marcel commited on
Commit
b5b833b
1 Parent(s): 32d34ee

25-25 cp 1200

Browse files
Files changed (7) hide show
  1. README.md +2 -2
  2. config.json +1 -1
  3. optimizer.pt +1 -1
  4. pytorch_model.bin +1 -1
  5. scheduler.pt +1 -1
  6. trainer_state.json +29 -1009
  7. training_args.bin +1 -1
README.md CHANGED
@@ -22,7 +22,7 @@ model-index:
22
  metrics:
23
  - name: Test WER
24
  type: wer
25
- value: 15.93
26
  ---
27
 
28
  # Wav2Vec2-Large-XLSR-53-German
@@ -140,7 +140,7 @@ result = test_dataset.map(evaluate, batched=True, batch_size=8)
140
  print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
141
  ```
142
 
143
- **Test Result**: 15.93 %
144
 
145
 
146
  ## Training
 
22
  metrics:
23
  - name: Test WER
24
  type: wer
25
+ value: 15.91
26
  ---
27
 
28
  # Wav2Vec2-Large-XLSR-53-German
 
140
  print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
141
  ```
142
 
143
+ **Test Result**: 15.91 %
144
 
145
 
146
  ## Training
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
3
  "activation_dropout": 0.0,
4
  "apply_spec_augment": true,
5
  "architectures": [
 
1
  {
2
+ "_name_or_path": "marcel/wav2vec2-large-xlsr-53-german",
3
  "activation_dropout": 0.0,
4
  "apply_spec_augment": true,
5
  "architectures": [
optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:49d87731bb26150050af1cdbd0e26840a767ab743ddc158b8c1597d81ce8a17e
3
  size 2490339591
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60c27fab01d40c3d3cf0a6311d0cf2cec0c7d7e44291f1c479b9ab0ecf1fac33
3
  size 2490339591
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c474421f8572350844e0e3d8950065489ef6f02008712116a41bd32948a4c91d
3
  size 1262065047
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:daf2875bd490c83c5a9c9e6ab12850d7591a8372f6e91189f2c87121e41d0555
3
  size 1262065047
scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2de514341497aeb2d969571853d3ddb165789f77a56da57480f7ae711d12e6df
3
  size 623
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bfdc2d2bf2e38aa9b79f44cff2f9aeb6010436b1e91d3f4e3dd4369fe0145158
3
  size 623
trainer_state.json CHANGED
@@ -1,1038 +1,58 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 29.70498474059003,
5
- "global_step": 29200,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
- "epoch": 0.41,
12
- "learning_rate": 0.00023999999999999998,
13
- "loss": 4.7996,
14
  "step": 400
15
  },
16
  {
17
- "epoch": 0.41,
18
- "eval_loss": 3.018367290496826,
19
- "eval_runtime": 230.2288,
20
- "eval_samples_per_second": 8.127,
21
- "eval_wer": 1.0,
22
  "step": 400
23
  },
24
  {
25
- "epoch": 0.81,
26
- "learning_rate": 0.0002968954812004139,
27
- "loss": 2.0295,
28
  "step": 800
29
  },
30
  {
31
- "epoch": 0.81,
32
- "eval_loss": 0.71845942735672,
33
- "eval_runtime": 234.0394,
34
- "eval_samples_per_second": 7.994,
35
- "eval_wer": 0.5917076167076167,
36
  "step": 800
37
  },
38
  {
39
- "epoch": 1.22,
40
- "learning_rate": 0.0002927561228009658,
41
- "loss": 0.4632,
42
  "step": 1200
43
  },
44
  {
45
- "epoch": 1.22,
46
- "eval_loss": 0.4624195396900177,
47
- "eval_runtime": 419.7523,
48
- "eval_samples_per_second": 4.457,
49
- "eval_wer": 0.4606879606879607,
50
  "step": 1200
51
- },
52
- {
53
- "epoch": 1.63,
54
- "learning_rate": 0.0002886167644015177,
55
- "loss": 0.3595,
56
- "step": 1600
57
- },
58
- {
59
- "epoch": 1.63,
60
- "eval_loss": 0.4238700568675995,
61
- "eval_runtime": 235.4681,
62
- "eval_samples_per_second": 7.946,
63
- "eval_wer": 0.4348894348894349,
64
- "step": 1600
65
- },
66
- {
67
- "epoch": 2.03,
68
- "learning_rate": 0.0002844774060020697,
69
- "loss": 0.3181,
70
- "step": 2000
71
- },
72
- {
73
- "epoch": 2.03,
74
- "eval_loss": 0.4156078100204468,
75
- "eval_runtime": 234.1423,
76
- "eval_samples_per_second": 7.991,
77
- "eval_wer": 0.42106879606879605,
78
- "step": 2000
79
- },
80
- {
81
- "epoch": 2.44,
82
- "learning_rate": 0.0002803380476026216,
83
- "loss": 0.2606,
84
- "step": 2400
85
- },
86
- {
87
- "epoch": 2.44,
88
- "eval_loss": 0.39613404870033264,
89
- "eval_runtime": 233.8224,
90
- "eval_samples_per_second": 8.002,
91
- "eval_wer": 0.4057125307125307,
92
- "step": 2400
93
- },
94
- {
95
- "epoch": 2.85,
96
- "learning_rate": 0.0002761986892031735,
97
- "loss": 0.2475,
98
- "step": 2800
99
- },
100
- {
101
- "epoch": 2.85,
102
- "eval_loss": 0.3680589497089386,
103
- "eval_runtime": 235.255,
104
- "eval_samples_per_second": 7.953,
105
- "eval_wer": 0.39256756756756755,
106
- "step": 2800
107
- },
108
- {
109
- "epoch": 3.26,
110
- "learning_rate": 0.0002720593308037254,
111
- "loss": 0.2208,
112
- "step": 3200
113
- },
114
- {
115
- "epoch": 3.26,
116
- "eval_loss": 0.40793511271476746,
117
- "eval_runtime": 233.2565,
118
- "eval_samples_per_second": 8.021,
119
- "eval_wer": 0.3984029484029484,
120
- "step": 3200
121
- },
122
- {
123
- "epoch": 3.66,
124
- "learning_rate": 0.0002679199724042773,
125
- "loss": 0.2168,
126
- "step": 3600
127
- },
128
- {
129
- "epoch": 3.66,
130
- "eval_loss": 0.40217384696006775,
131
- "eval_runtime": 233.5419,
132
- "eval_samples_per_second": 8.011,
133
- "eval_wer": 0.40958230958230957,
134
- "step": 3600
135
- },
136
- {
137
- "epoch": 4.07,
138
- "learning_rate": 0.0002637806140048292,
139
- "loss": 0.2059,
140
- "step": 4000
141
- },
142
- {
143
- "epoch": 4.07,
144
- "eval_loss": 0.42218777537345886,
145
- "eval_runtime": 233.047,
146
- "eval_samples_per_second": 8.028,
147
- "eval_wer": 0.40128992628992627,
148
- "step": 4000
149
- },
150
- {
151
- "epoch": 4.48,
152
- "learning_rate": 0.0002596412556053811,
153
- "loss": 0.1808,
154
- "step": 4400
155
- },
156
- {
157
- "epoch": 4.48,
158
- "eval_loss": 0.43475455045700073,
159
- "eval_runtime": 235.2511,
160
- "eval_samples_per_second": 7.953,
161
- "eval_wer": 0.39213759213759214,
162
- "step": 4400
163
- },
164
- {
165
- "epoch": 4.88,
166
- "learning_rate": 0.0002555018972059331,
167
- "loss": 0.1874,
168
- "step": 4800
169
- },
170
- {
171
- "epoch": 4.88,
172
- "eval_loss": 0.40788909792900085,
173
- "eval_runtime": 237.3453,
174
- "eval_samples_per_second": 7.883,
175
- "eval_wer": 0.3885749385749386,
176
- "step": 4800
177
- },
178
- {
179
- "epoch": 5.29,
180
- "learning_rate": 0.000251362538806485,
181
- "loss": 0.1671,
182
- "step": 5200
183
- },
184
- {
185
- "epoch": 5.29,
186
- "eval_loss": 0.42425239086151123,
187
- "eval_runtime": 235.3303,
188
- "eval_samples_per_second": 7.951,
189
- "eval_wer": 0.3864864864864865,
190
- "step": 5200
191
- },
192
- {
193
- "epoch": 5.7,
194
- "learning_rate": 0.0002472231804070369,
195
- "loss": 0.168,
196
- "step": 5600
197
- },
198
- {
199
- "epoch": 5.7,
200
- "eval_loss": 0.39308613538742065,
201
- "eval_runtime": 234.7832,
202
- "eval_samples_per_second": 7.969,
203
- "eval_wer": 0.38114250614250617,
204
- "step": 5600
205
- },
206
- {
207
- "epoch": 6.1,
208
- "learning_rate": 0.0002430838220075888,
209
- "loss": 0.1603,
210
- "step": 6000
211
- },
212
- {
213
- "epoch": 6.1,
214
- "eval_loss": 0.4091956615447998,
215
- "eval_runtime": 234.4154,
216
- "eval_samples_per_second": 7.982,
217
- "eval_wer": 0.38175675675675674,
218
- "step": 6000
219
- },
220
- {
221
- "epoch": 6.51,
222
- "learning_rate": 0.00023894446360814072,
223
- "loss": 0.1458,
224
- "step": 6400
225
- },
226
- {
227
- "epoch": 6.51,
228
- "eval_loss": 0.4052203893661499,
229
- "eval_runtime": 236.4087,
230
- "eval_samples_per_second": 7.914,
231
- "eval_wer": 0.3844594594594595,
232
- "step": 6400
233
- },
234
- {
235
- "epoch": 6.92,
236
- "learning_rate": 0.00023480510520869263,
237
- "loss": 0.1466,
238
- "step": 6800
239
- },
240
- {
241
- "epoch": 6.92,
242
- "eval_loss": 0.4150303900241852,
243
- "eval_runtime": 234.9457,
244
- "eval_samples_per_second": 7.964,
245
- "eval_wer": 0.3832923832923833,
246
- "step": 6800
247
- },
248
- {
249
- "epoch": 7.32,
250
- "learning_rate": 0.00023066574680924453,
251
- "loss": 0.1347,
252
- "step": 7200
253
- },
254
- {
255
- "epoch": 7.32,
256
- "eval_loss": 0.40190571546554565,
257
- "eval_runtime": 232.7808,
258
- "eval_samples_per_second": 8.038,
259
- "eval_wer": 0.3687346437346437,
260
- "step": 7200
261
- },
262
- {
263
- "epoch": 7.73,
264
- "learning_rate": 0.00022652638840979646,
265
- "loss": 0.1358,
266
- "step": 7600
267
- },
268
- {
269
- "epoch": 7.73,
270
- "eval_loss": 0.40535062551498413,
271
- "eval_runtime": 234.3447,
272
- "eval_samples_per_second": 7.984,
273
- "eval_wer": 0.3745085995085995,
274
- "step": 7600
275
- },
276
- {
277
- "epoch": 8.14,
278
- "learning_rate": 0.0002223870300103484,
279
- "loss": 0.1334,
280
- "step": 8000
281
- },
282
- {
283
- "epoch": 8.14,
284
- "eval_loss": 0.3982257843017578,
285
- "eval_runtime": 234.9167,
286
- "eval_samples_per_second": 7.965,
287
- "eval_wer": 0.371990171990172,
288
- "step": 8000
289
- },
290
- {
291
- "epoch": 8.55,
292
- "learning_rate": 0.0002182476716109003,
293
- "loss": 0.1228,
294
- "step": 8400
295
- },
296
- {
297
- "epoch": 8.55,
298
- "eval_loss": 0.38730743527412415,
299
- "eval_runtime": 236.5628,
300
- "eval_samples_per_second": 7.909,
301
- "eval_wer": 0.3726044226044226,
302
- "step": 8400
303
- },
304
- {
305
- "epoch": 8.95,
306
- "learning_rate": 0.0002141083132114522,
307
- "loss": 0.1226,
308
- "step": 8800
309
- },
310
- {
311
- "epoch": 8.95,
312
- "eval_loss": 0.3983035087585449,
313
- "eval_runtime": 236.8485,
314
- "eval_samples_per_second": 7.9,
315
- "eval_wer": 0.36437346437346435,
316
- "step": 8800
317
- },
318
- {
319
- "epoch": 9.36,
320
- "learning_rate": 0.0002099689548120041,
321
- "loss": 0.1091,
322
- "step": 9200
323
- },
324
- {
325
- "epoch": 9.36,
326
- "eval_loss": 0.4045446813106537,
327
- "eval_runtime": 234.4792,
328
- "eval_samples_per_second": 7.979,
329
- "eval_wer": 0.36406633906633906,
330
- "step": 9200
331
- },
332
- {
333
- "epoch": 9.77,
334
- "learning_rate": 0.00020582959641255602,
335
- "loss": 0.1124,
336
- "step": 9600
337
- },
338
- {
339
- "epoch": 9.77,
340
- "eval_loss": 0.41637349128723145,
341
- "eval_runtime": 234.3486,
342
- "eval_samples_per_second": 7.984,
343
- "eval_wer": 0.36812039312039313,
344
- "step": 9600
345
- },
346
- {
347
- "epoch": 10.17,
348
- "learning_rate": 0.00020169023801310795,
349
- "loss": 0.1117,
350
- "step": 10000
351
- },
352
- {
353
- "epoch": 10.17,
354
- "eval_loss": 0.42693623900413513,
355
- "eval_runtime": 235.3,
356
- "eval_samples_per_second": 7.952,
357
- "eval_wer": 0.36732186732186733,
358
- "step": 10000
359
- },
360
- {
361
- "epoch": 10.58,
362
- "learning_rate": 0.00019755087961365985,
363
- "loss": 0.1015,
364
- "step": 10400
365
- },
366
- {
367
- "epoch": 10.58,
368
- "eval_loss": 0.45069122314453125,
369
- "eval_runtime": 236.7508,
370
- "eval_samples_per_second": 7.903,
371
- "eval_wer": 0.36357493857493856,
372
- "step": 10400
373
- },
374
- {
375
- "epoch": 10.99,
376
- "learning_rate": 0.0001934115212142118,
377
- "loss": 0.1062,
378
- "step": 10800
379
- },
380
- {
381
- "epoch": 10.99,
382
- "eval_loss": 0.4379144608974457,
383
- "eval_runtime": 237.8786,
384
- "eval_samples_per_second": 7.865,
385
- "eval_wer": 0.3648034398034398,
386
- "step": 10800
387
- },
388
- {
389
- "epoch": 11.39,
390
- "learning_rate": 0.00018927216281476372,
391
- "loss": 0.0945,
392
- "step": 11200
393
- },
394
- {
395
- "epoch": 11.39,
396
- "eval_loss": 0.4285587966442108,
397
- "eval_runtime": 234.2891,
398
- "eval_samples_per_second": 7.986,
399
- "eval_wer": 0.36437346437346435,
400
- "step": 11200
401
- },
402
- {
403
- "epoch": 11.8,
404
- "learning_rate": 0.00018513280441531562,
405
- "loss": 0.0996,
406
- "step": 11600
407
- },
408
- {
409
- "epoch": 11.8,
410
- "eval_loss": 0.4398898780345917,
411
- "eval_runtime": 235.5528,
412
- "eval_samples_per_second": 7.943,
413
- "eval_wer": 0.367014742014742,
414
- "step": 11600
415
- },
416
- {
417
- "epoch": 12.21,
418
- "learning_rate": 0.00018099344601586753,
419
- "loss": 0.0911,
420
- "step": 12000
421
- },
422
- {
423
- "epoch": 12.21,
424
- "eval_loss": 0.4221023619174957,
425
- "eval_runtime": 234.4385,
426
- "eval_samples_per_second": 7.981,
427
- "eval_wer": 0.3568181818181818,
428
- "step": 12000
429
- },
430
- {
431
- "epoch": 12.61,
432
- "learning_rate": 0.00017685408761641943,
433
- "loss": 0.09,
434
- "step": 12400
435
- },
436
- {
437
- "epoch": 12.61,
438
- "eval_loss": 0.4387373924255371,
439
- "eval_runtime": 238.8017,
440
- "eval_samples_per_second": 7.835,
441
- "eval_wer": 0.3641891891891892,
442
- "step": 12400
443
- },
444
- {
445
- "epoch": 13.02,
446
- "learning_rate": 0.00017271472921697134,
447
- "loss": 0.0894,
448
- "step": 12800
449
- },
450
- {
451
- "epoch": 13.02,
452
- "eval_loss": 0.4392118752002716,
453
- "eval_runtime": 235.2986,
454
- "eval_samples_per_second": 7.952,
455
- "eval_wer": 0.3574938574938575,
456
- "step": 12800
457
- },
458
- {
459
- "epoch": 13.43,
460
- "learning_rate": 0.00016857537081752327,
461
- "loss": 0.0816,
462
- "step": 13200
463
- },
464
- {
465
- "epoch": 13.43,
466
- "eval_loss": 0.44565021991729736,
467
- "eval_runtime": 235.1666,
468
- "eval_samples_per_second": 7.956,
469
- "eval_wer": 0.3616707616707617,
470
- "step": 13200
471
- },
472
- {
473
- "epoch": 13.84,
474
- "learning_rate": 0.00016443601241807518,
475
- "loss": 0.0813,
476
- "step": 13600
477
- },
478
- {
479
- "epoch": 13.84,
480
- "eval_loss": 0.4428552985191345,
481
- "eval_runtime": 236.4759,
482
- "eval_samples_per_second": 7.912,
483
- "eval_wer": 0.3593980343980344,
484
- "step": 13600
485
- },
486
- {
487
- "epoch": 14.24,
488
- "learning_rate": 0.0001602966540186271,
489
- "loss": 0.078,
490
- "step": 14000
491
- },
492
- {
493
- "epoch": 14.24,
494
- "eval_loss": 0.46921756863594055,
495
- "eval_runtime": 236.081,
496
- "eval_samples_per_second": 7.925,
497
- "eval_wer": 0.35909090909090907,
498
- "step": 14000
499
- },
500
- {
501
- "epoch": 14.65,
502
- "learning_rate": 0.00015615729561917901,
503
- "loss": 0.0777,
504
- "step": 14400
505
- },
506
- {
507
- "epoch": 14.65,
508
- "eval_loss": 0.45899245142936707,
509
- "eval_runtime": 236.6098,
510
- "eval_samples_per_second": 7.908,
511
- "eval_wer": 0.35448402948402946,
512
- "step": 14400
513
- },
514
- {
515
- "epoch": 15.06,
516
- "learning_rate": 0.00015201793721973095,
517
- "loss": 0.076,
518
- "step": 14800
519
- },
520
- {
521
- "epoch": 15.06,
522
- "eval_loss": 0.48056069016456604,
523
- "eval_runtime": 234.5593,
524
- "eval_samples_per_second": 7.977,
525
- "eval_wer": 0.3517199017199017,
526
- "step": 14800
527
- },
528
- {
529
- "epoch": 15.46,
530
- "learning_rate": 0.00014787857882028285,
531
- "loss": 0.0717,
532
- "step": 15200
533
- },
534
- {
535
- "epoch": 15.46,
536
- "eval_loss": 0.49670103192329407,
537
- "eval_runtime": 236.0843,
538
- "eval_samples_per_second": 7.925,
539
- "eval_wer": 0.3600737100737101,
540
- "step": 15200
541
- },
542
- {
543
- "epoch": 15.87,
544
- "learning_rate": 0.00014373922042083476,
545
- "loss": 0.0708,
546
- "step": 15600
547
- },
548
- {
549
- "epoch": 15.87,
550
- "eval_loss": 0.46058785915374756,
551
- "eval_runtime": 242.6156,
552
- "eval_samples_per_second": 7.712,
553
- "eval_wer": 0.35982800982800983,
554
- "step": 15600
555
- },
556
- {
557
- "epoch": 16.28,
558
- "learning_rate": 0.00013959986202138666,
559
- "loss": 0.0673,
560
- "step": 16000
561
- },
562
- {
563
- "epoch": 16.28,
564
- "eval_loss": 0.45084264874458313,
565
- "eval_runtime": 234.0195,
566
- "eval_samples_per_second": 7.995,
567
- "eval_wer": 0.3546683046683047,
568
- "step": 16000
569
- },
570
- {
571
- "epoch": 16.68,
572
- "learning_rate": 0.0001354605036219386,
573
- "loss": 0.0664,
574
- "step": 16400
575
- },
576
- {
577
- "epoch": 16.68,
578
- "eval_loss": 0.4838450849056244,
579
- "eval_runtime": 237.8079,
580
- "eval_samples_per_second": 7.868,
581
- "eval_wer": 0.35614250614250614,
582
- "step": 16400
583
- },
584
- {
585
- "epoch": 17.09,
586
- "learning_rate": 0.0001313211452224905,
587
- "loss": 0.0639,
588
- "step": 16800
589
- },
590
- {
591
- "epoch": 17.09,
592
- "eval_loss": 0.4703587591648102,
593
- "eval_runtime": 235.9394,
594
- "eval_samples_per_second": 7.93,
595
- "eval_wer": 0.35657248157248156,
596
- "step": 16800
597
- },
598
- {
599
- "epoch": 17.5,
600
- "learning_rate": 0.0001271817868230424,
601
- "loss": 0.0597,
602
- "step": 17200
603
- },
604
- {
605
- "epoch": 17.5,
606
- "eval_loss": 0.47286155819892883,
607
- "eval_runtime": 237.4364,
608
- "eval_samples_per_second": 7.88,
609
- "eval_wer": 0.35657248157248156,
610
- "step": 17200
611
- },
612
- {
613
- "epoch": 17.9,
614
- "learning_rate": 0.00012304242842359434,
615
- "loss": 0.068,
616
- "step": 17600
617
- },
618
- {
619
- "epoch": 17.9,
620
- "eval_loss": 0.42778506875038147,
621
- "eval_runtime": 235.8032,
622
- "eval_samples_per_second": 7.935,
623
- "eval_wer": 0.3484029484029484,
624
- "step": 17600
625
- },
626
- {
627
- "epoch": 18.31,
628
- "learning_rate": 0.00011890307002414625,
629
- "loss": 0.0582,
630
- "step": 18000
631
- },
632
- {
633
- "epoch": 18.31,
634
- "eval_loss": 0.47386595606803894,
635
- "eval_runtime": 238.3257,
636
- "eval_samples_per_second": 7.851,
637
- "eval_wer": 0.34864864864864864,
638
- "step": 18000
639
- },
640
- {
641
- "epoch": 18.72,
642
- "learning_rate": 0.00011476371162469816,
643
- "loss": 0.0593,
644
- "step": 18400
645
- },
646
- {
647
- "epoch": 18.72,
648
- "eval_loss": 0.46009212732315063,
649
- "eval_runtime": 238.7244,
650
- "eval_samples_per_second": 7.837,
651
- "eval_wer": 0.3468673218673219,
652
- "step": 18400
653
- },
654
- {
655
- "epoch": 19.13,
656
- "learning_rate": 0.00011062435322525008,
657
- "loss": 0.0554,
658
- "step": 18800
659
- },
660
- {
661
- "epoch": 19.13,
662
- "eval_loss": 0.506970226764679,
663
- "eval_runtime": 236.9238,
664
- "eval_samples_per_second": 7.897,
665
- "eval_wer": 0.34594594594594597,
666
- "step": 18800
667
- },
668
- {
669
- "epoch": 19.53,
670
- "learning_rate": 0.00010648499482580198,
671
- "loss": 0.0567,
672
- "step": 19200
673
- },
674
- {
675
- "epoch": 19.53,
676
- "eval_loss": 0.47986453771591187,
677
- "eval_runtime": 237.1448,
678
- "eval_samples_per_second": 7.89,
679
- "eval_wer": 0.3435503685503685,
680
- "step": 19200
681
- },
682
- {
683
- "epoch": 19.94,
684
- "learning_rate": 0.00010234563642635392,
685
- "loss": 0.0554,
686
- "step": 19600
687
- },
688
- {
689
- "epoch": 19.94,
690
- "eval_loss": 0.47748732566833496,
691
- "eval_runtime": 236.6967,
692
- "eval_samples_per_second": 7.905,
693
- "eval_wer": 0.34877149877149877,
694
- "step": 19600
695
- },
696
- {
697
- "epoch": 20.35,
698
- "learning_rate": 9.820627802690582e-05,
699
- "loss": 0.0532,
700
- "step": 20000
701
- },
702
- {
703
- "epoch": 20.35,
704
- "eval_loss": 0.5038613677024841,
705
- "eval_runtime": 237.1224,
706
- "eval_samples_per_second": 7.89,
707
- "eval_wer": 0.34342751842751845,
708
- "step": 20000
709
- },
710
- {
711
- "epoch": 20.75,
712
- "learning_rate": 9.406691962745773e-05,
713
- "loss": 0.0503,
714
- "step": 20400
715
- },
716
- {
717
- "epoch": 20.75,
718
- "eval_loss": 0.48931312561035156,
719
- "eval_runtime": 236.3616,
720
- "eval_samples_per_second": 7.916,
721
- "eval_wer": 0.3406633906633907,
722
- "step": 20400
723
- },
724
- {
725
- "epoch": 21.16,
726
- "learning_rate": 8.992756122800964e-05,
727
- "loss": 0.049,
728
- "step": 20800
729
- },
730
- {
731
- "epoch": 21.16,
732
- "eval_loss": 0.484068900346756,
733
- "eval_runtime": 237.5327,
734
- "eval_samples_per_second": 7.877,
735
- "eval_wer": 0.3410933660933661,
736
- "step": 20800
737
- },
738
- {
739
- "epoch": 21.57,
740
- "learning_rate": 8.578820282856158e-05,
741
- "loss": 0.048,
742
- "step": 21200
743
- },
744
- {
745
- "epoch": 21.57,
746
- "eval_loss": 0.4956786334514618,
747
- "eval_runtime": 240.2049,
748
- "eval_samples_per_second": 7.789,
749
- "eval_wer": 0.3418304668304668,
750
- "step": 21200
751
- },
752
- {
753
- "epoch": 21.97,
754
- "learning_rate": 8.164884442911348e-05,
755
- "loss": 0.0453,
756
- "step": 21600
757
- },
758
- {
759
- "epoch": 21.97,
760
- "eval_loss": 0.5122228860855103,
761
- "eval_runtime": 237.8707,
762
- "eval_samples_per_second": 7.866,
763
- "eval_wer": 0.3369164619164619,
764
- "step": 21600
765
- },
766
- {
767
- "epoch": 22.38,
768
- "learning_rate": 7.750948602966539e-05,
769
- "loss": 0.0466,
770
- "step": 22000
771
- },
772
- {
773
- "epoch": 22.38,
774
- "eval_loss": 0.5171410441398621,
775
- "eval_runtime": 254.3593,
776
- "eval_samples_per_second": 7.356,
777
- "eval_wer": 0.34121621621621623,
778
- "step": 22000
779
- },
780
- {
781
- "epoch": 22.79,
782
- "learning_rate": 7.33701276302173e-05,
783
- "loss": 0.0456,
784
- "step": 22400
785
- },
786
- {
787
- "epoch": 22.79,
788
- "eval_loss": 0.5072354674339294,
789
- "eval_runtime": 240.0638,
790
- "eval_samples_per_second": 7.794,
791
- "eval_wer": 0.33507371007371006,
792
- "step": 22400
793
- },
794
- {
795
- "epoch": 23.19,
796
- "learning_rate": 6.923076923076922e-05,
797
- "loss": 0.0459,
798
- "step": 22800
799
- },
800
- {
801
- "epoch": 23.19,
802
- "eval_loss": 0.521458625793457,
803
- "eval_runtime": 457.6179,
804
- "eval_samples_per_second": 4.089,
805
- "eval_wer": 0.34170761670761673,
806
- "step": 22800
807
- },
808
- {
809
- "epoch": 23.6,
810
- "learning_rate": 6.509141083132114e-05,
811
- "loss": 0.0404,
812
- "step": 23200
813
- },
814
- {
815
- "epoch": 23.6,
816
- "eval_loss": 0.5035232901573181,
817
- "eval_runtime": 991.4067,
818
- "eval_samples_per_second": 1.887,
819
- "eval_wer": 0.33746928746928745,
820
- "step": 23200
821
- },
822
- {
823
- "epoch": 24.01,
824
- "learning_rate": 6.095205243187306e-05,
825
- "loss": 0.0426,
826
- "step": 23600
827
- },
828
- {
829
- "epoch": 24.01,
830
- "eval_loss": 0.5209127068519592,
831
- "eval_runtime": 809.3899,
832
- "eval_samples_per_second": 2.312,
833
- "eval_wer": 0.3356879606879607,
834
- "step": 23600
835
- },
836
- {
837
- "epoch": 24.42,
838
- "learning_rate": 5.6812694032424966e-05,
839
- "loss": 0.0407,
840
- "step": 24000
841
- },
842
- {
843
- "epoch": 24.42,
844
- "eval_loss": 0.521318256855011,
845
- "eval_runtime": 685.3958,
846
- "eval_samples_per_second": 2.73,
847
- "eval_wer": 0.3367936117936118,
848
- "step": 24000
849
- },
850
- {
851
- "epoch": 24.82,
852
- "learning_rate": 5.2673335632976885e-05,
853
- "loss": 0.0384,
854
- "step": 24400
855
- },
856
- {
857
- "epoch": 24.82,
858
- "eval_loss": 0.533201277256012,
859
- "eval_runtime": 995.2009,
860
- "eval_samples_per_second": 1.88,
861
- "eval_wer": 0.3352579852579853,
862
- "step": 24400
863
- },
864
- {
865
- "epoch": 25.23,
866
- "learning_rate": 4.8533977233528796e-05,
867
- "loss": 0.0381,
868
- "step": 24800
869
- },
870
- {
871
- "epoch": 25.23,
872
- "eval_loss": 0.5235264897346497,
873
- "eval_runtime": 297.7118,
874
- "eval_samples_per_second": 6.285,
875
- "eval_wer": 0.33175675675675675,
876
- "step": 24800
877
- },
878
- {
879
- "epoch": 25.64,
880
- "learning_rate": 4.4394618834080715e-05,
881
- "loss": 0.0392,
882
- "step": 25200
883
- },
884
- {
885
- "epoch": 25.64,
886
- "eval_loss": 0.5155506134033203,
887
- "eval_runtime": 236.6244,
888
- "eval_samples_per_second": 7.907,
889
- "eval_wer": 0.3305896805896806,
890
- "step": 25200
891
- },
892
- {
893
- "epoch": 26.04,
894
- "learning_rate": 4.025526043463263e-05,
895
- "loss": 0.0374,
896
- "step": 25600
897
- },
898
- {
899
- "epoch": 26.04,
900
- "eval_loss": 0.52613765001297,
901
- "eval_runtime": 238.4752,
902
- "eval_samples_per_second": 7.846,
903
- "eval_wer": 0.3332309582309582,
904
- "step": 25600
905
- },
906
- {
907
- "epoch": 26.45,
908
- "learning_rate": 3.611590203518454e-05,
909
- "loss": 0.0352,
910
- "step": 26000
911
- },
912
- {
913
- "epoch": 26.45,
914
- "eval_loss": 0.5194851756095886,
915
- "eval_runtime": 237.5911,
916
- "eval_samples_per_second": 7.875,
917
- "eval_wer": 0.33132678132678134,
918
- "step": 26000
919
- },
920
- {
921
- "epoch": 26.86,
922
- "learning_rate": 3.197654363573646e-05,
923
- "loss": 0.0357,
924
- "step": 26400
925
- },
926
- {
927
- "epoch": 26.86,
928
- "eval_loss": 0.5337009429931641,
929
- "eval_runtime": 236.9359,
930
- "eval_samples_per_second": 7.897,
931
- "eval_wer": 0.33218673218673217,
932
- "step": 26400
933
- },
934
- {
935
- "epoch": 27.26,
936
- "learning_rate": 2.7837185236288373e-05,
937
- "loss": 0.0342,
938
- "step": 26800
939
- },
940
- {
941
- "epoch": 27.26,
942
- "eval_loss": 0.5221392512321472,
943
- "eval_runtime": 237.1237,
944
- "eval_samples_per_second": 7.89,
945
- "eval_wer": 0.3324324324324324,
946
- "step": 26800
947
- },
948
- {
949
- "epoch": 27.67,
950
- "learning_rate": 2.3697826836840288e-05,
951
- "loss": 0.0346,
952
- "step": 27200
953
- },
954
- {
955
- "epoch": 27.67,
956
- "eval_loss": 0.5248429775238037,
957
- "eval_runtime": 238.7028,
958
- "eval_samples_per_second": 7.838,
959
- "eval_wer": 0.32843980343980345,
960
- "step": 27200
961
- },
962
- {
963
- "epoch": 28.08,
964
- "learning_rate": 1.95584684373922e-05,
965
- "loss": 0.0343,
966
- "step": 27600
967
- },
968
- {
969
- "epoch": 28.08,
970
- "eval_loss": 0.5455237030982971,
971
- "eval_runtime": 238.0797,
972
- "eval_samples_per_second": 7.859,
973
- "eval_wer": 0.3278869778869779,
974
- "step": 27600
975
- },
976
- {
977
- "epoch": 28.48,
978
- "learning_rate": 1.541911003794412e-05,
979
- "loss": 0.034,
980
- "step": 28000
981
- },
982
- {
983
- "epoch": 28.48,
984
- "eval_loss": 0.5291683077812195,
985
- "eval_runtime": 300.3419,
986
- "eval_samples_per_second": 6.23,
987
- "eval_wer": 0.3269041769041769,
988
- "step": 28000
989
- },
990
- {
991
- "epoch": 28.89,
992
- "learning_rate": 1.1279751638496032e-05,
993
- "loss": 0.0337,
994
- "step": 28400
995
- },
996
- {
997
- "epoch": 28.89,
998
- "eval_loss": 0.5360307693481445,
999
- "eval_runtime": 238.3743,
1000
- "eval_samples_per_second": 7.849,
1001
- "eval_wer": 0.3269041769041769,
1002
- "step": 28400
1003
- },
1004
- {
1005
- "epoch": 29.3,
1006
- "learning_rate": 7.140393239047947e-06,
1007
- "loss": 0.032,
1008
- "step": 28800
1009
- },
1010
- {
1011
- "epoch": 29.3,
1012
- "eval_loss": 0.5361535549163818,
1013
- "eval_runtime": 851.2223,
1014
- "eval_samples_per_second": 2.198,
1015
- "eval_wer": 0.3257985257985258,
1016
- "step": 28800
1017
- },
1018
- {
1019
- "epoch": 29.7,
1020
- "learning_rate": 3.001034839599862e-06,
1021
- "loss": 0.0314,
1022
- "step": 29200
1023
- },
1024
- {
1025
- "epoch": 29.7,
1026
- "eval_loss": 0.5325431823730469,
1027
- "eval_runtime": 1467.4683,
1028
- "eval_samples_per_second": 1.275,
1029
- "eval_wer": 0.32524570024570026,
1030
- "step": 29200
1031
  }
1032
  ],
1033
- "max_steps": 29490,
1034
- "num_train_epochs": 30,
1035
- "total_flos": 1.620248435122493e+20,
1036
  "trial_name": null,
1037
  "trial_params": null
1038
  }
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.6230529595015576,
5
+ "global_step": 1200,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
+ "epoch": 0.21,
12
+ "learning_rate": 2.842473646947527e-06,
13
+ "loss": 0.1561,
14
  "step": 400
15
  },
16
  {
17
+ "epoch": 0.21,
18
+ "eval_loss": 0.16617341339588165,
19
+ "eval_runtime": 237.5111,
20
+ "eval_samples_per_second": 7.878,
21
+ "eval_wer": 0.25878220140515223,
22
  "step": 400
23
  },
24
  {
25
+ "epoch": 0.42,
26
+ "learning_rate": 2.63105872341108e-06,
27
+ "loss": 0.152,
28
  "step": 800
29
  },
30
  {
31
+ "epoch": 0.42,
32
+ "eval_loss": 0.16523481905460358,
33
+ "eval_runtime": 238.503,
34
+ "eval_samples_per_second": 7.845,
35
+ "eval_wer": 0.25843091334894613,
36
  "step": 800
37
  },
38
  {
39
+ "epoch": 0.62,
40
+ "learning_rate": 2.4196437998746334e-06,
41
+ "loss": 0.1537,
42
  "step": 1200
43
  },
44
  {
45
+ "epoch": 0.62,
46
+ "eval_loss": 0.16467183828353882,
47
+ "eval_runtime": 240.7886,
48
+ "eval_samples_per_second": 7.77,
49
+ "eval_wer": 0.2573770491803279,
50
  "step": 1200
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
51
  }
52
  ],
53
+ "max_steps": 5778,
54
+ "num_train_epochs": 3,
55
+ "total_flos": 6.372500637308682e+18,
56
  "trial_name": null,
57
  "trial_params": null
58
  }
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:db301806c4ec290ebc2bb9eeb602193eaa2657d9ae8f5e00fefcdc2cf5f00e60
3
  size 2351
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d2f7a2ddf41246ad0cbcd373df0ff8f1dc4300c370d379996d9428542d8e4a12
3
  size 2351