,
526
  "step": 830
527
  },
528
  {
529
- "epoch": 1.24,
530
- "learning_rate": 0.00022974965449160906,
531
- "loss": 0.2144,
532
  "step": 840
533
  },
534
  {
535
- "epoch": 1.26,
536
- "learning_rate": 0.0002296919052319842,
537
- "loss": 0.2157,
538
  "step": 850
539
  },
540
  {
541
- "epoch": 1.27,
542
- "learning_rate": 0.0002296341559723593,
543
- "loss": 0.2149,
544
  "step": 860
545
  },
546
  {
547
- "epoch": 1.29,
548
- "learning_rate": 0.00022957640671273443,
549
- "loss": 0.2186,
550
  "step": 870
551
  },
552
  {
553
- "epoch": 1.3,
554
- "learning_rate": 0.00022951865745310957,
555
- "loss": 0.2686,
556
  "step": 880
557
  },
558
  {
559
- "epoch": 1.31,
560
- "learning_rate": 0.00022946668311944716,
561
- "loss": 0.2136,
562
  "step": 890
563
  },
564
  {
565
- "epoch": 1.33,
566
- "learning_rate": 0.00022940893385982233,
567
- "loss": 0.1865,
568
  "step": 900
569
  },
570
  {
571
- "epoch": 1.34,
572
- "learning_rate": 0.00022935118460019744,
573
- "loss": 0.1724,
574
  "step": 910
575
  },
576
  {
577
- "epoch": 1.36,
578
- "learning_rate": 0.00022929343534057255,
579
- "loss": 0.1752,
580
  "step": 920
581
  },
582
  {
583
- "epoch": 1.37,
584
- "learning_rate": 0.0002292356860809477,
585
- "loss": 0.2561,
586
  "step": 930
587
  },
588
  {
589
- "epoch": 1.39,
590
- "learning_rate": 0.0002291779368213228,
591
- "loss": 0.1831,
592
  "step": 940
593
  },
594
  {
595
- "epoch": 1.4,
596
- "learning_rate": 0.00022912018756169792,
597
- "loss": 0.1773,
598
  "step": 950
599
  },
600
  {
601
- "epoch": 1.42,
602
- "learning_rate": 0.00022906243830207306,
603
- "loss": 0.1994,
604
  "step": 960
605
  },
606
  {
607
- "epoch": 1.43,
608
- "learning_rate": 0.00022900468904244817,
609
- "loss": 0.1943,
610
  "step": 970
611
  },
612
  {
613
- "epoch": 1.45,
614
- "learning_rate": 0.0002289469397828233,
615
- "loss": 0.27,
616
  "step": 980
617
  },
618
  {
619
- "epoch": 1.46,
620
- "learning_rate": 0.00022888919052319842,
621
- "loss": 0.1959,
622
  "step": 990
623
  },
624
  {
625
- "epoch": 1.48,
626
- "learning_rate": 0.00022883144126357354,
627
- "loss": 0.1822,
628
  "step": 1000
629
  },
630
  {
631
- "epoch": 1.49,
632
- "learning_rate": 0.00022877369200394868,
633
- "loss": 0.2212,
634
  "step": 1010
635
  },
636
  {
637
- "epoch": 1.51,
638
- "learning_rate": 0.0002287159427443238,
639
- "loss": 0.2142,
640
  "step": 1020
641
  },
642
  {
643
- "epoch": 1.52,
644
- "learning_rate": 0.0002286581934846989,
645
- "loss": 0.2741,
646
  "step": 1030
647
  },
648
  {
649
- "epoch": 1.54,
650
- "learning_rate": 0.00022860044422507404,
651
- "loss": 0.1909,
652
  "step": 1040
653
  },
654
  {
655
- "epoch": 1.55,
656
- "learning_rate": 0.00022854269496544916,
657
- "loss": 0.1991,
658
  "step": 1050
659
  },
660
  {
661
- "epoch": 1.57,
662
- "learning_rate": 0.00022848494570582427,
663
- "loss": 0.1728,
664
  "step": 1060
665
  },
666
  {
667
- "epoch": 1.58,
668
- "learning_rate": 0.0002284271964461994,
669
- "loss": 0.2023,
670
  "step": 1070
671
  },
672
  {
673
- "epoch": 1.6,
674
- "learning_rate": 0.00022836944718657452,
675
- "loss": 0.2307,
676
  "step": 1080
677
  },
678
  {
679
- "epoch": 1.61,
680
- "learning_rate": 0.00022831169792694963,
681
- "loss": 0.1708,
682
  "step": 1090
683
  },
684
  {
685
- "epoch": 1.62,
686
- "learning_rate": 0.00022825394866732477,
687
- "loss": 0.2045,
688
  "step": 1100
689
  },
690
  {
691
- "epoch": 1.64,
692
- "learning_rate": 0.0002281961994076999,
693
- "loss": 0.1962,
694
  "step": 1110
695
  },
696
  {
697
- "epoch": 1.65,
698
- "learning_rate": 0.00022813845014807503,
699
- "loss": 0.2261,
700
  "step": 1120
701
  },
702
  {
703
- "epoch": 1.67,
704
- "learning_rate": 0.00022808070088845014,
705
- "loss": 0.2592,
706
  "step": 1130
707
  },
708
  {
709
- "epoch": 1.68,
710
- "learning_rate": 0.00022802295162882525,
711
- "loss": 0.1846,
712
  "step": 1140
713
  },
714
  {
715
- "epoch": 1.7,
716
- "learning_rate": 0.0002279652023692004,
717
- "loss": 0.1764,
718
  "step": 1150
719
  },
720
  {
721
- "epoch": 1.71,
722
- "learning_rate": 0.0002279074531095755,
723
- "loss": 0.173,
724
  "step": 1160
725
  },
726
  {
727
- "epoch": 1.73,
728
- "learning_rate": 0.00022784970384995062,
729
- "loss": 0.1976,
730
  "step": 1170
731
  },
732
  {
733
- "epoch": 1.74,
734
- "learning_rate": 0.00022779195459032576,
735
- "loss": 0.2587,
736
  "step": 1180
737
  },
738
  {
739
- "epoch": 1.76,
740
- "learning_rate": 0.00022773420533070087,
741
- "loss": 0.171,
742
  "step": 1190
743
  },
744
  {
745
- "epoch": 1.77,
746
- "learning_rate": 0.00022767645607107599,
747
- "loss": 0.1942,
748
  "step": 1200
749
  },
750
  {
751
- "epoch": 1.77,
752
- "eval_cer": 0.14407103988518807,
753
- "eval_loss": 0.09608560800552368,
754
- "eval_runtime": 165.7261,
755
- "eval_samples_per_second": 14.367,
756
- "eval_steps_per_second": 0.483,
757
- "eval_wer": 0.9865602687946241,
758
  "step": 1200
759
  },
760
  {
761
- "epoch": 1.79,
762
- "learning_rate": 0.00022761870681145113,
763
- "loss": 0.1729,
764
  "step": 1210
765
  },
766
  {
767
- "epoch": 1.8,
768
- "learning_rate": 0.00022756095755182627,
769
- "loss": 0.1849,
770
  "step": 1220
771
  },
772
  {
773
- "epoch": 1.82,
774
- "learning_rate": 0.00022750320829220138,
775
- "loss": 0.2318,
776
  "step": 1230
777
  },
778
  {
779
- "epoch": 1.83,
780
- "learning_rate": 0.00022744545903257652,
781
- "loss": 0.1915,
782
  "step": 1240
783
  },
784
  {
785
- "epoch": 1.85,
786
- "learning_rate": 0.00022738770977295163,
787
- "loss": 0.1969,
788
  "step": 1250
789
  },
790
  {
791
- "epoch": 1.86,
792
- "learning_rate": 0.00022732996051332674,
793
- "loss": 0.1764,
794
  "step": 1260
795
  },
796
  {
797
- "epoch": 1.88,
798
- "learning_rate": 0.00022727221125370188,
799
- "loss": 0.1896,
800
  "step": 1270
801
  },
802
  {
803
- "epoch": 1.89,
804
- "learning_rate": 0.000227214461994077,
805
- "loss": 0.2015,
806
  "step": 1280
807
  },
808
  {
809
- "epoch": 1.91,
810
- "learning_rate": 0.00022715671273445214,
811
- "loss": 0.1796,
812
  "step": 1290
813
  },
814
  {
815
- "epoch": 1.92,
816
- "learning_rate": 0.00022709896347482725,
817
- "loss": 0.205,
818
  "step": 1300
819
  },
820
  {
821
- "epoch": 1.94,
822
- "learning_rate": 0.00022704121421520236,
823
- "loss": 0.1773,
824
  "step": 1310
825
  },
826
  {
827
- "epoch": 1.95,
828
- "learning_rate": 0.0002269834649555775,
829
- "loss": 0.1917,
830
  "step": 1320
831
  },
832
  {
833
- "epoch": 1.96,
834
- "learning_rate": 0.00022692571569595262,
835
- "loss": 0.233,
836
  "step": 1330
837
  },
838
  {
839
- "epoch": 1.98,
840
- "learning_rate": 0.00022686796643632773,
841
- "loss": 0.1977,
842
  "step": 1340
843
  },
844
  {
845
- "epoch": 1.99,
846
- "learning_rate": 0.00022681021717670287,
847
- "loss": 0.176,
848
  "step": 1350
849
  },
850
  {
851
- "epoch": 2.01,
852
- "learning_rate": 0.00022675246791707798,
853
- "loss": 0.2263,
854
  "step": 1360
855
  },
856
  {
857
- "epoch": 2.02,
858
- "learning_rate": 0.0002266947186574531,
859
- "loss": 0.1504,
860
  "step": 1370
861
  },
862
  {
863
- "epoch": 2.04,
864
- "learning_rate": 0.00022663696939782823,
865
- "loss": 0.1572,
866
  "step": 1380
867
  },
868
  {
869
- "epoch": 2.05,
870
- "learning_rate": 0.00022657922013820335,
871
- "loss": 0.1765,
872
  "step": 1390
873
  },
874
  {
875
- "epoch": 2.07,
876
- "learning_rate": 0.00022652147087857846,
877
- "loss": 0.1634,
878
  "step": 1400
879
  },
880
  {
881
- "epoch": 2.08,
882
- "learning_rate": 0.0002264637216189536,
883
- "loss": 0.2366,
884
  "step": 1410
885
  },
886
  {
887
- "epoch": 2.1,
888
- "learning_rate": 0.00022640597235932871,
889
- "loss": 0.1596,
890
  "step": 1420
891
  },
892
  {
893
- "epoch": 2.11,
894
- "learning_rate": 0.00022634822309970385,
895
- "loss": 0.1419,
896
  "step": 1430
897
  },
898
  {
899
- "epoch": 2.13,
900
- "learning_rate": 0.00022629047384007897,
901
- "loss": 0.1785,
902
  "step": 1440
903
  },
904
  {
905
- "epoch": 2.14,
906
- "learning_rate": 0.00022623272458045408,
907
- "loss": 0.1351,
908
  "step": 1450
909
  },
910
  {
911
- "epoch": 2.16,
912
- "learning_rate": 0.00022617497532082922,
913
- "loss": 0.2157,
914
  "step": 1460
915
  },
916
  {
917
- "epoch": 2.17,
918
- "learning_rate": 0.00022611722606120433,
919
- "loss": 0.1547,
920
  "step": 1470
921
  },
922
  {
923
- "epoch": 2.19,
924
- "learning_rate": 0.00022605947680157945,
925
- "loss": 0.1938,
926
  "step": 1480
927
  },
928
  {
929
- "epoch": 2.2,
930
- "learning_rate": 0.00022600172754195459,
931
- "loss": 0.1561,
932
  "step": 1490
933
  },
934
  {
935
- "epoch": 2.22,
936
- "learning_rate": 0.0002259439782823297,
937
- "loss": 0.2033,
938
  "step": 1500
939
  },
940
  {
941
- "epoch": 2.23,
942
- "learning_rate": 0.0002258862290227048,
943
- "loss": 0.2139,
944
  "step": 1510
945
  },
946
  {
947
- "epoch": 2.25,
948
- "learning_rate": 0.00022582847976307995,
949
- "loss": 0.1894,
950
  "step": 1520
951
  },
952
  {
953
- "epoch": 2.26,
954
- "learning_rate": 0.0002257707305034551,
955
- "loss": 0.1765,
956
  "step": 1530
957
  },
958
  {
959
- "epoch": 2.27,
960
- "learning_rate": 0.0002257129812438302,
961
- "loss": 0.1716,
962
  "step": 1540
963
  },
964
  {
965
- "epoch": 2.29,
966
- "learning_rate": 0.00022565523198420534,
967
- "loss": 0.1856,
968
  "step": 1550
969
  },
970
  {
971
- "epoch": 2.3,
972
- "learning_rate": 0.00022559748272458046,
973
- "loss": 0.2059,
974
  "step": 1560
975
  },
976
  {
977
- "epoch": 2.32,
978
- "learning_rate": 0.00022553973346495557,
979
- "loss": 0.1688,
980
  "step": 1570
981
  },
982
  {
983
- "epoch": 2.33,
984
- "learning_rate": 0.0002254819842053307,
985
- "loss": 0.1618,
986
  "step": 1580
987
  },
988
  {
989
- "epoch": 2.35,
990
- "learning_rate": 0.00022542423494570582,
991
- "loss": 0.1745,
992
  "step": 1590
993
  },
994
  {
995
- "epoch": 2.36,
996
- "learning_rate": 0.00022536648568608096,
997
- "loss": 0.1825,
998
  "step": 1600
999
  },
1000
  {
1001
- "epoch": 2.36,
1002
- "eval_cer": 0.1480535789033068,
1003
- "eval_loss": 0.1054612547159195,
1004
- "eval_runtime": 165.9306,
1005
- "eval_samples_per_second": 14.349,
1006
- "eval_steps_per_second": 0.482,
1007
- "eval_wer": 0.9861402771944561,
1008
  "step": 1600
1009
  },
1010
  {
1011
- "epoch": 2.38,
1012
- "learning_rate": 0.00022530873642645608,
1013
- "loss": 0.2008,
1014
  "step": 1610
1015
  },
1016
  {
1017
- "epoch": 2.39,
1018
- "learning_rate": 0.0002252509871668312,
1019
- "loss": 0.1951,
1020
  "step": 1620
1021
  },
1022
  {
1023
- "epoch": 2.41,
1024
- "learning_rate": 0.00022519323790720633,
1025
- "loss": 0.1537,
1026
  "step": 1630
1027
  },
1028
  {
1029
- "epoch": 2.42,
1030
- "learning_rate": 0.00022513548864758144,
1031
- "loss": 0.171,
1032
  "step": 1640
1033
  },
1034
  {
1035
- "epoch": 2.44,
1036
- "learning_rate": 0.00022507773938795655,
1037
- "loss": 0.1947,
1038
  "step": 1650
1039
  },
1040
  {
1041
- "epoch": 2.45,
1042
- "learning_rate": 0.0002250199901283317,
1043
- "loss": 0.1987,
1044
  "step": 1660
1045
  },
1046
  {
1047
- "epoch": 2.47,
1048
- "learning_rate": 0.0002249622408687068,
1049
- "loss": 0.1625,
1050
  "step": 1670
1051
  },
1052
  {
1053
- "epoch": 2.48,
1054
- "learning_rate": 0.00022490449160908192,
1055
- "loss": 0.1478,
1056
  "step": 1680
1057
  },
1058
  {
1059
- "epoch": 2.5,
1060
- "learning_rate": 0.00022484674234945706,
1061
- "loss": 0.1909,
1062
  "step": 1690
1063
  },
1064
  {
1065
- "epoch": 2.51,
1066
- "learning_rate": 0.00022478899308983217,
1067
- "loss": 0.1789,
1068
  "step": 1700
1069
  },
1070
  {
1071
- "epoch": 2.53,
1072
- "learning_rate": 0.0002247312438302073,
1073
- "loss": 0.1984,
1074
  "step": 1710
1075
  },
1076
  {
1077
- "epoch": 2.54,
1078
- "learning_rate": 0.00022467349457058243,
1079
- "loss": 0.1525,
1080
  "step": 1720
1081
  },
1082
  {
1083
- "epoch": 2.56,
1084
- "learning_rate": 0.00022461574531095754,
1085
- "loss": 0.1623,
1086
  "step": 1730
1087
  },
1088
  {
1089
- "epoch": 2.57,
1090
- "learning_rate": 0.00022455799605133268,
1091
- "loss": 0.1666,
1092
  "step": 1740
1093
  },
1094
  {
1095
- "epoch": 2.58,
1096
- "learning_rate": 0.0002245002467917078,
1097
- "loss": 0.1731,
1098
  "step": 1750
1099
  },
1100
  {
1101
- "epoch": 2.6,
1102
- "learning_rate": 0.0002244424975320829,
1103
- "loss": 0.211,
1104
  "step": 1760
1105
  },
1106
  {
1107
- "epoch": 2.61,
1108
- "learning_rate": 0.00022438474827245805,
1109
- "loss": 0.184,
1110
  "step": 1770
1111
  },
1112
  {
1113
- "epoch": 2.63,
1114
- "learning_rate": 0.00022432699901283316,
1115
- "loss": 0.1317,
1116
  "step": 1780
1117
  },
1118
  {
1119
- "epoch": 2.64,
1120
- "learning_rate": 0.00022426924975320827,
1121
- "loss": 0.1603,
1122
  "step": 1790
1123
  },
1124
  {
1125
- "epoch": 2.66,
1126
- "learning_rate": 0.0002242115004935834,
1127
- "loss": 0.186,
1128
  "step": 1800
1129
  },
1130
  {
1131
- "epoch": 2.67,
1132
- "learning_rate": 0.00022415375123395852,
1133
- "loss": 0.2378,
1134
  "step": 1810
1135
  },
1136
  {
1137
- "epoch": 2.69,
1138
- "learning_rate": 0.00022409600197433364,
1139
- "loss": 0.1734,
1140
  "step": 1820
1141
  },
1142
  {
1143
- "epoch": 2.7,
1144
- "learning_rate": 0.00022403825271470878,
1145
- "loss": 0.1453,
1146
  "step": 1830
1147
  },
1148
  {
1149
- "epoch": 2.72,
1150
- "learning_rate": 0.0002239805034550839,
1151
- "loss": 0.171,
1152
  "step": 1840
1153
  },
1154
  {
1155
- "epoch": 2.73,
1156
- "learning_rate": 0.00022392275419545903,
1157
- "loss": 0.2007,
1158
  "step": 1850
1159
  },
1160
  {
1161
- "epoch": 2.75,
1162
- "learning_rate": 0.00022386500493583417,
1163
- "loss": 0.2472,
1164
  "step": 1860
1165
  },
1166
  {
1167
- "epoch": 2.76,
1168
- "learning_rate": 0.00022380725567620928,
1169
- "loss": 0.1497,
1170
  "step": 1870
1171
  },
1172
  {
1173
- "epoch": 2.78,
1174
- "learning_rate": 0.0002237495064165844,
1175
- "loss": 0.1762,
1176
  "step": 1880
1177
  },
1178
  {
1179
- "epoch": 2.79,
1180
- "learning_rate": 0.00022369175715695954,
1181
- "loss": 0.142,
1182
  "step": 1890
1183
  },
1184
  {
1185
- "epoch": 2.81,
1186
- "learning_rate": 0.00022363400789733465,
1187
- "loss": 0.176,
1188
  "step": 1900
1189
  },
1190
  {
1191
- "epoch": 2.82,
1192
- "learning_rate": 0.0002235762586377098,
1193
- "loss": 0.2183,
1194
  "step": 1910
1195
  },
1196
  {
1197
- "epoch": 2.84,
1198
- "learning_rate": 0.0002235185093780849,
1199
- "loss": 0.1707,
1200
  "step": 1920
1201
  },
1202
  {
1203
- "epoch": 2.85,
1204
- "learning_rate": 0.00022346076011846001,
1205
- "loss": 0.1642,
1206
  "step": 1930
1207
  },
1208
  {
1209
- "epoch": 2.87,
1210
- "learning_rate": 0.00022340301085883515,
1211
- "loss": 0.1454,
1212
  "step": 1940
1213
  },
1214
  {
1215
- "epoch": 2.88,
1216
- "learning_rate": 0.00022334526159921027,
1217
- "loss": 0.1854,
1218
  "step": 1950
1219
  },
1220
  {
1221
- "epoch": 2.9,
1222
- "learning_rate": 0.00022328751233958538,
1223
- "loss": 0.2147,
1224
  "step": 1960
1225
  },
1226
  {
1227
- "epoch": 2.91,
1228
- "learning_rate": 0.00022322976307996052,
1229
- "loss": 0.1664,
1230
  "step": 1970
1231
  },
1232
  {
1233
- "epoch": 2.92,
1234
- "learning_rate": 0.00022317201382033563,
1235
- "loss": 0.1495,
1236
  "step": 1980
1237
  },
1238
  {
1239
- "epoch": 2.94,
1240
- "learning_rate": 0.00022311426456071075,
1241
- "loss": 0.1694,
1242
  "step": 1990
1243
  },
1244
  {
1245
- "epoch": 2.95,
1246
- "learning_rate": 0.0002230565153010859,
1247
- "loss": 0.1664,
1248
  "step": 2000
1249
  },
1250
  {
1251
- "epoch": 2.95,
1252
- "eval_cer": 0.15189858279016924,
1253
- "eval_loss": 0.09597988426685333,
1254
- "eval_runtime": 165.3281,
1255
- "eval_samples_per_second": 14.402,
1256
- "eval_steps_per_second": 0.484,
1257
- "eval_wer": 0.9882402351952961,
1258
  "step": 2000
1259
  },
1260
  {
1261
- "epoch": 2.97,
1262
- "learning_rate": 0.000222998766041461,
1263
- "loss": 0.2587,
1264
  "step": 2010
1265
  },
1266
  {
1267
- "epoch": 2.98,
1268
- "learning_rate": 0.0002229410167818361,
1269
- "loss": 0.1783,
1270
  "step": 2020
1271
  },
1272
  {
1273
- "epoch": 3.0,
1274
- "learning_rate": 0.00022288326752221125,
1275
- "loss": 0.1742,
1276
  "step": 2030
1277
  },
1278
  {
1279
- "epoch": 3.01,
1280
- "learning_rate": 0.00022282551826258637,
1281
- "loss": 0.1699,
1282
  "step": 2040
1283
  },
1284
  {
1285
- "epoch": 3.03,
1286
- "learning_rate": 0.0002227677690029615,
1287
- "loss": 0.1536,
1288
  "step": 2050
1289
  },
1290
  {
1291
- "epoch": 3.04,
1292
- "learning_rate": 0.00022271001974333662,
1293
- "loss": 0.155,
1294
  "step": 2060
1295
  },
1296
  {
1297
- "epoch": 3.06,
1298
- "learning_rate": 0.00022265227048371173,
1299
- "loss": 0.1595,
1300
  "step": 2070
1301
  },
1302
  {
1303
- "epoch": 3.07,
1304
- "learning_rate": 0.00022259452122408687,
1305
- "loss": 0.1566,
1306
  "step": 2080
1307
  },
1308
  {
1309
- "epoch": 3.09,
1310
- "learning_rate": 0.00022253677196446198,
1311
- "loss": 0.2007,
1312
  "step": 2090
1313
  },
1314
  {
1315
- "epoch": 3.1,
1316
- "learning_rate": 0.0002224790227048371,
1317
- "loss": 0.1453,
1318
  "step": 2100
1319
  },
1320
  {
1321
- "epoch": 3.12,
1322
- "learning_rate": 0.00022242127344521224,
1323
- "loss": 0.1786,
1324
  "step": 2110
1325
  },
1326
  {
1327
- "epoch": 3.13,
1328
- "learning_rate": 0.00022236352418558735,
1329
- "loss": 0.1646,
1330
  "step": 2120
1331
  },
1332
  {
1333
- "epoch": 3.15,
1334
- "learning_rate": 0.00022230577492596246,
1335
- "loss": 0.2085,
1336
  "step": 2130
1337
  },
1338
  {
1339
- "epoch": 3.16,
1340
- "learning_rate": 0.0002222480256663376,
1341
- "loss": 0.1987,
1342
  "step": 2140
1343
  },
1344
  {
1345
- "epoch": 3.18,
1346
- "learning_rate": 0.00022219027640671272,
1347
- "loss": 0.1479,
1348
  "step": 2150
1349
  },
1350
  {
1351
- "epoch": 3.19,
1352
- "learning_rate": 0.00022213252714708783,
1353
- "loss": 0.1693,
1354
  "step": 2160
1355
  },
1356
  {
1357
- "epoch": 3.21,
1358
- "learning_rate": 0.000222074777887463,
1359
- "loss": 0.1579,
1360
  "step": 2170
1361
  },
1362
  {
1363
- "epoch": 3.22,
1364
- "learning_rate": 0.0002220170286278381,
1365
- "loss": 0.1905,
1366
  "step": 2180
1367
  },
1368
  {
1369
- "epoch": 3.23,
1370
- "learning_rate": 0.00022195927936821322,
1371
- "loss": 0.1969,
1372
  "step": 2190
1373
  },
1374
  {
1375
- "epoch": 3.25,
1376
- "learning_rate": 0.00022190153010858836,
1377
- "loss": 0.1504,
1378
  "step": 2200
1379
  },
1380
  {
1381
- "epoch": 3.26,
1382
- "learning_rate": 0.00022184378084896347,
1383
- "loss": 0.1784,
1384
  "step": 2210
1385
  },
1386
  {
1387
- "epoch": 3.28,
1388
- "learning_rate": 0.00022178603158933861,
1389
- "loss": 0.1693,
1390
  "step": 2220
1391
  },
1392
  {
1393
- "epoch": 3.29,
1394
- "learning_rate": 0.00022172828232971373,
1395
- "loss": 0.1799,
1396
  "step": 2230
1397
  },
1398
  {
1399
- "epoch": 3.31,
1400
- "learning_rate": 0.00022167053307008884,
1401
- "loss": 0.19,
1402
  "step": 2240
1403
  },
1404
  {
1405
- "epoch": 3.32,
1406
- "learning_rate": 0.00022161278381046398,
1407
- "loss": 0.1471,
1408
  "step": 2250
1409
  },
1410
  {
1411
- "epoch": 3.34,
1412
- "learning_rate": 0.0002215550345508391,
1413
- "loss": 0.154,
1414
  "step": 2260
1415
  },
1416
  {
1417
- "epoch": 3.35,
1418
- "learning_rate": 0.0002214972852912142,
1419
- "loss": 0.1579,
1420
  "step": 2270
1421
  },
1422
  {
1423
- "epoch": 3.37,
1424
- "learning_rate": 0.00022143953603158935,
1425
- "loss": 0.2001,
1426
  "step": 2280
1427
  },
1428
  {
1429
- "epoch": 3.38,
1430
- "learning_rate": 0.00022138178677196446,
1431
- "loss": 0.186,
1432
  "step": 2290
1433
  },
1434
  {
1435
- "epoch": 3.4,
1436
- "learning_rate": 0.00022132403751233957,
1437
- "loss": 0.1495,
1438
  "step": 2300
1439
  },
1440
  {
1441
- "epoch": 3.41,
1442
- "learning_rate": 0.0002212662882527147,
1443
- "loss": 0.1514,
1444
  "step": 2310
1445
  },
1446
  {
1447
- "epoch": 3.43,
1448
- "learning_rate": 0.00022120853899308983,
1449
- "loss": 0.1803,
1450
  "step": 2320
1451
  },
1452
  {
1453
- "epoch": 3.44,
1454
- "learning_rate": 0.00022115078973346494,
1455
- "loss": 0.1838,
1456
  "step": 2330
1457
  },
1458
  {
1459
- "epoch": 3.46,
1460
- "learning_rate": 0.00022109304047384008,
1461
- "loss": 0.2408,
1462
  "step": 2340
1463
  },
1464
  {
1465
- "epoch": 3.47,
1466
- "learning_rate": 0.0002210352912142152,
1467
- "loss": 0.137,
1468
  "step": 2350
1469
  },
1470
  {
1471
- "epoch": 3.49,
1472
- "learning_rate": 0.00022097754195459033,
1473
- "loss": 0.1435,
1474
  "step": 2360
1475
  },
1476
  {
1477
- "epoch": 3.5,
1478
- "learning_rate": 0.00022091979269496544,
1479
- "loss": 0.1597,
1480
  "step": 2370
1481
  },
1482
  {
1483
- "epoch": 3.52,
1484
- "learning_rate": 0.00022086204343534056,
1485
- "loss": 0.2343,
1486
  "step": 2380
1487
  },
1488
  {
1489
- "epoch": 3.53,
1490
- "learning_rate": 0.0002208042941757157,
1491
- "loss": 0.1752,
1492
  "step": 2390
1493
  },
1494
  {
1495
- "epoch": 3.55,
1496
- "learning_rate": 0.0002207465449160908,
1497
- "loss": 0.145,
1498
  "step": 2400
1499
  },
1500
  {
1501
- "epoch": 3.55,
1502
- "eval_cer": 0.16321832207139866,
1503
- "eval_loss": 0.10200775414705276,
1504
- "eval_runtime": 165.6528,
1505
- "eval_samples_per_second": 14.373,
1506
- "eval_steps_per_second": 0.483,
1507
- "eval_wer": 0.989920201595968,
1508
  "step": 2400
1509
- },
1510
- {
1511
- "epoch": 3.56,
1512
- "learning_rate": 0.00022068879565646592,
1513
- "loss": 0.1755,
1514
- "step": 2410
1515
- },
1516
- {
1517
- "epoch": 3.57,
1518
- "learning_rate": 0.00022063104639684106,
1519
- "loss": 0.1808,
1520
- "step": 2420
1521
- },
1522
- {
1523
- "epoch": 3.59,
1524
- "learning_rate": 0.00022057329713721618,
1525
- "loss": 0.1764,
1526
- "step": 2430
1527
- },
1528
- {
1529
- "epoch": 3.6,
1530
- "learning_rate": 0.0002205155478775913,
1531
- "loss": 0.1693,
1532
- "step": 2440
1533
- },
1534
- {
1535
- "epoch": 3.62,
1536
- "learning_rate": 0.00022045779861796643,
1537
- "loss": 0.1472,
1538
- "step": 2450
1539
- },
1540
- {
1541
- "epoch": 3.63,
1542
- "learning_rate": 0.00022040004935834154,
1543
- "loss": 0.1485,
1544
- "step": 2460
1545
- },
1546
- {
1547
- "epoch": 3.65,
1548
- "learning_rate": 0.00022034230009871665,
1549
- "loss": 0.1508,
1550
- "step": 2470
1551
- },
1552
- {
1553
- "epoch": 3.66,
1554
- "learning_rate": 0.00022028455083909182,
1555
- "loss": 0.1928,
1556
- "step": 2480
1557
- },
1558
- {
1559
- "epoch": 3.68,
1560
- "learning_rate": 0.00022022680157946693,
1561
- "loss": 0.1844,
1562
- "step": 2490
1563
- },
1564
- {
1565
- "epoch": 3.69,
1566
- "learning_rate": 0.00022016905231984205,
1567
- "loss": 0.1618,
1568
- "step": 2500
1569
- },
1570
- {
1571
- "epoch": 3.71,
1572
- "learning_rate": 0.0002201113030602172,
1573
- "loss": 0.1525,
1574
- "step": 2510
1575
- },
1576
- {
1577
- "epoch": 3.72,
1578
- "learning_rate": 0.0002200535538005923,
1579
- "loss": 0.1799,
1580
- "step": 2520
1581
- },
1582
- {
1583
- "epoch": 3.74,
1584
- "learning_rate": 0.00021999580454096744,
1585
- "loss": 0.203,
1586
- "step": 2530
1587
- },
1588
- {
1589
- "epoch": 3.75,
1590
- "learning_rate": 0.00021993805528134255,
1591
- "loss": 0.1789,
1592
- "step": 2540
1593
- },
1594
- {
1595
- "epoch": 3.77,
1596
- "learning_rate": 0.00021988030602171767,
1597
- "loss": 0.1593,
1598
- "step": 2550
1599
- },
1600
- {
1601
- "epoch": 3.78,
1602
- "learning_rate": 0.0002198225567620928,
1603
- "loss": 0.1344,
1604
- "step": 2560
1605
- },
1606
- {
1607
- "epoch": 3.8,
1608
- "learning_rate": 0.00021976480750246792,
1609
- "loss": 0.1924,
1610
- "step": 2570
1611
- },
1612
- {
1613
- "epoch": 3.81,
1614
- "learning_rate": 0.00021970705824284303,
1615
- "loss": 0.2305,
1616
- "step": 2580
1617
- },
1618
- {
1619
- "epoch": 3.83,
1620
- "learning_rate": 0.00021964930898321817,
1621
- "loss": 0.1798,
1622
- "step": 2590
1623
- },
1624
- {
1625
- "epoch": 3.84,
1626
- "learning_rate": 0.00021959155972359329,
1627
- "loss": 0.1414,
1628
- "step": 2600
1629
- },
1630
- {
1631
- "epoch": 3.86,
1632
- "learning_rate": 0.0002195338104639684,
1633
- "loss": 0.1392,
1634
- "step": 2610
1635
- },
1636
- {
1637
- "epoch": 3.87,
1638
- "learning_rate": 0.00021947606120434354,
1639
- "loss": 0.1533,
1640
- "step": 2620
1641
- },
1642
- {
1643
- "epoch": 3.88,
1644
- "learning_rate": 0.00021941831194471865,
1645
- "loss": 0.1981,
1646
- "step": 2630
1647
- },
1648
- {
1649
- "epoch": 3.9,
1650
- "learning_rate": 0.00021936056268509376,
1651
- "loss": 0.179,
1652
- "step": 2640
1653
- },
1654
- {
1655
- "epoch": 3.91,
1656
- "learning_rate": 0.0002193028134254689,
1657
- "loss": 0.1409,
1658
- "step": 2650
1659
- },
1660
- {
1661
- "epoch": 3.93,
1662
- "learning_rate": 0.00021924506416584402,
1663
- "loss": 0.1624,
1664
- "step": 2660
1665
- },
1666
- {
1667
- "epoch": 3.94,
1668
- "learning_rate": 0.00021918731490621916,
1669
- "loss": 0.1619,
1670
- "step": 2670
1671
- },
1672
- {
1673
- "epoch": 3.96,
1674
- "learning_rate": 0.00021912956564659427,
1675
- "loss": 0.1727,
1676
- "step": 2680
1677
- },
1678
- {
1679
- "epoch": 3.97,
1680
- "learning_rate": 0.00021907181638696938,
1681
- "loss": 0.1602,
1682
- "step": 2690
1683
- },
1684
- {
1685
- "epoch": 3.99,
1686
- "learning_rate": 0.00021901406712734452,
1687
- "loss": 0.1295,
1688
- "step": 2700
1689
- },
1690
- {
1691
- "epoch": 4.0,
1692
- "learning_rate": 0.00021895631786771964,
1693
- "loss": 0.1825,
1694
- "step": 2710
1695
- },
1696
- {
1697
- "epoch": 4.02,
1698
- "learning_rate": 0.00021889856860809475,
1699
- "loss": 0.1724,
1700
- "step": 2720
1701
- },
1702
- {
1703
- "epoch": 4.03,
1704
- "learning_rate": 0.0002188408193484699,
1705
- "loss": 0.1503,
1706
- "step": 2730
1707
- },
1708
- {
1709
- "epoch": 4.05,
1710
- "learning_rate": 0.000218783070088845,
1711
- "loss": 0.1364,
1712
- "step": 2740
1713
- },
1714
- {
1715
- "epoch": 4.06,
1716
- "learning_rate": 0.00021872532082922011,
1717
- "loss": 0.144,
1718
- "step": 2750
1719
- },
1720
- {
1721
- "epoch": 4.08,
1722
- "learning_rate": 0.00021866757156959525,
1723
- "loss": 0.178,
1724
- "step": 2760
1725
- },
1726
- {
1727
- "epoch": 4.09,
1728
- "learning_rate": 0.00021860982230997037,
1729
- "loss": 0.1625,
1730
- "step": 2770
1731
- },
1732
- {
1733
- "epoch": 4.11,
1734
- "learning_rate": 0.00021855207305034548,
1735
- "loss": 0.1344,
1736
- "step": 2780
1737
- },
1738
- {
1739
- "epoch": 4.12,
1740
- "learning_rate": 0.00021849432379072062,
1741
- "loss": 0.1355,
1742
- "step": 2790
1743
- },
1744
- {
1745
- "epoch": 4.14,
1746
- "learning_rate": 0.00021843657453109576,
1747
- "loss": 0.1534,
1748
- "step": 2800
1749
- },
1750
- {
1751
- "epoch": 4.14,
1752
- "eval_cer": 0.15161155295102555,
1753
- "eval_loss": 0.10632243752479553,
1754
- "eval_runtime": 166.0979,
1755
- "eval_samples_per_second": 14.335,
1756
- "eval_steps_per_second": 0.482,
1757
- "eval_wer": 0.9764804703905922,
1758
- "step": 2800
1759
- },
1760
- {
1761
- "epoch": 4.15,
1762
- "learning_rate": 0.00021837882527147087,
1763
- "loss": 0.1826,
1764
- "step": 2810
1765
- },
1766
- {
1767
- "epoch": 4.17,
1768
- "learning_rate": 0.000218321076011846,
1769
- "loss": 0.1615,
1770
- "step": 2820
1771
- },
1772
- {
1773
- "epoch": 4.18,
1774
- "learning_rate": 0.00021826332675222113,
1775
- "loss": 0.1218,
1776
- "step": 2830
1777
- },
1778
- {
1779
- "epoch": 4.19,
1780
- "learning_rate": 0.00021820557749259627,
1781
- "loss": 0.1326,
1782
- "step": 2840
1783
- },
1784
- {
1785
- "epoch": 4.21,
1786
- "learning_rate": 0.00021814782823297138,
1787
- "loss": 0.1623,
1788
- "step": 2850
1789
- },
1790
- {
1791
- "epoch": 4.22,
1792
- "learning_rate": 0.0002180900789733465,
1793
- "loss": 0.1729,
1794
- "step": 2860
1795
- },
1796
- {
1797
- "epoch": 4.24,
1798
- "learning_rate": 0.00021803232971372163,
1799
- "loss": 0.1435,
1800
- "step": 2870
1801
- },
1802
- {
1803
- "epoch": 4.25,
1804
- "learning_rate": 0.00021797458045409675,
1805
- "loss": 0.1547,
1806
- "step": 2880
1807
- },
1808
- {
1809
- "epoch": 4.27,
1810
- "learning_rate": 0.00021791683119447186,
1811
- "loss": 0.156,
1812
- "step": 2890
1813
- },
1814
- {
1815
- "epoch": 4.28,
1816
- "learning_rate": 0.000217859081934847,
1817
- "loss": 0.1387,
1818
- "step": 2900
1819
- },
1820
- {
1821
- "epoch": 4.3,
1822
- "learning_rate": 0.0002178013326752221,
1823
- "loss": 0.2081,
1824
- "step": 2910
1825
- },
1826
- {
1827
- "epoch": 4.31,
1828
- "learning_rate": 0.00021774358341559722,
1829
- "loss": 0.1335,
1830
- "step": 2920
1831
- },
1832
- {
1833
- "epoch": 4.33,
1834
- "learning_rate": 0.00021768583415597236,
1835
- "loss": 0.1493,
1836
- "step": 2930
1837
- },
1838
- {
1839
- "epoch": 4.34,
1840
- "learning_rate": 0.00021762808489634748,
1841
- "loss": 0.1436,
1842
- "step": 2940
1843
- },
1844
- {
1845
- "epoch": 4.36,
1846
- "learning_rate": 0.0002175703356367226,
1847
- "loss": 0.1497,
1848
- "step": 2950
1849
- },
1850
- {
1851
- "epoch": 4.37,
1852
- "learning_rate": 0.00021751258637709773,
1853
- "loss": 0.1468,
1854
- "step": 2960
1855
- },
1856
- {
1857
- "epoch": 4.39,
1858
- "learning_rate": 0.00021745483711747284,
1859
- "loss": 0.1292,
1860
- "step": 2970
1861
- },
1862
- {
1863
- "epoch": 4.4,
1864
- "learning_rate": 0.00021739708785784798,
1865
- "loss": 0.1211,
1866
- "step": 2980
1867
- },
1868
- {
1869
- "epoch": 4.42,
1870
- "learning_rate": 0.0002173393385982231,
1871
- "loss": 0.1397,
1872
- "step": 2990
1873
- },
1874
- {
1875
- "epoch": 4.43,
1876
- "learning_rate": 0.0002172815893385982,
1877
- "loss": 0.1508,
1878
- "step": 3000
1879
- },
1880
- {
1881
- "epoch": 4.45,
1882
- "learning_rate": 0.00021722384007897335,
1883
- "loss": 0.2053,
1884
- "step": 3010
1885
- },
1886
- {
1887
- "epoch": 4.46,
1888
- "learning_rate": 0.00021716609081934846,
1889
- "loss": 0.1531,
1890
- "step": 3020
1891
- },
1892
- {
1893
- "epoch": 4.48,
1894
- "learning_rate": 0.00021710834155972357,
1895
- "loss": 0.132,
1896
- "step": 3030
1897
- },
1898
- {
1899
- "epoch": 4.49,
1900
- "learning_rate": 0.00021705059230009871,
1901
- "loss": 0.1469,
1902
- "step": 3040
1903
- },
1904
- {
1905
- "epoch": 4.51,
1906
- "learning_rate": 0.00021699284304047383,
1907
- "loss": 0.1471,
1908
- "step": 3050
1909
- },
1910
- {
1911
- "epoch": 4.52,
1912
- "learning_rate": 0.00021693509378084894,
1913
- "loss": 0.161,
1914
- "step": 3060
1915
- },
1916
- {
1917
- "epoch": 4.53,
1918
- "learning_rate": 0.00021687734452122408,
1919
- "loss": 0.1464,
1920
- "step": 3070
1921
- },
1922
- {
1923
- "epoch": 4.55,
1924
- "learning_rate": 0.0002168195952615992,
1925
- "loss": 0.1569,
1926
- "step": 3080
1927
- },
1928
- {
1929
- "epoch": 4.56,
1930
- "learning_rate": 0.0002167618460019743,
1931
- "loss": 0.1392,
1932
- "step": 3090
1933
- },
1934
- {
1935
- "epoch": 4.58,
1936
- "learning_rate": 0.00021670409674234945,
1937
- "loss": 0.1533,
1938
- "step": 3100
1939
- },
1940
- {
1941
- "epoch": 4.59,
1942
- "learning_rate": 0.00021664634748272456,
1943
- "loss": 0.1944,
1944
- "step": 3110
1945
- },
1946
- {
1947
- "epoch": 4.61,
1948
- "learning_rate": 0.0002165885982230997,
1949
- "loss": 0.1442,
1950
- "step": 3120
1951
- },
1952
- {
1953
- "epoch": 4.62,
1954
- "learning_rate": 0.00021653084896347484,
1955
- "loss": 0.1265,
1956
- "step": 3130
1957
- },
1958
- {
1959
- "epoch": 4.64,
1960
- "learning_rate": 0.00021647309970384995,
1961
- "loss": 0.1276,
1962
- "step": 3140
1963
- },
1964
- {
1965
- "epoch": 4.65,
1966
- "learning_rate": 0.0002164153504442251,
1967
- "loss": 0.1633,
1968
- "step": 3150
1969
- },
1970
- {
1971
- "epoch": 4.67,
1972
- "learning_rate": 0.0002163576011846002,
1973
- "loss": 0.1694,
1974
- "step": 3160
1975
- },
1976
- {
1977
- "epoch": 4.68,
1978
- "learning_rate": 0.00021629985192497532,
1979
- "loss": 0.1232,
1980
- "step": 3170
1981
- },
1982
- {
1983
- "epoch": 4.7,
1984
- "learning_rate": 0.00021624210266535046,
1985
- "loss": 0.1466,
1986
- "step": 3180
1987
- },
1988
- {
1989
- "epoch": 4.71,
1990
- "learning_rate": 0.00021618435340572557,
1991
- "loss": 0.1246,
1992
- "step": 3190
1993
- },
1994
- {
1995
- "epoch": 4.73,
1996
- "learning_rate": 0.00021612660414610068,
1997
- "loss": 0.1327,
1998
- "step": 3200
1999
- },
2000
- {
2001
- "epoch": 4.73,
2002
- "eval_cer": 0.12719009747054955,
2003
- "eval_loss": 0.09979347139596939,
2004
- "eval_runtime": 166.6582,
2005
- "eval_samples_per_second": 14.287,
2006
- "eval_steps_per_second": 0.48,
2007
- "eval_wer": 0.9764804703905922,
2008
- "step": 3200
2009
- },
2010
- {
2011
- "epoch": 4.74,
2012
- "learning_rate": 0.00021606885488647582,
2013
- "loss": 0.2253,
2014
- "step": 3210
2015
- },
2016
- {
2017
- "epoch": 4.76,
2018
- "learning_rate": 0.00021601110562685094,
2019
- "loss": 0.1384,
2020
- "step": 3220
2021
- },
2022
- {
2023
- "epoch": 4.77,
2024
- "learning_rate": 0.00021595335636722605,
2025
- "loss": 0.1735,
2026
- "step": 3230
2027
- },
2028
- {
2029
- "epoch": 4.79,
2030
- "learning_rate": 0.0002158956071076012,
2031
- "loss": 0.121,
2032
- "step": 3240
2033
- },
2034
- {
2035
- "epoch": 4.8,
2036
- "learning_rate": 0.0002158378578479763,
2037
- "loss": 0.1469,
2038
- "step": 3250
2039
- },
2040
- {
2041
- "epoch": 4.82,
2042
- "learning_rate": 0.00021578010858835142,
2043
- "loss": 0.1975,
2044
- "step": 3260
2045
- },
2046
- {
2047
- "epoch": 4.83,
2048
- "learning_rate": 0.00021572235932872656,
2049
- "loss": 0.133,
2050
- "step": 3270
2051
- },
2052
- {
2053
- "epoch": 4.84,
2054
- "learning_rate": 0.00021566461006910167,
2055
- "loss": 0.1317,
2056
- "step": 3280
2057
- },
2058
- {
2059
- "epoch": 4.86,
2060
- "learning_rate": 0.0002156068608094768,
2061
- "loss": 0.1458,
2062
- "step": 3290
2063
- },
2064
- {
2065
- "epoch": 4.87,
2066
- "learning_rate": 0.00021554911154985192,
2067
- "loss": 0.1423,
2068
- "step": 3300
2069
- },
2070
- {
2071
- "epoch": 4.89,
2072
- "learning_rate": 0.00021549136229022703,
2073
- "loss": 0.1852,
2074
- "step": 3310
2075
- },
2076
- {
2077
- "epoch": 4.9,
2078
- "learning_rate": 0.00021543361303060217,
2079
- "loss": 0.1516,
2080
- "step": 3320
2081
- },
2082
- {
2083
- "epoch": 4.92,
2084
- "learning_rate": 0.0002153758637709773,
2085
- "loss": 0.1418,
2086
- "step": 3330
2087
- },
2088
- {
2089
- "epoch": 4.93,
2090
- "learning_rate": 0.0002153181145113524,
2091
- "loss": 0.1275,
2092
- "step": 3340
2093
- },
2094
- {
2095
- "epoch": 4.95,
2096
- "learning_rate": 0.00021526036525172754,
2097
- "loss": 0.1199,
2098
- "step": 3350
2099
- },
2100
- {
2101
- "epoch": 4.96,
2102
- "learning_rate": 0.00021520261599210265,
2103
- "loss": 0.135,
2104
- "step": 3360
2105
- },
2106
- {
2107
- "epoch": 4.98,
2108
- "learning_rate": 0.00021514486673247777,
2109
- "loss": 0.1277,
2110
- "step": 3370
2111
- },
2112
- {
2113
- "epoch": 4.99,
2114
- "learning_rate": 0.0002150871174728529,
2115
- "loss": 0.1338,
2116
- "step": 3380
2117
- },
2118
- {
2119
- "epoch": 5.01,
2120
- "learning_rate": 0.00021502936821322802,
2121
- "loss": 0.2228,
2122
- "step": 3390
2123
- },
2124
- {
2125
- "epoch": 5.02,
2126
- "learning_rate": 0.00021497161895360313,
2127
- "loss": 0.138,
2128
- "step": 3400
2129
- },
2130
- {
2131
- "epoch": 5.04,
2132
- "learning_rate": 0.00021491386969397827,
2133
- "loss": 0.1168,
2134
- "step": 3410
2135
- },
2136
- {
2137
- "epoch": 5.05,
2138
- "learning_rate": 0.00021485612043435339,
2139
- "loss": 0.1339,
2140
- "step": 3420
2141
- },
2142
- {
2143
- "epoch": 5.07,
2144
- "learning_rate": 0.00021479837117472853,
2145
- "loss": 0.1233,
2146
- "step": 3430
2147
- },
2148
- {
2149
- "epoch": 5.08,
2150
- "learning_rate": 0.00021474062191510367,
2151
- "loss": 0.1767,
2152
- "step": 3440
2153
- },
2154
- {
2155
- "epoch": 5.1,
2156
- "learning_rate": 0.00021468287265547878,
2157
- "loss": 0.126,
2158
- "step": 3450
2159
- },
2160
- {
2161
- "epoch": 5.11,
2162
- "learning_rate": 0.00021462512339585392,
2163
- "loss": 0.1149,
2164
- "step": 3460
2165
- },
2166
- {
2167
- "epoch": 5.13,
2168
- "learning_rate": 0.00021456737413622903,
2169
- "loss": 0.1379,
2170
- "step": 3470
2171
- },
2172
- {
2173
- "epoch": 5.14,
2174
- "learning_rate": 0.00021450962487660414,
2175
- "loss": 0.1406,
2176
- "step": 3480
2177
- },
2178
- {
2179
- "epoch": 5.16,
2180
- "learning_rate": 0.00021445187561697928,
2181
- "loss": 0.2308,
2182
- "step": 3490
2183
- },
2184
- {
2185
- "epoch": 5.17,
2186
- "learning_rate": 0.0002143941263573544,
2187
- "loss": 0.1305,
2188
- "step": 3500
2189
- },
2190
- {
2191
- "epoch": 5.18,
2192
- "learning_rate": 0.0002143363770977295,
2193
- "loss": 0.1516,
2194
- "step": 3510
2195
- },
2196
- {
2197
- "epoch": 5.2,
2198
- "learning_rate": 0.00021427862783810465,
2199
- "loss": 0.1331,
2200
- "step": 3520
2201
- },
2202
- {
2203
- "epoch": 5.21,
2204
- "learning_rate": 0.00021422087857847976,
2205
- "loss": 0.1383,
2206
- "step": 3530
2207
- },
2208
- {
2209
- "epoch": 5.23,
2210
- "learning_rate": 0.00021416312931885488,
2211
- "loss": 0.1725,
2212
- "step": 3540
2213
- },
2214
- {
2215
- "epoch": 5.24,
2216
- "learning_rate": 0.00021410538005923002,
2217
- "loss": 0.1231,
2218
- "step": 3550
2219
- },
2220
- {
2221
- "epoch": 5.26,
2222
- "learning_rate": 0.00021404763079960513,
2223
- "loss": 0.1221,
2224
- "step": 3560
2225
- },
2226
- {
2227
- "epoch": 5.27,
2228
- "learning_rate": 0.00021398988153998024,
2229
- "loss": 0.1338,
2230
- "step": 3570
2231
- },
2232
- {
2233
- "epoch": 5.29,
2234
- "learning_rate": 0.00021393213228035538,
2235
- "loss": 0.149,
2236
- "step": 3580
2237
- },
2238
- {
2239
- "epoch": 5.3,
2240
- "learning_rate": 0.0002138743830207305,
2241
- "loss": 0.1826,
2242
- "step": 3590
2243
- },
2244
- {
2245
- "epoch": 5.32,
2246
- "learning_rate": 0.00021381663376110563,
2247
- "loss": 0.1227,
2248
- "step": 3600
2249
- },
2250
- {
2251
- "epoch": 5.32,
2252
- "eval_cer": 0.1120133947258267,
2253
- "eval_loss": 0.10438049584627151,
2254
- "eval_runtime": 166.8135,
2255
- "eval_samples_per_second": 14.273,
2256
- "eval_steps_per_second": 0.48,
2257
- "eval_wer": 0.9706005879882402,
2258
- "step": 3600
2259
- },
2260
- {
2261
- "epoch": 5.33,
2262
- "learning_rate": 0.00021375888450148075,
2263
- "loss": 0.1549,
2264
- "step": 3610
2265
- },
2266
- {
2267
- "epoch": 5.35,
2268
- "learning_rate": 0.00021370113524185586,
2269
- "loss": 0.1344,
2270
- "step": 3620
2271
- },
2272
- {
2273
- "epoch": 5.36,
2274
- "learning_rate": 0.000213643385982231,
2275
- "loss": 0.1332,
2276
- "step": 3630
2277
- },
2278
- {
2279
- "epoch": 5.38,
2280
- "learning_rate": 0.0002135856367226061,
2281
- "loss": 0.1962,
2282
- "step": 3640
2283
- },
2284
- {
2285
- "epoch": 5.39,
2286
- "learning_rate": 0.00021352788746298123,
2287
- "loss": 0.1639,
2288
- "step": 3650
2289
- },
2290
- {
2291
- "epoch": 5.41,
2292
- "learning_rate": 0.00021347013820335637,
2293
- "loss": 0.1347,
2294
- "step": 3660
2295
- },
2296
- {
2297
- "epoch": 5.42,
2298
- "learning_rate": 0.00021341238894373148,
2299
- "loss": 0.1247,
2300
- "step": 3670
2301
- },
2302
- {
2303
- "epoch": 5.44,
2304
- "learning_rate": 0.0002133546396841066,
2305
- "loss": 0.1204,
2306
- "step": 3680
2307
- },
2308
- {
2309
- "epoch": 5.45,
2310
- "learning_rate": 0.00021329689042448173,
2311
- "loss": 0.1719,
2312
- "step": 3690
2313
- },
2314
- {
2315
- "epoch": 5.47,
2316
- "learning_rate": 0.00021323914116485685,
2317
- "loss": 0.1347,
2318
- "step": 3700
2319
- },
2320
- {
2321
- "epoch": 5.48,
2322
- "learning_rate": 0.00021318139190523196,
2323
- "loss": 0.1365,
2324
- "step": 3710
2325
- },
2326
- {
2327
- "epoch": 5.49,
2328
- "learning_rate": 0.0002131236426456071,
2329
- "loss": 0.1205,
2330
- "step": 3720
2331
- },
2332
- {
2333
- "epoch": 5.51,
2334
- "learning_rate": 0.0002130658933859822,
2335
- "loss": 0.115,
2336
- "step": 3730
2337
- },
2338
- {
2339
- "epoch": 5.52,
2340
- "learning_rate": 0.00021300814412635735,
2341
- "loss": 0.185,
2342
- "step": 3740
2343
- },
2344
- {
2345
- "epoch": 5.54,
2346
- "learning_rate": 0.0002129503948667325,
2347
- "loss": 0.142,
2348
- "step": 3750
2349
- },
2350
- {
2351
- "epoch": 5.55,
2352
- "learning_rate": 0.0002128926456071076,
2353
- "loss": 0.1271,
2354
- "step": 3760
2355
- },
2356
- {
2357
- "epoch": 5.57,
2358
- "learning_rate": 0.00021283489634748274,
2359
- "loss": 0.1131,
2360
- "step": 3770
2361
- },
2362
- {
2363
- "epoch": 5.58,
2364
- "learning_rate": 0.00021277714708785786,
2365
- "loss": 0.1687,
2366
- "step": 3780
2367
- },
2368
- {
2369
- "epoch": 5.6,
2370
- "learning_rate": 0.00021271939782823297,
2371
- "loss": 0.1643,
2372
- "step": 3790
2373
- },
2374
- {
2375
- "epoch": 5.61,
2376
- "learning_rate": 0.0002126616485686081,
2377
- "loss": 0.1241,
2378
- "step": 3800
2379
- },
2380
- {
2381
- "epoch": 5.63,
2382
- "learning_rate": 0.00021260389930898322,
2383
- "loss": 0.1178,
2384
- "step": 3810
2385
- },
2386
- {
2387
- "epoch": 5.64,
2388
- "learning_rate": 0.00021254615004935834,
2389
- "loss": 0.1287,
2390
- "step": 3820
2391
- },
2392
- {
2393
- "epoch": 5.66,
2394
- "learning_rate": 0.00021248840078973348,
2395
- "loss": 0.1464,
2396
- "step": 3830
2397
- },
2398
- {
2399
- "epoch": 5.67,
2400
- "learning_rate": 0.0002124306515301086,
2401
- "loss": 0.1721,
2402
- "step": 3840
2403
- },
2404
- {
2405
- "epoch": 5.69,
2406
- "learning_rate": 0.0002123729022704837,
2407
- "loss": 0.1249,
2408
- "step": 3850
2409
- },
2410
- {
2411
- "epoch": 5.7,
2412
- "learning_rate": 0.00021231515301085884,
2413
- "loss": 0.1276,
2414
- "step": 3860
2415
- },
2416
- {
2417
- "epoch": 5.72,
2418
- "learning_rate": 0.00021225740375123395,
2419
- "loss": 0.1106,
2420
- "step": 3870
2421
- },
2422
- {
2423
- "epoch": 5.73,
2424
- "learning_rate": 0.00021219965449160907,
2425
- "loss": 0.1401,
2426
- "step": 3880
2427
- },
2428
- {
2429
- "epoch": 5.75,
2430
- "learning_rate": 0.0002121419052319842,
2431
- "loss": 0.1725,
2432
- "step": 3890
2433
- },
2434
- {
2435
- "epoch": 5.76,
2436
- "learning_rate": 0.00021208415597235932,
2437
- "loss": 0.1278,
2438
- "step": 3900
2439
- },
2440
- {
2441
- "epoch": 5.78,
2442
- "learning_rate": 0.00021202640671273446,
2443
- "loss": 0.122,
2444
- "step": 3910
2445
- },
2446
- {
2447
- "epoch": 5.79,
2448
- "learning_rate": 0.00021196865745310957,
2449
- "loss": 0.1494,
2450
- "step": 3920
2451
- },
2452
- {
2453
- "epoch": 5.81,
2454
- "learning_rate": 0.00021191090819348469,
2455
- "loss": 0.123,
2456
- "step": 3930
2457
- },
2458
- {
2459
- "epoch": 5.82,
2460
- "learning_rate": 0.00021185315893385983,
2461
- "loss": 0.1845,
2462
- "step": 3940
2463
- },
2464
- {
2465
- "epoch": 5.83,
2466
- "learning_rate": 0.00021179540967423494,
2467
- "loss": 0.1123,
2468
- "step": 3950
2469
- },
2470
- {
2471
- "epoch": 5.85,
2472
- "learning_rate": 0.00021173766041461005,
2473
- "loss": 0.1372,
2474
- "step": 3960
2475
- },
2476
- {
2477
- "epoch": 5.86,
2478
- "learning_rate": 0.0002116799111549852,
2479
- "loss": 0.1328,
2480
- "step": 3970
2481
- },
2482
- {
2483
- "epoch": 5.88,
2484
- "learning_rate": 0.0002116221618953603,
2485
- "loss": 0.1381,
2486
- "step": 3980
2487
- },
2488
- {
2489
- "epoch": 5.89,
2490
- "learning_rate": 0.00021156441263573542,
2491
- "loss": 0.1655,
2492
- "step": 3990
2493
- },
2494
- {
2495
- "epoch": 5.91,
2496
- "learning_rate": 0.00021150666337611056,
2497
- "loss": 0.1214,
2498
- "step": 4000
2499
- },
2500
- {
2501
- "epoch": 5.91,
2502
- "eval_cer": 0.17154218740656582,
2503
- "eval_loss": 0.0955534353852272,
2504
- "eval_runtime": 165.9473,
2505
- "eval_samples_per_second": 14.348,
2506
- "eval_steps_per_second": 0.482,
2507
- "eval_wer": 0.9815203695926081,
2508
- "step": 4000
2509
- },
2510
- {
2511
- "epoch": 5.92,
2512
- "learning_rate": 0.00021144891411648567,
2513
- "loss": 0.1512,
2514
- "step": 4010
2515
- },
2516
- {
2517
- "epoch": 5.94,
2518
- "learning_rate": 0.00021139116485686078,
2519
- "loss": 0.1352,
2520
- "step": 4020
2521
- },
2522
- {
2523
- "epoch": 5.95,
2524
- "learning_rate": 0.00021133341559723592,
2525
- "loss": 0.1583,
2526
- "step": 4030
2527
- },
2528
- {
2529
- "epoch": 5.97,
2530
- "learning_rate": 0.00021127566633761104,
2531
- "loss": 0.1663,
2532
- "step": 4040
2533
- },
2534
- {
2535
- "epoch": 5.98,
2536
- "learning_rate": 0.00021121791707798618,
2537
- "loss": 0.1244,
2538
- "step": 4050
2539
- },
2540
- {
2541
- "epoch": 6.0,
2542
- "learning_rate": 0.0002111601678183613,
2543
- "loss": 0.1249,
2544
- "step": 4060
2545
- },
2546
- {
2547
- "epoch": 6.01,
2548
- "learning_rate": 0.00021110241855873643,
2549
- "loss": 0.1476,
2550
- "step": 4070
2551
- },
2552
- {
2553
- "epoch": 6.03,
2554
- "learning_rate": 0.00021104466929911157,
2555
- "loss": 0.1145,
2556
- "step": 4080
2557
- },
2558
- {
2559
- "epoch": 6.04,
2560
- "learning_rate": 0.00021098692003948668,
2561
- "loss": 0.125,
2562
- "step": 4090
2563
- },
2564
- {
2565
- "epoch": 6.06,
2566
- "learning_rate": 0.0002109291707798618,
2567
- "loss": 0.1257,
2568
- "step": 4100
2569
- },
2570
- {
2571
- "epoch": 6.07,
2572
- "learning_rate": 0.00021087142152023694,
2573
- "loss": 0.1312,
2574
- "step": 4110
2575
- },
2576
- {
2577
- "epoch": 6.09,
2578
- "learning_rate": 0.00021081367226061205,
2579
- "loss": 0.1667,
2580
- "step": 4120
2581
- },
2582
- {
2583
- "epoch": 6.1,
2584
- "learning_rate": 0.00021075592300098716,
2585
- "loss": 0.1385,
2586
- "step": 4130
2587
- },
2588
- {
2589
- "epoch": 6.12,
2590
- "learning_rate": 0.0002106981737413623,
2591
- "loss": 0.1052,
2592
- "step": 4140
2593
- },
2594
- {
2595
- "epoch": 6.13,
2596
- "learning_rate": 0.00021064042448173741,
2597
- "loss": 0.1436,
2598
- "step": 4150
2599
- },
2600
- {
2601
- "epoch": 6.14,
2602
- "learning_rate": 0.00021058267522211253,
2603
- "loss": 0.1238,
2604
- "step": 4160
2605
- },
2606
- {
2607
- "epoch": 6.16,
2608
- "learning_rate": 0.00021052492596248767,
2609
- "loss": 0.1597,
2610
- "step": 4170
2611
- },
2612
- {
2613
- "epoch": 6.17,
2614
- "learning_rate": 0.00021046717670286278,
2615
- "loss": 0.1195,
2616
- "step": 4180
2617
- },
2618
- {
2619
- "epoch": 6.19,
2620
- "learning_rate": 0.0002104094274432379,
2621
- "loss": 0.1236,
2622
- "step": 4190
2623
- },
2624
- {
2625
- "epoch": 6.2,
2626
- "learning_rate": 0.00021035167818361303,
2627
- "loss": 0.131,
2628
- "step": 4200
2629
- },
2630
- {
2631
- "epoch": 6.22,
2632
- "learning_rate": 0.00021029392892398815,
2633
- "loss": 0.1599,
2634
- "step": 4210
2635
- },
2636
- {
2637
- "epoch": 6.23,
2638
- "learning_rate": 0.00021023617966436329,
2639
- "loss": 0.1528,
2640
- "step": 4220
2641
- },
2642
- {
2643
- "epoch": 6.25,
2644
- "learning_rate": 0.0002101784304047384,
2645
- "loss": 0.1046,
2646
- "step": 4230
2647
- },
2648
- {
2649
- "epoch": 6.26,
2650
- "learning_rate": 0.0002101206811451135,
2651
- "loss": 0.1251,
2652
- "step": 4240
2653
- },
2654
- {
2655
- "epoch": 6.28,
2656
- "learning_rate": 0.00021006293188548865,
2657
- "loss": 0.1403,
2658
- "step": 4250
2659
- },
2660
- {
2661
- "epoch": 6.29,
2662
- "learning_rate": 0.00021000518262586376,
2663
- "loss": 0.1351,
2664
- "step": 4260
2665
- },
2666
- {
2667
- "epoch": 6.31,
2668
- "learning_rate": 0.00020994743336623888,
2669
- "loss": 0.1858,
2670
- "step": 4270
2671
- },
2672
- {
2673
- "epoch": 6.32,
2674
- "learning_rate": 0.00020988968410661402,
2675
- "loss": 0.1426,
2676
- "step": 4280
2677
- },
2678
- {
2679
- "epoch": 6.34,
2680
- "learning_rate": 0.00020983193484698913,
2681
- "loss": 0.121,
2682
- "step": 4290
2683
- },
2684
- {
2685
- "epoch": 6.35,
2686
- "learning_rate": 0.00020977418558736424,
2687
- "loss": 0.1312,
2688
- "step": 4300
2689
- },
2690
- {
2691
- "epoch": 6.37,
2692
- "learning_rate": 0.00020971643632773938,
2693
- "loss": 0.1339,
2694
- "step": 4310
2695
- },
2696
- {
2697
- "epoch": 6.38,
2698
- "learning_rate": 0.0002096586870681145,
2699
- "loss": 0.1572,
2700
- "step": 4320
2701
- },
2702
- {
2703
- "epoch": 6.4,
2704
- "learning_rate": 0.0002096009378084896,
2705
- "loss": 0.1363,
2706
- "step": 4330
2707
- },
2708
- {
2709
- "epoch": 6.41,
2710
- "learning_rate": 0.00020954318854886475,
2711
- "loss": 0.1165,
2712
- "step": 4340
2713
- },
2714
- {
2715
- "epoch": 6.43,
2716
- "learning_rate": 0.00020948543928923986,
2717
- "loss": 0.1348,
2718
- "step": 4350
2719
- },
2720
- {
2721
- "epoch": 6.44,
2722
- "learning_rate": 0.000209427690029615,
2723
- "loss": 0.1251,
2724
- "step": 4360
2725
- },
2726
- {
2727
- "epoch": 6.45,
2728
- "learning_rate": 0.00020936994076999012,
2729
- "loss": 0.1619,
2730
- "step": 4370
2731
- },
2732
- {
2733
- "epoch": 6.47,
2734
- "learning_rate": 0.00020931219151036526,
2735
- "loss": 0.1307,
2736
- "step": 4380
2737
- },
2738
- {
2739
- "epoch": 6.48,
2740
- "learning_rate": 0.0002092544422507404,
2741
- "loss": 0.1258,
2742
- "step": 4390
2743
- },
2744
- {
2745
- "epoch": 6.5,
2746
- "learning_rate": 0.0002091966929911155,
2747
- "loss": 0.1169,
2748
- "step": 4400
2749
- },
2750
- {
2751
- "epoch": 6.5,
2752
- "eval_cer": 0.1349219637624828,
2753
- "eval_loss": 0.0982118621468544,
2754
- "eval_runtime": 165.5517,
2755
- "eval_samples_per_second": 14.382,
2756
- "eval_steps_per_second": 0.483,
2757
- "eval_wer": 0.9815203695926081,
2758
- "step": 4400
2759
- },
2760
- {
2761
- "epoch": 6.51,
2762
- "learning_rate": 0.00020913894373149062,
2763
- "loss": 0.1491,
2764
- "step": 4410
2765
- },
2766
- {
2767
- "epoch": 6.53,
2768
- "learning_rate": 0.00020908119447186576,
2769
- "loss": 0.1562,
2770
- "step": 4420
2771
- },
2772
- {
2773
- "epoch": 6.54,
2774
- "learning_rate": 0.00020902344521224087,
2775
- "loss": 0.1651,
2776
- "step": 4430
2777
- },
2778
- {
2779
- "epoch": 6.56,
2780
- "learning_rate": 0.000208965695952616,
2781
- "loss": 0.1447,
2782
- "step": 4440
2783
- },
2784
- {
2785
- "epoch": 6.57,
2786
- "learning_rate": 0.00020890794669299113,
2787
- "loss": 0.1284,
2788
- "step": 4450
2789
- },
2790
- {
2791
- "epoch": 6.59,
2792
- "learning_rate": 0.00020885019743336624,
2793
- "loss": 0.1374,
2794
- "step": 4460
2795
- },
2796
- {
2797
- "epoch": 6.6,
2798
- "learning_rate": 0.00020879244817374135,
2799
- "loss": 0.1388,
2800
- "step": 4470
2801
- },
2802
- {
2803
- "epoch": 6.62,
2804
- "learning_rate": 0.0002087346989141165,
2805
- "loss": 0.1255,
2806
- "step": 4480
2807
- },
2808
- {
2809
- "epoch": 6.63,
2810
- "learning_rate": 0.0002086769496544916,
2811
- "loss": 0.1194,
2812
- "step": 4490
2813
- },
2814
- {
2815
- "epoch": 6.65,
2816
- "learning_rate": 0.00020861920039486672,
2817
- "loss": 0.1333,
2818
- "step": 4500
2819
- },
2820
- {
2821
- "epoch": 6.66,
2822
- "learning_rate": 0.00020856145113524186,
2823
- "loss": 0.1433,
2824
- "step": 4510
2825
- },
2826
- {
2827
- "epoch": 6.68,
2828
- "learning_rate": 0.00020850370187561697,
2829
- "loss": 0.143,
2830
- "step": 4520
2831
- },
2832
- {
2833
- "epoch": 6.69,
2834
- "learning_rate": 0.0002084459526159921,
2835
- "loss": 0.1392,
2836
- "step": 4530
2837
- },
2838
- {
2839
- "epoch": 6.71,
2840
- "learning_rate": 0.00020838820335636722,
2841
- "loss": 0.1325,
2842
- "step": 4540
2843
- },
2844
- {
2845
- "epoch": 6.72,
2846
- "learning_rate": 0.00020833045409674234,
2847
- "loss": 0.1197,
2848
- "step": 4550
2849
- },
2850
- {
2851
- "epoch": 6.74,
2852
- "learning_rate": 0.00020827270483711748,
2853
- "loss": 0.1346,
2854
- "step": 4560
2855
- },
2856
- {
2857
- "epoch": 6.75,
2858
- "learning_rate": 0.0002082149555774926,
2859
- "loss": 0.1403,
2860
- "step": 4570
2861
- },
2862
- {
2863
- "epoch": 6.77,
2864
- "learning_rate": 0.0002081572063178677,
2865
- "loss": 0.1181,
2866
- "step": 4580
2867
- },
2868
- {
2869
- "epoch": 6.78,
2870
- "learning_rate": 0.00020809945705824284,
2871
- "loss": 0.1239,
2872
- "step": 4590
2873
- },
2874
- {
2875
- "epoch": 6.79,
2876
- "learning_rate": 0.00020804170779861796,
2877
- "loss": 0.1675,
2878
- "step": 4600
2879
- },
2880
- {
2881
- "epoch": 6.81,
2882
- "learning_rate": 0.00020798395853899307,
2883
- "loss": 0.1435,
2884
- "step": 4610
2885
- },
2886
- {
2887
- "epoch": 6.82,
2888
- "learning_rate": 0.0002079262092793682,
2889
- "loss": 0.1441,
2890
- "step": 4620
2891
- },
2892
- {
2893
- "epoch": 6.84,
2894
- "learning_rate": 0.00020786846001974332,
2895
- "loss": 0.1482,
2896
- "step": 4630
2897
- },
2898
- {
2899
- "epoch": 6.85,
2900
- "learning_rate": 0.00020781071076011844,
2901
- "loss": 0.1217,
2902
- "step": 4640
2903
- },
2904
- {
2905
- "epoch": 6.87,
2906
- "learning_rate": 0.00020775296150049358,
2907
- "loss": 0.1092,
2908
- "step": 4650
2909
- },
2910
- {
2911
- "epoch": 6.88,
2912
- "learning_rate": 0.0002076952122408687,
2913
- "loss": 0.1154,
2914
- "step": 4660
2915
- },
2916
- {
2917
- "epoch": 6.9,
2918
- "learning_rate": 0.00020763746298124383,
2919
- "loss": 0.1945,
2920
- "step": 4670
2921
- },
2922
- {
2923
- "epoch": 6.91,
2924
- "learning_rate": 0.00020757971372161894,
2925
- "loss": 0.1312,
2926
- "step": 4680
2927
- },
2928
- {
2929
- "epoch": 6.93,
2930
- "learning_rate": 0.00020752196446199405,
2931
- "loss": 0.1162,
2932
- "step": 4690
2933
- },
2934
- {
2935
- "epoch": 6.94,
2936
- "learning_rate": 0.00020746421520236922,
2937
- "loss": 0.1188,
2938
- "step": 4700
2939
- },
2940
- {
2941
- "epoch": 6.96,
2942
- "learning_rate": 0.00020740646594274433,
2943
- "loss": 0.1391,
2944
- "step": 4710
2945
- },
2946
- {
2947
- "epoch": 6.97,
2948
- "learning_rate": 0.00020734871668311945,
2949
- "loss": 0.1672,
2950
- "step": 4720
2951
- },
2952
- {
2953
- "epoch": 6.99,
2954
- "learning_rate": 0.0002072909674234946,
2955
- "loss": 0.1429,
2956
- "step": 4730
2957
- },
2958
- {
2959
- "epoch": 7.0,
2960
- "learning_rate": 0.0002072332181638697,
2961
- "loss": 0.1438,
2962
- "step": 4740
2963
- },
2964
- {
2965
- "epoch": 7.02,
2966
- "learning_rate": 0.0002071754689042448,
2967
- "loss": 0.1274,
2968
- "step": 4750
2969
- },
2970
- {
2971
- "epoch": 7.03,
2972
- "learning_rate": 0.00020711771964461995,
2973
- "loss": 0.1086,
2974
- "step": 4760
2975
- },
2976
- {
2977
- "epoch": 7.05,
2978
- "learning_rate": 0.00020705997038499507,
2979
- "loss": 0.1207,
2980
- "step": 4770
2981
- },
2982
- {
2983
- "epoch": 7.06,
2984
- "learning_rate": 0.00020700222112537018,
2985
- "loss": 0.1179,
2986
- "step": 4780
2987
- },
2988
- {
2989
- "epoch": 7.08,
2990
- "learning_rate": 0.00020694447186574532,
2991
- "loss": 0.1464,
2992
- "step": 4790
2993
- },
2994
- {
2995
- "epoch": 7.09,
2996
- "learning_rate": 0.00020688672260612043,
2997
- "loss": 0.1115,
2998
- "step": 4800
2999
- },
3000
- {
3001
- "epoch": 7.09,
3002
- "eval_cer": 0.15084614004664235,
3003
- "eval_loss": 0.10529889166355133,
3004
- "eval_runtime": 167.4152,
3005
- "eval_samples_per_second": 14.222,
3006
- "eval_steps_per_second": 0.478,
3007
- "eval_wer": 0.9844603107937842,
3008
- "step": 4800
3009
- },
3010
- {
3011
- "epoch": 7.1,
3012
- "learning_rate": 0.00020682897334649554,
3013
- "loss": 0.1201,
3014
- "step": 4810
3015
- },
3016
- {
3017
- "epoch": 7.12,
3018
- "learning_rate": 0.00020677122408687068,
3019
- "loss": 0.1359,
3020
- "step": 4820
3021
- },
3022
- {
3023
- "epoch": 7.13,
3024
- "learning_rate": 0.0002067134748272458,
3025
- "loss": 0.1161,
3026
- "step": 4830
3027
- },
3028
- {
3029
- "epoch": 7.15,
3030
- "learning_rate": 0.00020665572556762094,
3031
- "loss": 0.1579,
3032
- "step": 4840
3033
- },
3034
- {
3035
- "epoch": 7.16,
3036
- "learning_rate": 0.00020659797630799605,
3037
- "loss": 0.1196,
3038
- "step": 4850
3039
- },
3040
- {
3041
- "epoch": 7.18,
3042
- "learning_rate": 0.00020654022704837116,
3043
- "loss": 0.1122,
3044
- "step": 4860
3045
- },
3046
- {
3047
- "epoch": 7.19,
3048
- "learning_rate": 0.0002064824777887463,
3049
- "loss": 0.1323,
3050
- "step": 4870
3051
- },
3052
- {
3053
- "epoch": 7.21,
3054
- "learning_rate": 0.00020642472852912142,
3055
- "loss": 0.14,
3056
- "step": 4880
3057
- },
3058
- {
3059
- "epoch": 7.22,
3060
- "learning_rate": 0.00020636697926949653,
3061
- "loss": 0.1683,
3062
- "step": 4890
3063
- },
3064
- {
3065
- "epoch": 7.24,
3066
- "learning_rate": 0.00020630923000987167,
3067
- "loss": 0.1203,
3068
- "step": 4900
3069
- },
3070
- {
3071
- "epoch": 7.25,
3072
- "learning_rate": 0.00020625148075024678,
3073
- "loss": 0.1275,
3074
- "step": 4910
3075
- },
3076
- {
3077
- "epoch": 7.27,
3078
- "learning_rate": 0.0002061937314906219,
3079
- "loss": 0.1206,
3080
- "step": 4920
3081
- },
3082
- {
3083
- "epoch": 7.28,
3084
- "learning_rate": 0.00020613598223099704,
3085
- "loss": 0.127,
3086
- "step": 4930
3087
- },
3088
- {
3089
- "epoch": 7.3,
3090
- "learning_rate": 0.00020607823297137215,
3091
- "loss": 0.1551,
3092
- "step": 4940
3093
- },
3094
- {
3095
- "epoch": 7.31,
3096
- "learning_rate": 0.00020602048371174726,
3097
- "loss": 0.1446,
3098
- "step": 4950
3099
- },
3100
- {
3101
- "epoch": 7.33,
3102
- "learning_rate": 0.0002059627344521224,
3103
- "loss": 0.1283,
3104
- "step": 4960
3105
- },
3106
- {
3107
- "epoch": 7.34,
3108
- "learning_rate": 0.00020590498519249751,
3109
- "loss": 0.1253,
3110
- "step": 4970
3111
- },
3112
- {
3113
- "epoch": 7.36,
3114
- "learning_rate": 0.00020584723593287265,
3115
- "loss": 0.1143,
3116
- "step": 4980
3117
- },
3118
- {
3119
- "epoch": 7.37,
3120
- "learning_rate": 0.00020578948667324777,
3121
- "loss": 0.1261,
3122
- "step": 4990
3123
- },
3124
- {
3125
- "epoch": 7.39,
3126
- "learning_rate": 0.00020573173741362288,
3127
- "loss": 0.1282,
3128
- "step": 5000
3129
- },
3130
- {
3131
- "epoch": 7.4,
3132
- "learning_rate": 0.00020567398815399802,
3133
- "loss": 0.1288,
3134
- "step": 5010
3135
- },
3136
- {
3137
- "epoch": 7.42,
3138
- "learning_rate": 0.00020561623889437316,
3139
- "loss": 0.105,
3140
- "step": 5020
3141
- },
3142
- {
3143
- "epoch": 7.43,
3144
- "learning_rate": 0.00020555848963474827,
3145
- "loss": 0.1218,
3146
- "step": 5030
3147
- },
3148
- {
3149
- "epoch": 7.44,
3150
- "learning_rate": 0.0002055007403751234,
3151
- "loss": 0.1393,
3152
- "step": 5040
3153
- },
3154
- {
3155
- "epoch": 7.46,
3156
- "learning_rate": 0.00020544299111549853,
3157
- "loss": 0.1122,
3158
- "step": 5050
3159
- },
3160
- {
3161
- "epoch": 7.47,
3162
- "learning_rate": 0.00020538524185587364,
3163
- "loss": 0.1103,
3164
- "step": 5060
3165
- },
3166
- {
3167
- "epoch": 7.49,
3168
- "learning_rate": 0.00020532749259624878,
3169
- "loss": 0.0996,
3170
- "step": 5070
3171
- },
3172
- {
3173
- "epoch": 7.5,
3174
- "learning_rate": 0.0002052697433366239,
3175
- "loss": 0.1136,
3176
- "step": 5080
3177
- },
3178
- {
3179
- "epoch": 7.52,
3180
- "learning_rate": 0.000205211994076999,
3181
- "loss": 0.1593,
3182
- "step": 5090
3183
- },
3184
- {
3185
- "epoch": 7.53,
3186
- "learning_rate": 0.00020515424481737414,
3187
- "loss": 0.1223,
3188
- "step": 5100
3189
- },
3190
- {
3191
- "epoch": 7.55,
3192
- "learning_rate": 0.00020509649555774926,
3193
- "loss": 0.1164,
3194
- "step": 5110
3195
- },
3196
- {
3197
- "epoch": 7.56,
3198
- "learning_rate": 0.00020503874629812437,
3199
- "loss": 0.1101,
3200
- "step": 5120
3201
- },
3202
- {
3203
- "epoch": 7.58,
3204
- "learning_rate": 0.0002049809970384995,
3205
- "loss": 0.1082,
3206
- "step": 5130
3207
- },
3208
- {
3209
- "epoch": 7.59,
3210
- "learning_rate": 0.00020492324777887462,
3211
- "loss": 0.1355,
3212
- "step": 5140
3213
- },
3214
- {
3215
- "epoch": 7.61,
3216
- "learning_rate": 0.00020486549851924976,
3217
- "loss": 0.1187,
3218
- "step": 5150
3219
- },
3220
- {
3221
- "epoch": 7.62,
3222
- "learning_rate": 0.00020480774925962488,
3223
- "loss": 0.1147,
3224
- "step": 5160
3225
- },
3226
- {
3227
- "epoch": 7.64,
3228
- "learning_rate": 0.00020475,
3229
- "loss": 0.1072,
3230
- "step": 5170
3231
- },
3232
- {
3233
- "epoch": 7.65,
3234
- "learning_rate": 0.00020469225074037513,
3235
- "loss": 0.1298,
3236
- "step": 5180
3237
- },
3238
- {
3239
- "epoch": 7.67,
3240
- "learning_rate": 0.00020463450148075024,
3241
- "loss": 0.1563,
3242
- "step": 5190
3243
- },
3244
- {
3245
- "epoch": 7.68,
3246
- "learning_rate": 0.00020457675222112536,
3247
- "loss": 0.1066,
3248
- "step": 5200
3249
- },
3250
- {
3251
- "epoch": 7.68,
3252
- "eval_cer": 0.10776774502182623,
3253
- "eval_loss": 0.09926649183034897,
3254
- "eval_runtime": 165.9352,
3255
- "eval_samples_per_second": 14.349,
3256
- "eval_steps_per_second": 0.482,
3257
- "eval_wer": 0.9601007979840404,
3258
- "step": 5200
3259
  }
3260
  ],
3261
- "max_steps": 40620,
3262
  "num_train_epochs": 60,
3263
- "total_flos": 5.489095218329608e+19,
3264
  "trial_name": null,
3265
  "trial_params": null
3266
  }
 
1
  {
2
+ "best_metric": 0.15721334517002106,
3
+ "best_model_checkpoint": "voidful/wav2vec2-xlsr-53-espeak-librispeech-ft-all/checkpoint-2400",
4
+ "epoch": 0.35046728971962615,
5
+ "global_step": 2400,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2.34e-05,
13
+ "loss": 0.2473,
14
  "step": 10
15
  },
16
  {
17
+ "epoch": 0.0,
18
+ "learning_rate": 4.68e-05,
19
+ "loss": 0.2506,
20
  "step": 20
21
  },
22
  {
23
+ "epoch": 0.0,
24
+ "learning_rate": 7.02e-05,
25
+ "loss": 0.2243,
26
  "step": 30
27
  },
28
  {
29
+ "epoch": 0.01,
30
+ "learning_rate": 9.36e-05,
31
+ "loss": 0.2569,
32
  "step": 40
33
  },
34
  {
35
+ "epoch": 0.01,
36
+ "learning_rate": 0.000117,
37
+ "loss": 0.429,
38
  "step": 50
39
  },
40
  {
41
+ "epoch": 0.01,
42
+ "learning_rate": 0.0001404,
43
+ "loss": 0.2216,
44
  "step": 60
45
  },
46
  {
47
+ "epoch": 0.01,
48
+ "learning_rate": 0.0001638,
49
+ "loss": 0.2044,
50
  "step": 70
51
  },
52
  {
53
+ "epoch": 0.01,
54
+ "learning_rate": 0.0001872,
55
+ "loss": 0.2037,
56
  "step": 80
57
  },
58
  {
59
+ "epoch": 0.01,
60
+ "learning_rate": 0.0002106,
61
+ "loss": 0.2444,
62
  "step": 90
63
  },
64
  {
65
+ "epoch": 0.01,
66
+ "learning_rate": 0.000234,
67
+ "loss": 0.4259,
68
  "step": 100
69
  },
70
  {
71
+ "epoch": 0.02,
72
+ "learning_rate": 0.00023399430352013243,
73
+ "loss": 0.2284,
74
  "step": 110
75
  },
76
  {
77
+ "epoch": 0.02,
78
+ "learning_rate": 0.00023398860704026487,
79
+ "loss": 0.2131,
80
  "step": 120
81
  },
82
  {
83
+ "epoch": 0.02,
84
+ "learning_rate": 0.00023398291056039728,
85
+ "loss": 0.2125,
86
  "step": 130
87
  },
88
  {
89
+ "epoch": 0.02,
90
+ "learning_rate": 0.0002339772140805297,
91
+ "loss": 0.2449,
92
  "step": 140
93
  },
94
  {
95
+ "epoch": 0.02,
96
+ "learning_rate": 0.00023397151760066215,
97
+ "loss": 0.3585,
98
  "step": 150
99
  },
100
  {
101
+ "epoch": 0.02,
102
+ "learning_rate": 0.00023396582112079458,
103
+ "loss": 0.2464,
104
  "step": 160
105
  },
106
  {
107
+ "epoch": 0.02,
108
+ "learning_rate": 0.00023396012464092702,
109
+ "loss": 0.2342,
110
  "step": 170
111
  },
112
  {
113
+ "epoch": 0.03,
114
+ "learning_rate": 0.00023395442816105943,
115
+ "loss": 0.2261,
116
  "step": 180
117
  },
118
  {
119
+ "epoch": 0.03,
120
+ "learning_rate": 0.00023394873168119186,
121
+ "loss": 0.2682,
122
  "step": 190
123
  },
124
  {
125
+ "epoch": 0.03,
126
+ "learning_rate": 0.0002339430352013243,
127
+ "loss": 0.3853,
128
  "step": 200
129
  },
130
  {
131
+ "epoch": 0.03,
132
+ "learning_rate": 0.00023393733872145673,
133
+ "loss": 0.2098,
134
  "step": 210
135
  },
136
  {
137
+ "epoch": 0.03,
138
+ "learning_rate": 0.00023393164224158917,
139
+ "loss": 0.2502,
140
  "step": 220
141
  },
142
  {
143
+ "epoch": 0.03,
144
+ "learning_rate": 0.00023392594576172158,
145
+ "loss": 0.2289,
146
  "step": 230
147
  },
148
  {
149
+ "epoch": 0.04,
150
+ "learning_rate": 0.000233920249281854,
151
+ "loss": 0.2539,
152
  "step": 240
153
  },
154
  {
155
+ "epoch": 0.04,
156
+ "learning_rate": 0.00023391455280198645,
157
+ "loss": 0.3958,
158
  "step": 250
159
  },
160
  {
161
+ "epoch": 0.04,
162
+ "learning_rate": 0.00023390885632211888,
163
+ "loss": 0.2165,
164
  "step": 260
165
  },
166
  {
167
+ "epoch": 0.04,
168
+ "learning_rate": 0.00023390315984225135,
169
+ "loss": 0.215,
170
  "step": 270
171
  },
172
  {
173
+ "epoch": 0.04,
174
+ "learning_rate": 0.00023389746336238375,
175
+ "loss": 0.2255,
176
  "step": 280
177
  },
178
  {
179
+ "epoch": 0.04,
180
+ "learning_rate": 0.0002338917668825162,
181
+ "loss": 0.2072,
182
  "step": 290
183
  },
184
  {
185
+ "epoch": 0.04,
186
+ "learning_rate": 0.00023388607040264863,
187
+ "loss": 0.3403,
188
  "step": 300
189
  },
190
  {
191
+ "epoch": 0.05,
192
+ "learning_rate": 0.00023388037392278106,
193
+ "loss": 0.2147,
194
  "step": 310
195
  },
196
  {
197
+ "epoch": 0.05,
198
+ "learning_rate": 0.0002338746774429135,
199
+ "loss": 0.1936,
200
  "step": 320
201
  },
202
  {
203
+ "epoch": 0.05,
204
+ "learning_rate": 0.0002338689809630459,
205
+ "loss": 0.2352,
206
  "step": 330
207
  },
208
  {
209
+ "epoch": 0.05,
210
+ "learning_rate": 0.00023386328448317834,
211
+ "loss": 0.2696,
212
  "step": 340
213
  },
214
  {
215
+ "epoch": 0.05,
216
+ "learning_rate": 0.00023385758800331078,
217
+ "loss": 0.3343,
218
  "step": 350
219
  },
220
  {
221
+ "epoch": 0.05,
222
+ "learning_rate": 0.0002338518915234432,
223
+ "loss": 0.202,
224
  "step": 360
225
  },
226
  {
227
+ "epoch": 0.05,
228
+ "learning_rate": 0.00023384619504357565,
229
+ "loss": 0.2138,
230
  "step": 370
231
  },
232
  {
233
+ "epoch": 0.06,
234
+ "learning_rate": 0.00023384049856370806,
235
+ "loss": 0.2142,
236
  "step": 380
237
  },
238
  {
239
+ "epoch": 0.06,
240
+ "learning_rate": 0.0002338348020838405,
241
+ "loss": 0.2535,
242
  "step": 390
243
  },
244
  {
245
+ "epoch": 0.06,
246
+ "learning_rate": 0.00023382910560397293,
247
+ "loss": 0.392,
248
  "step": 400
249
  },
250
  {
251
+ "epoch": 0.06,
252
+ "eval_cer": 0.13186392030593033,
253
+ "eval_loss": 0.17454230785369873,
254
+ "eval_runtime": 360.3554,
255
+ "eval_samples_per_second": 14.516,
256
+ "eval_steps_per_second": 0.486,
257
+ "eval_wer": 0.9709424584209521,
258
  "step": 400
259
  },
260
  {
261
+ "epoch": 0.06,
262
+ "learning_rate": 0.00023382340912410536,
263
+ "loss": 0.225,
264
  "step": 410
265
  },
266
  {
267
+ "epoch": 0.06,
268
+ "learning_rate": 0.0002338177126442378,
269
+ "loss": 0.2077,
270
  "step": 420
271
  },
272
  {
273
+ "epoch": 0.06,
274
+ "learning_rate": 0.0002338120161643702,
275
+ "loss": 0.2306,
276
  "step": 430
277
  },
278
  {
279
+ "epoch": 0.06,
280
+ "learning_rate": 0.00023380631968450264,
281
+ "loss": 0.2642,
282
  "step": 440
283
  },
284
  {
285
+ "epoch": 0.07,
286
+ "learning_rate": 0.00023380119285262182,
287
+ "loss": 0.3968,
288
  "step": 450
289
  },
290
  {
291
+ "epoch": 0.07,
292
+ "learning_rate": 0.00023379549637275425,
293
+ "loss": 0.1817,
294
  "step": 460
295
  },
296
  {
297
+ "epoch": 0.07,
298
+ "learning_rate": 0.0002337897998928867,
299
+ "loss": 0.2048,
300
  "step": 470
301
  },
302
  {
303
+ "epoch": 0.07,
304
+ "learning_rate": 0.00023378410341301912,
305
+ "loss": 0.211,
306
  "step": 480
307
  },
308
  {
309
+ "epoch": 0.07,
310
+ "learning_rate": 0.00023377840693315156,
311
+ "loss": 0.2555,
312
  "step": 490
313
  },
314
  {
315
+ "epoch": 0.07,
316
+ "learning_rate": 0.000233772710453284,
317
+ "loss": 0.3959,
318
  "step": 500
319
  },
320
  {
321
+ "epoch": 0.07,
322
+ "learning_rate": 0.00023376701397341643,
323
+ "loss": 0.2082,
324
  "step": 510
325
  },
326
  {
327
+ "epoch": 0.08,
328
+ "learning_rate": 0.00023376131749354887,
329
+ "loss": 0.1888,
330
  "step": 520
331
  },
332
  {
333
+ "epoch": 0.08,
334
+ "learning_rate": 0.0002337556210136813,
335
+ "loss": 0.2242,
336
  "step": 530
337
  },
338
  {
339
+ "epoch": 0.08,
340
+ "learning_rate": 0.0002337499245338137,
341
+ "loss": 0.2365,
342
  "step": 540
343
  },
344
  {
345
+ "epoch": 0.08,
346
+ "learning_rate": 0.00023374422805394615,
347
+ "loss": 0.3182,
348
  "step": 550
349
  },
350
  {
351
+ "epoch": 0.08,
352
+ "learning_rate": 0.00023373853157407858,
353
+ "loss": 0.2074,
354
  "step": 560
355
  },
356
  {
357
+ "epoch": 0.08,
358
+ "learning_rate": 0.00023373283509421102,
359
+ "loss": 0.2005,
360
  "step": 570
361
  },
362
  {
363
+ "epoch": 0.08,
364
+ "learning_rate": 0.00023372713861434345,
365
+ "loss": 0.2163,
366
  "step": 580
367
  },
368
  {
369
+ "epoch": 0.09,
370
+ "learning_rate": 0.0002337214421344759,
371
+ "loss": 0.2461,
372
  "step": 590
373
  },
374
  {
375
+ "epoch": 0.09,
376
+ "learning_rate": 0.0002337157456546083,
377
+ "loss": 0.3741,
378
  "step": 600
379
  },
380
  {
381
+ "epoch": 0.09,
382
+ "learning_rate": 0.00023371061882272747,
383
+ "loss": 0.21,
384
  "step": 610
385
  },
386
  {
387
+ "epoch": 0.09,
388
+ "learning_rate": 0.0002337049223428599,
389
+ "loss": 0.2095,
390
  "step": 620
391
  },
392
  {
393
+ "epoch": 0.09,
394
+ "learning_rate": 0.00023369922586299234,
395
+ "loss": 0.1971,
396
  "step": 630
397
  },
398
  {
399
+ "epoch": 0.09,
400
+ "learning_rate": 0.00023369352938312478,
401
+ "loss": 0.2462,
402
  "step": 640
403
  },
404
  {
405
+ "epoch": 0.09,
406
+ "learning_rate": 0.00023368783290325722,
407
+ "loss": 0.3378,
408
  "step": 650
409
  },
410
  {
411
+ "epoch": 0.1,
412
+ "learning_rate": 0.00023368213642338962,
413
+ "loss": 0.2211,
414
  "step": 660
415
  },
416
  {
417
+ "epoch": 0.1,
418
+ "learning_rate": 0.00023367643994352206,
419
+ "loss": 0.1843,
420
  "step": 670
421
  },
422
  {
423
+ "epoch": 0.1,
424
+ "learning_rate": 0.0002336707434636545,
425
+ "loss": 0.198,
426
  "step": 680
427
  },
428
  {
429
+ "epoch": 0.1,
430
+ "learning_rate": 0.00023366504698378693,
431
+ "loss": 0.2688,
432
  "step": 690
433
  },
434
  {
435
+ "epoch": 0.1,
436
+ "learning_rate": 0.00023365935050391937,
437
+ "loss": 0.3775,
438
  "step": 700
439
  },
440
  {
441
+ "epoch": 0.1,
442
+ "learning_rate": 0.00023365365402405177,
443
+ "loss": 0.1945,
444
  "step": 710
445
  },
446
  {
447
+ "epoch": 0.11,
448
+ "learning_rate": 0.00023364795754418424,
449
+ "loss": 0.2172,
450
  "step": 720
451
  },
452
  {
453
+ "epoch": 0.11,
454
+ "learning_rate": 0.00023364226106431667,
455
+ "loss": 0.2145,
456
  "step": 730
457
  },
458
  {
459
+ "epoch": 0.11,
460
+ "learning_rate": 0.0002336365645844491,
461
+ "loss": 0.2712,
462
  "step": 740
463
  },
464
  {
465
+ "epoch": 0.11,
466
+ "learning_rate": 0.00023363086810458154,
467
+ "loss": 0.3371,
468
  "step": 750
469
  },
470
  {
471
+ "epoch": 0.11,
472
+ "learning_rate": 0.00023362517162471395,
473
+ "loss": 0.2176,
474
  "step": 760
475
  },
476
  {
477
+ "epoch": 0.11,
478
+ "learning_rate": 0.0002336194751448464,
479
+ "loss": 0.1926,
480
  "step": 770
481
  },
482
  {
483
+ "epoch": 0.11,
484
+ "learning_rate": 0.00023361377866497882,
485
+ "loss": 0.2002,
486
  "step": 780
487
  },
488
  {
489
+ "epoch": 0.12,
490
+ "learning_rate": 0.00023360808218511126,
491
+ "loss": 0.2345,
492
  "step": 790
493
  },
494
  {
495
+ "epoch": 0.12,
496
+ "learning_rate": 0.0002336023857052437,
497
+ "loss": 0.3921,
498
  "step": 800
499
  },
500
  {
501
+ "epoch": 0.12,
502
+ "eval_cer": 0.10110303888191077,
503
+ "eval_loss": 0.17119961977005005,
504
+ "eval_runtime": 359.8523,
505
+ "eval_samples_per_second": 14.537,
506
+ "eval_steps_per_second": 0.486,
507
+ "eval_wer": 0.941884916841904,
508
  "step": 800
509
  },
510
  {
511
+ "epoch": 0.12,
512
+ "learning_rate": 0.0002335966892253761,
513
+ "loss": 0.2154,
514
  "step": 810
515
  },
516
  {
517
+ "epoch": 0.12,
518
+ "learning_rate": 0.00023359099274550854,
519
+ "loss": 0.213,
520
  "step": 820
521
  },
522
  {
523
+ "epoch": 0.12,
524
+ "learning_rate": 0.00023358529626564097,
525
+ "loss": 0.1914,
526
  "step": 830
527
  },
528
  {
529
+ "epoch": 0.12,
530
+ "learning_rate": 0.0002335795997857734,
531
+ "loss": 0.2188,
532
  "step": 840
533
  },
534
  {
535
+ "epoch": 0.12,
536
+ "learning_rate": 0.00023357390330590585,
537
+ "loss": 0.344,
538
  "step": 850
539
  },
540
  {
541
+ "epoch": 0.13,
542
+ "learning_rate": 0.00023356820682603825,
543
+ "loss": 0.1963,
544
  "step": 860
545
  },
546
  {
547
+ "epoch": 0.13,
548
+ "learning_rate": 0.0002335625103461707,
549
+ "loss": 0.198,
550
  "step": 870
551
  },
552
  {
553
+ "epoch": 0.13,
554
+ "learning_rate": 0.00023355681386630312,
555
+ "loss": 0.2368,
556
  "step": 880
557
  },
558
  {
559
+ "epoch": 0.13,
560
+ "learning_rate": 0.00023355111738643556,
561
+ "loss": 0.2333,
562
  "step": 890
563
  },
564
  {
565
+ "epoch": 0.13,
566
+ "learning_rate": 0.000233545420906568,
567
+ "loss": 0.3454,
568
  "step": 900
569
  },
570
  {
571
+ "epoch": 0.13,
572
+ "learning_rate": 0.0002335397244267004,
573
+ "loss": 0.21,
574
  "step": 910
575
  },
576
  {
577
+ "epoch": 0.13,
578
+ "learning_rate": 0.00023353402794683284,
579
+ "loss": 0.1849,
580
  "step": 920
581
  },
582
  {
583
+ "epoch": 0.14,
584
+ "learning_rate": 0.00023352833146696528,
585
+ "loss": 0.1982,
586
  "step": 930
587
  },
588
  {
589
+ "epoch": 0.14,
590
+ "learning_rate": 0.0002335226349870977,
591
+ "loss": 0.2371,
592
  "step": 940
593
  },
594
  {
595
+ "epoch": 0.14,
596
+ "learning_rate": 0.00023351750815521691,
597
+ "loss": 0.346,
598
  "step": 950
599
  },
600
  {
601
+ "epoch": 0.14,
602
+ "learning_rate": 0.00023351181167534935,
603
+ "loss": 0.2024,
604
  "step": 960
605
  },
606
  {
607
+ "epoch": 0.14,
608
+ "learning_rate": 0.00023350611519548179,
609
+ "loss": 0.1852,
610
  "step": 970
611
  },
612
  {
613
+ "epoch": 0.14,
614
+ "learning_rate": 0.0002335004187156142,
615
+ "loss": 0.2196,
616
  "step": 980
617
  },
618
  {
619
+ "epoch": 0.14,
620
+ "learning_rate": 0.00023349472223574663,
621
+ "loss": 0.2333,
622
  "step": 990
623
  },
624
  {
625
+ "epoch": 0.15,
626
+ "learning_rate": 0.00023348902575587906,
627
+ "loss": 0.3275,
628
  "step": 1000
629
  },
630
  {
631
+ "epoch": 0.15,
632
+ "learning_rate": 0.0002334833292760115,
633
+ "loss": 0.1848,
634
  "step": 1010
635
  },
636
  {
637
+ "epoch": 0.15,
638
+ "learning_rate": 0.00023347763279614394,
639
+ "loss": 0.1784,
640
  "step": 1020
641
  },
642
  {
643
+ "epoch": 0.15,
644
+ "learning_rate": 0.00023347193631627634,
645
+ "loss": 0.1954,
646
  "step": 1030
647
  },
648
  {
649
+ "epoch": 0.15,
650
+ "learning_rate": 0.00023346623983640878,
651
+ "loss": 0.211,
652
  "step": 1040
653
  },
654
  {
655
+ "epoch": 0.15,
656
+ "learning_rate": 0.00023346054335654122,
657
+ "loss": 0.3037,
658
  "step": 1050
659
  },
660
  {
661
+ "epoch": 0.15,
662
+ "learning_rate": 0.00023345484687667365,
663
+ "loss": 0.2249,
664
  "step": 1060
665
  },
666
  {
667
+ "epoch": 0.16,
668
+ "learning_rate": 0.0002334491503968061,
669
+ "loss": 0.1877,
670
  "step": 1070
671
  },
672
  {
673
+ "epoch": 0.16,
674
+ "learning_rate": 0.0002334434539169385,
675
+ "loss": 0.1877,
676
  "step": 1080
677
  },
678
  {
679
+ "epoch": 0.16,
680
+ "learning_rate": 0.00023343775743707093,
681
+ "loss": 0.2514,
682
  "step": 1090
683
  },
684
  {
685
+ "epoch": 0.16,
686
+ "learning_rate": 0.00023343206095720337,
687
+ "loss": 0.3478,
688
  "step": 1100
689
  },
690
  {
691
+ "epoch": 0.16,
692
+ "learning_rate": 0.0002334263644773358,
693
+ "loss": 0.2016,
694
  "step": 1110
695
  },
696
  {
697
+ "epoch": 0.16,
698
+ "learning_rate": 0.00023342066799746824,
699
+ "loss": 0.1919,
700
  "step": 1120
701
  },
702
  {
703
+ "epoch": 0.17,
704
+ "learning_rate": 0.00023341497151760065,
705
+ "loss": 0.2027,
706
  "step": 1130
707
  },
708
  {
709
+ "epoch": 0.17,
710
+ "learning_rate": 0.00023340927503773308,
711
+ "loss": 0.2362,
712
  "step": 1140
713
  },
714
  {
715
+ "epoch": 0.17,
716
+ "learning_rate": 0.00023340357855786552,
717
+ "loss": 0.3776,
718
  "step": 1150
719
  },
720
  {
721
+ "epoch": 0.17,
722
+ "learning_rate": 0.00023339788207799795,
723
+ "loss": 0.2072,
724
  "step": 1160
725
  },
726
  {
727
+ "epoch": 0.17,
728
+ "learning_rate": 0.0002333921855981304,
729
+ "loss": 0.21,
730
  "step": 1170
731
  },
732
  {
733
+ "epoch": 0.17,
734
+ "learning_rate": 0.0002333864891182628,
735
+ "loss": 0.2136,
736
  "step": 1180
737
  },
738
  {
739
+ "epoch": 0.17,
740
+ "learning_rate": 0.00023338079263839523,
741
+ "loss": 0.2236,
742
  "step": 1190
743
  },
744
  {
745
+ "epoch": 0.18,
746
+ "learning_rate": 0.00023337509615852767,
747
+ "loss": 0.3523,
748
  "step": 1200
749
  },
750
  {
751
+ "epoch": 0.18,
752
+ "eval_cer": 0.10916811901518182,
753
+ "eval_loss": 0.16797101497650146,
754
+ "eval_runtime": 357.2006,
755
+ "eval_samples_per_second": 14.644,
756
+ "eval_steps_per_second": 0.49,
757
+ "eval_wer": 0.9625310648059644,
758
  "step": 1200
759
  },
760
  {
761
+ "epoch": 0.18,
762
+ "learning_rate": 0.0002333693996786601,
763
+ "loss": 0.1918,
764
  "step": 1210
765
  },
766
  {
767
+ "epoch": 0.18,
768
+ "learning_rate": 0.00023336370319879254,
769
+ "loss": 0.2046,
770
  "step": 1220
771
  },
772
  {
773
+ "epoch": 0.18,
774
+ "learning_rate": 0.00023335800671892495,
775
+ "loss": 0.2033,
776
  "step": 1230
777
  },
778
  {
779
+ "epoch": 0.18,
780
+ "learning_rate": 0.00023335231023905738,
781
+ "loss": 0.2217,
782
  "step": 1240
783
  },
784
  {
785
+ "epoch": 0.18,
786
+ "learning_rate": 0.00023334661375918982,
787
+ "loss": 0.332,
788
  "step": 1250
789
  },
790
  {
791
+ "epoch": 0.18,
792
+ "learning_rate": 0.00023334091727932228,
793
+ "loss": 0.2168,
794
  "step": 1260
795
  },
796
  {
797
+ "epoch": 0.19,
798
+ "learning_rate": 0.00023333522079945472,
799
+ "loss": 0.1957,
800
  "step": 1270
801
  },
802
  {
803
+ "epoch": 0.19,
804
+ "learning_rate": 0.00023332952431958713,
805
+ "loss": 0.1912,
806
  "step": 1280
807
  },
808
  {
809
+ "epoch": 0.19,
810
+ "learning_rate": 0.00023332382783971956,
811
+ "loss": 0.2479,
812
  "step": 1290
813
  },
814
  {
815
+ "epoch": 0.19,
816
+ "learning_rate": 0.000233318131359852,
817
+ "loss": 0.3616,
818
  "step": 1300
819
  },
820
  {
821
+ "epoch": 0.19,
822
+ "learning_rate": 0.00023331243487998443,
823
+ "loss": 0.2171,
824
  "step": 1310
825
  },
826
  {
827
+ "epoch": 0.19,
828
+ "learning_rate": 0.00023330673840011687,
829
+ "loss": 0.1931,
830
  "step": 1320
831
  },
832
  {
833
+ "epoch": 0.19,
834
+ "learning_rate": 0.00023330104192024928,
835
+ "loss": 0.2406,
836
  "step": 1330
837
  },
838
  {
839
+ "epoch": 0.2,
840
+ "learning_rate": 0.0002332953454403817,
841
+ "loss": 0.2337,
842
  "step": 1340
843
  },
844
  {
845
+ "epoch": 0.2,
846
+ "learning_rate": 0.00023328964896051415,
847
+ "loss": 0.3704,
848
  "step": 1350
849
  },
850
  {
851
+ "epoch": 0.2,
852
+ "learning_rate": 0.00023328395248064658,
853
+ "loss": 0.2246,
854
  "step": 1360
855
  },
856
  {
857
+ "epoch": 0.2,
858
+ "learning_rate": 0.00023327825600077902,
859
+ "loss": 0.2027,
860
  "step": 1370
861
  },
862
  {
863
+ "epoch": 0.2,
864
+ "learning_rate": 0.00023327255952091143,
865
+ "loss": 0.2162,
866
  "step": 1380
867
  },
868
  {
869
+ "epoch": 0.2,
870
+ "learning_rate": 0.00023326686304104386,
871
+ "loss": 0.238,
872
  "step": 1390
873
  },
874
  {
875
+ "epoch": 0.2,
876
+ "learning_rate": 0.0002332611665611763,
877
+ "loss": 0.3327,
878
  "step": 1400
879
  },
880
  {
881
+ "epoch": 0.21,
882
+ "learning_rate": 0.00023325547008130873,
883
+ "loss": 0.2206,
884
  "step": 1410
885
  },
886
  {
887
+ "epoch": 0.21,
888
+ "learning_rate": 0.00023324977360144117,
889
+ "loss": 0.2087,
890
  "step": 1420
891
  },
892
  {
893
+ "epoch": 0.21,
894
+ "learning_rate": 0.00023324407712157358,
895
+ "loss": 0.2095,
896
  "step": 1430
897
  },
898
  {
899
+ "epoch": 0.21,
900
+ "learning_rate": 0.000233238380641706,
901
+ "loss": 0.2335,
902
  "step": 1440
903
  },
904
  {
905
+ "epoch": 0.21,
906
+ "learning_rate": 0.00023323268416183845,
907
+ "loss": 0.3777,
908
  "step": 1450
909
  },
910
  {
911
+ "epoch": 0.21,
912
+ "learning_rate": 0.00023322698768197088,
913
+ "loss": 0.1873,
914
  "step": 1460
915
  },
916
  {
917
+ "epoch": 0.21,
918
+ "learning_rate": 0.00023322129120210332,
919
+ "loss": 0.1953,
920
  "step": 1470
921
  },
922
  {
923
+ "epoch": 0.22,
924
+ "learning_rate": 0.00023321559472223573,
925
+ "loss": 0.2137,
926
  "step": 1480
927
  },
928
  {
929
+ "epoch": 0.22,
930
+ "learning_rate": 0.00023320989824236816,
931
+ "loss": 0.2647,
932
  "step": 1490
933
  },
934
  {
935
+ "epoch": 0.22,
936
+ "learning_rate": 0.0002332042017625006,
937
+ "loss": 0.3354,
938
  "step": 1500
939
  },
940
  {
941
+ "epoch": 0.22,
942
+ "learning_rate": 0.00023319850528263303,
943
+ "loss": 0.2053,
944
  "step": 1510
945
  },
946
  {
947
+ "epoch": 0.22,
948
+ "learning_rate": 0.00023319280880276547,
949
+ "loss": 0.2,
950
  "step": 1520
951
  },
952
  {
953
+ "epoch": 0.22,
954
+ "learning_rate": 0.00023318711232289788,
955
+ "loss": 0.2223,
956
  "step": 1530
957
  },
958
  {
959
+ "epoch": 0.22,
960
+ "learning_rate": 0.00023318141584303031,
961
+ "loss": 0.2289,
962
  "step": 1540
963
  },
964
  {
965
+ "epoch": 0.23,
966
+ "learning_rate": 0.00023317571936316275,
967
+ "loss": 0.3805,
968
  "step": 1550
969
  },
970
  {
971
+ "epoch": 0.23,
972
+ "learning_rate": 0.00023317002288329519,
973
+ "loss": 0.2144,
974
  "step": 1560
975
  },
976
  {
977
+ "epoch": 0.23,
978
+ "learning_rate": 0.00023316432640342762,
979
+ "loss": 0.1852,
980
  "step": 1570
981
  },
982
  {
983
+ "epoch": 0.23,
984
+ "learning_rate": 0.00023315862992356006,
985
+ "loss": 0.1926,
986
  "step": 1580
987
  },
988
  {
989
+ "epoch": 0.23,
990
+ "learning_rate": 0.0002331529334436925,
991
+ "loss": 0.2233,
992
  "step": 1590
993
  },
994
  {
995
+ "epoch": 0.23,
996
+ "learning_rate": 0.00023314723696382493,
997
+ "loss": 0.3165,
998
  "step": 1600
999
  },
1000
  {
1001
+ "epoch": 0.23,
1002
+ "eval_cer": 0.10628673416007632,
1003
+ "eval_loss": 0.16792798042297363,
1004
+ "eval_runtime": 368.4708,
1005
+ "eval_samples_per_second": 14.197,
1006
+ "eval_steps_per_second": 0.475,
1007
+ "eval_wer": 0.9545020072643854,
1008
  "step": 1600
1009
  },
1010
  {
1011
+ "epoch": 0.24,
1012
+ "learning_rate": 0.00023314154048395736,
1013
+ "loss": 0.2055,
1014
  "step": 1610
1015
  },
1016
  {
1017
+ "epoch": 0.24,
1018
+ "learning_rate": 0.0002331358440040898,
1019
+ "loss": 0.1796,
1020
  "step": 1620
1021
  },
1022
  {
1023
+ "epoch": 0.24,
1024
+ "learning_rate": 0.0002331301475242222,
1025
+ "loss": 0.2126,
1026
  "step": 1630
1027
  },
1028
  {
1029
+ "epoch": 0.24,
1030
+ "learning_rate": 0.00023312445104435464,
1031
+ "loss": 0.2269,
1032
  "step": 1640
1033
  },
1034
  {
1035
+ "epoch": 0.24,
1036
+ "learning_rate": 0.00023311875456448708,
1037
+ "loss": 0.3827,
1038
  "step": 1650
1039
  },
1040
  {
1041
+ "epoch": 0.24,
1042
+ "learning_rate": 0.00023311305808461951,
1043
+ "loss": 0.2356,
1044
  "step": 1660
1045
  },
1046
  {
1047
+ "epoch": 0.24,
1048
+ "learning_rate": 0.00023310736160475195,
1049
+ "loss": 0.2162,
1050
  "step": 1670
1051
  },
1052
  {
1053
+ "epoch": 0.25,
1054
+ "learning_rate": 0.00023310166512488436,
1055
+ "loss": 0.2021,
1056
  "step": 1680
1057
  },
1058
  {
1059
+ "epoch": 0.25,
1060
+ "learning_rate": 0.0002330959686450168,
1061
+ "loss": 0.2629,
1062
  "step": 1690
1063
  },
1064
  {
1065
+ "epoch": 0.25,
1066
+ "learning_rate": 0.00023309027216514923,
1067
+ "loss": 0.3539,
1068
  "step": 1700
1069
  },
1070
  {
1071
+ "epoch": 0.25,
1072
+ "learning_rate": 0.00023308457568528166,
1073
+ "loss": 0.1938,
1074
  "step": 1710
1075
  },
1076
  {
1077
+ "epoch": 0.25,
1078
+ "learning_rate": 0.0002330788792054141,
1079
+ "loss": 0.2117,
1080
  "step": 1720
1081
  },
1082
  {
1083
+ "epoch": 0.25,
1084
+ "learning_rate": 0.0002330731827255465,
1085
+ "loss": 0.2037,
1086
  "step": 1730
1087
  },
1088
  {
1089
+ "epoch": 0.25,
1090
+ "learning_rate": 0.00023306748624567894,
1091
+ "loss": 0.2314,
1092
  "step": 1740
1093
  },
1094
  {
1095
+ "epoch": 0.26,
1096
+ "learning_rate": 0.00023306178976581138,
1097
+ "loss": 0.3471,
1098
  "step": 1750
1099
  },
1100
  {
1101
+ "epoch": 0.26,
1102
+ "learning_rate": 0.00023305609328594382,
1103
+ "loss": 0.206,
1104
  "step": 1760
1105
  },
1106
  {
1107
+ "epoch": 0.26,
1108
+ "learning_rate": 0.00023305039680607625,
1109
+ "loss": 0.1968,
1110
  "step": 1770
1111
  },
1112
  {
1113
+ "epoch": 0.26,
1114
+ "learning_rate": 0.00023304470032620866,
1115
+ "loss": 0.2046,
1116
  "step": 1780
1117
  },
1118
  {
1119
+ "epoch": 0.26,
1120
+ "learning_rate": 0.0002330390038463411,
1121
+ "loss": 0.2072,
1122
  "step": 1790
1123
  },
1124
  {
1125
+ "epoch": 0.26,
1126
+ "learning_rate": 0.00023303330736647353,
1127
+ "loss": 0.3442,
1128
  "step": 1800
1129
  },
1130
  {
1131
+ "epoch": 0.26,
1132
+ "learning_rate": 0.00023302761088660597,
1133
+ "loss": 0.2232,
1134
  "step": 1810
1135
  },
1136
  {
1137
+ "epoch": 0.27,
1138
+ "learning_rate": 0.0002330219144067384,
1139
+ "loss": 0.1899,
1140
  "step": 1820
1141
  },
1142
  {
1143
+ "epoch": 0.27,
1144
+ "learning_rate": 0.0002330162179268708,
1145
+ "loss": 0.2074,
1146
  "step": 1830
1147
  },
1148
  {
1149
+ "epoch": 0.27,
1150
+ "learning_rate": 0.00023301052144700325,
1151
+ "loss": 0.2778,
1152
  "step": 1840
1153
  },
1154
  {
1155
+ "epoch": 0.27,
1156
+ "learning_rate": 0.00023300482496713568,
1157
+ "loss": 0.367,
1158
  "step": 1850
1159
  },
1160
  {
1161
+ "epoch": 0.27,
1162
+ "learning_rate": 0.00023299912848726812,
1163
+ "loss": 0.2327,
1164
  "step": 1860
1165
  },
1166
  {
1167
+ "epoch": 0.27,
1168
+ "learning_rate": 0.00023299343200740055,
1169
+ "loss": 0.1955,
1170
  "step": 1870
1171
  },
1172
  {
1173
+ "epoch": 0.27,
1174
+ "learning_rate": 0.00023298773552753296,
1175
+ "loss": 0.2023,
1176
  "step": 1880
1177
  },
1178
  {
1179
+ "epoch": 0.28,
1180
+ "learning_rate": 0.0002329820390476654,
1181
+ "loss": 0.2603,
1182
  "step": 1890
1183
  },
1184
  {
1185
+ "epoch": 0.28,
1186
+ "learning_rate": 0.00023297634256779786,
1187
+ "loss": 0.3668,
1188
  "step": 1900
1189
  },
1190
  {
1191
+ "epoch": 0.28,
1192
+ "learning_rate": 0.0002329706460879303,
1193
+ "loss": 0.1981,
1194
  "step": 1910
1195
  },
1196
  {
1197
+ "epoch": 0.28,
1198
+ "learning_rate": 0.00023296494960806273,
1199
+ "loss": 0.2515,
1200
  "step": 1920
1201
  },
1202
  {
1203
+ "epoch": 0.28,
1204
+ "learning_rate": 0.00023295925312819514,
1205
+ "loss": 0.1854,
1206
  "step": 1930
1207
  },
1208
  {
1209
+ "epoch": 0.28,
1210
+ "learning_rate": 0.00023295355664832757,
1211
+ "loss": 0.2656,
1212
  "step": 1940
1213
  },
1214
  {
1215
+ "epoch": 0.28,
1216
+ "learning_rate": 0.00023294786016846,
1217
+ "loss": 0.3529,
1218
  "step": 1950
1219
  },
1220
  {
1221
+ "epoch": 0.29,
1222
+ "learning_rate": 0.00023294216368859245,
1223
+ "loss": 0.2059,
1224
  "step": 1960
1225
  },
1226
  {
1227
+ "epoch": 0.29,
1228
+ "learning_rate": 0.00023293646720872488,
1229
+ "loss": 0.1898,
1230
  "step": 1970
1231
  },
1232
  {
1233
+ "epoch": 0.29,
1234
+ "learning_rate": 0.0002329307707288573,
1235
+ "loss": 0.2393,
1236
  "step": 1980
1237
  },
1238
  {
1239
+ "epoch": 0.29,
1240
+ "learning_rate": 0.00023292507424898973,
1241
+ "loss": 0.2417,
1242
  "step": 1990
1243
  },
1244
  {
1245
+ "epoch": 0.29,
1246
+ "learning_rate": 0.00023291937776912216,
1247
+ "loss": 0.3633,
1248
  "step": 2000
1249
  },
1250
  {
1251
+ "epoch": 0.29,
1252
+ "eval_cer": 0.09183784796918316,
1253
+ "eval_loss": 0.1622667759656906,
1254
+ "eval_runtime": 378.6594,
1255
+ "eval_samples_per_second": 13.815,
1256
+ "eval_steps_per_second": 0.462,
1257
+ "eval_wer": 0.9124450391894475,
1258
  "step": 2000
1259
  },
1260
  {
1261
+ "epoch": 0.29,
1262
+ "learning_rate": 0.0002329136812892546,
1263
+ "loss": 0.2131,
1264
  "step": 2010
1265
  },
1266
  {
1267
+ "epoch": 0.29,
1268
+ "learning_rate": 0.00023290798480938703,
1269
+ "loss": 0.2096,
1270
  "step": 2020
1271
  },
1272
  {
1273
+ "epoch": 0.3,
1274
+ "learning_rate": 0.00023290228832951944,
1275
+ "loss": 0.2113,
1276
  "step": 2030
1277
  },
1278
  {
1279
+ "epoch": 0.3,
1280
+ "learning_rate": 0.00023289659184965188,
1281
+ "loss": 0.2428,
1282
  "step": 2040
1283
  },
1284
  {
1285
+ "epoch": 0.3,
1286
+ "learning_rate": 0.0002328908953697843,
1287
+ "loss": 0.364,
1288
  "step": 2050
1289
  },
1290
  {
1291
+ "epoch": 0.3,
1292
+ "learning_rate": 0.00023288519888991675,
1293
+ "loss": 0.1823,
1294
  "step": 2060
1295
  },
1296
  {
1297
+ "epoch": 0.3,
1298
+ "learning_rate": 0.00023287950241004918,
1299
+ "loss": 0.1999,
1300
  "step": 2070
1301
  },
1302
  {
1303
+ "epoch": 0.3,
1304
+ "learning_rate": 0.0002328738059301816,
1305
+ "loss": 0.1921,
1306
  "step": 2080
1307
  },
1308
  {
1309
+ "epoch": 0.31,
1310
+ "learning_rate": 0.00023286810945031403,
1311
+ "loss": 0.2185,
1312
  "step": 2090
1313
  },
1314
  {
1315
+ "epoch": 0.31,
1316
+ "learning_rate": 0.00023286241297044646,
1317
+ "loss": 0.3339,
1318
  "step": 2100
1319
  },
1320
  {
1321
+ "epoch": 0.31,
1322
+ "learning_rate": 0.0002328567164905789,
1323
+ "loss": 0.1856,
1324
  "step": 2110
1325
  },
1326
  {
1327
+ "epoch": 0.31,
1328
+ "learning_rate": 0.00023285102001071133,
1329
+ "loss": 0.1971,
1330
  "step": 2120
1331
  },
1332
  {
1333
+ "epoch": 0.31,
1334
+ "learning_rate": 0.00023284532353084374,
1335
+ "loss": 0.1749,
1336
  "step": 2130
1337
  },
1338
  {
1339
+ "epoch": 0.31,
1340
+ "learning_rate": 0.00023283962705097618,
1341
+ "loss": 0.202,
1342
  "step": 2140
1343
  },
1344
  {
1345
+ "epoch": 0.31,
1346
+ "learning_rate": 0.0002328339305711086,
1347
+ "loss": 0.3775,
1348
  "step": 2150
1349
  },
1350
  {
1351
+ "epoch": 0.32,
1352
+ "learning_rate": 0.00023282823409124105,
1353
+ "loss": 0.2029,
1354
  "step": 2160
1355
  },
1356
  {
1357
+ "epoch": 0.32,
1358
+ "learning_rate": 0.00023282253761137348,
1359
+ "loss": 0.2078,
1360
  "step": 2170
1361
  },
1362
  {
1363
+ "epoch": 0.32,
1364
+ "learning_rate": 0.0002328168411315059,
1365
+ "loss": 0.2169,
1366
  "step": 2180
1367
  },
1368
  {
1369
+ "epoch": 0.32,
1370
+ "learning_rate": 0.00023281114465163833,
1371
+ "loss": 0.2519,
1372
  "step": 2190
1373
  },
1374
  {
1375
+ "epoch": 0.32,
1376
+ "learning_rate": 0.00023280544817177076,
1377
+ "loss": 0.3249,
1378
  "step": 2200
1379
  },
1380
  {
1381
+ "epoch": 0.32,
1382
+ "learning_rate": 0.0002327997516919032,
1383
+ "loss": 0.1929,
1384
  "step": 2210
1385
  },
1386
  {
1387
+ "epoch": 0.32,
1388
+ "learning_rate": 0.00023279405521203563,
1389
+ "loss": 0.192,
1390
  "step": 2220
1391
  },
1392
  {
1393
+ "epoch": 0.33,
1394
+ "learning_rate": 0.00023278835873216807,
1395
+ "loss": 0.1953,
1396
  "step": 2230
1397
  },
1398
  {
1399
+ "epoch": 0.33,
1400
+ "learning_rate": 0.0002327826622523005,
1401
+ "loss": 0.2662,
1402
  "step": 2240
1403
  },
1404
  {
1405
+ "epoch": 0.33,
1406
+ "learning_rate": 0.00023277696577243294,
1407
+ "loss": 0.3353,
1408
  "step": 2250
1409
  },
1410
  {
1411
+ "epoch": 0.33,
1412
+ "learning_rate": 0.00023277126929256538,
1413
+ "loss": 0.2348,
1414
  "step": 2260
1415
  },
1416
  {
1417
+ "epoch": 0.33,
1418
+ "learning_rate": 0.0002327655728126978,
1419
+ "loss": 0.207,
1420
  "step": 2270
1421
  },
1422
  {
1423
+ "epoch": 0.33,
1424
+ "learning_rate": 0.00023275987633283022,
1425
+ "loss": 0.221,
1426
  "step": 2280
1427
  },
1428
  {
1429
+ "epoch": 0.33,
1430
+ "learning_rate": 0.00023275417985296266,
1431
+ "loss": 0.2328,
1432
  "step": 2290
1433
  },
1434
  {
1435
+ "epoch": 0.34,
1436
+ "learning_rate": 0.0002327484833730951,
1437
+ "loss": 0.3737,
1438
  "step": 2300
1439
  },
1440
  {
1441
+ "epoch": 0.34,
1442
+ "learning_rate": 0.00023274278689322753,
1443
+ "loss": 0.2214,
1444
  "step": 2310
1445
  },
1446
  {
1447
+ "epoch": 0.34,
1448
+ "learning_rate": 0.00023273709041335996,
1449
+ "loss": 0.2035,
1450
  "step": 2320
1451
  },
1452
  {
1453
+ "epoch": 0.34,
1454
+ "learning_rate": 0.00023273139393349237,
1455
+ "loss": 0.2373,
1456
  "step": 2330
1457
  },
1458
  {
1459
+ "epoch": 0.34,
1460
+ "learning_rate": 0.0002327256974536248,
1461
+ "loss": 0.2607,
1462
  "step": 2340
1463
  },
1464
  {
1465
+ "epoch": 0.34,
1466
+ "learning_rate": 0.00023272000097375724,
1467
+ "loss": 0.3317,
1468
  "step": 2350
1469
  },
1470
  {
1471
+ "epoch": 0.34,
1472
+ "learning_rate": 0.00023271430449388968,
1473
+ "loss": 0.1983,
1474
  "step": 2360
1475
  },
1476
  {
1477
+ "epoch": 0.35,
1478
+ "learning_rate": 0.00023270860801402211,
1479
+ "loss": 0.2118,
1480
  "step": 2370
1481
  },
1482
  {
1483
+ "epoch": 0.35,
1484
+ "learning_rate": 0.00023270291153415452,
1485
+ "loss": 0.2143,
1486
  "step": 2380
1487
  },
1488
  {
1489
+ "epoch": 0.35,
1490
+ "learning_rate": 0.00023269721505428696,
1491
+ "loss": 0.2277,
1492
  "step": 2390
1493
  },
1494
  {
1495
+ "epoch": 0.35,
1496
+ "learning_rate": 0.0002326915185744194,
1497
+ "loss": 0.3574,
1498
  "step": 2400
1499
  },
1500
  {
1501
+ "epoch": 0.35,
1502
+ "eval_cer": 0.0956312051182347,
1503
+ "eval_loss": 0.15721334517002106,
1504
+ "eval_runtime": 363.9845,
1505
+ "eval_samples_per_second": 14.371,
1506
+ "eval_steps_per_second": 0.481,
1507
+ "eval_wer": 0.9361498757407761,
1508
  "step": 2400
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1509
  }
1510
  ],
1511
+ "max_steps": 410880,
1512
  "num_train_epochs": 60,
1513
+ "total_flos": 2.442475321551867e+19,
1514
  "trial_name": null,
1515
  "trial_params": null
1516
  }
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f56f31de078a68ec0236b863c9f14d7668d3a1377ed31d3ae07ae71d6b4e959d
3
  size 3119
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0669da01203e40a96dd70c7f4d165950d4d8b01b55ce70f75ac298893a8254e9
3
  size 3119