anuragshas commited on
Commit
5a340fd
1 Parent(s): c71bac1

End of training

Browse files
Files changed (4) hide show
  1. all_results.json +15 -0
  2. eval_results.json +10 -0
  3. train_results.json +8 -0
  4. trainer_state.json +541 -0
all_results.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 197.55,
3
+ "eval_cer": 0.0946218197167965,
4
+ "eval_loss": 0.3641946017742157,
5
+ "eval_runtime": 96.4471,
6
+ "eval_samples": 1806,
7
+ "eval_samples_per_second": 18.725,
8
+ "eval_steps_per_second": 0.301,
9
+ "eval_wer": 0.41896696619044543,
10
+ "train_loss": 1.5647427196365646,
11
+ "train_runtime": 50885.6899,
12
+ "train_samples": 3919,
13
+ "train_samples_per_second": 15.405,
14
+ "train_steps_per_second": 0.12
15
+ }
eval_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 197.55,
3
+ "eval_cer": 0.0946218197167965,
4
+ "eval_loss": 0.3641946017742157,
5
+ "eval_runtime": 96.4471,
6
+ "eval_samples": 1806,
7
+ "eval_samples_per_second": 18.725,
8
+ "eval_steps_per_second": 0.301,
9
+ "eval_wer": 0.41896696619044543
10
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 197.55,
3
+ "train_loss": 1.5647427196365646,
4
+ "train_runtime": 50885.6899,
5
+ "train_samples": 3919,
6
+ "train_samples_per_second": 15.405,
7
+ "train_steps_per_second": 0.12
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,541 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 197.5483870967742,
5
+ "global_step": 6124,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 3.23,
12
+ "learning_rate": 1.2639705882352941e-05,
13
+ "loss": 14.7415,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 6.45,
18
+ "learning_rate": 2.477205882352941e-05,
19
+ "loss": 6.4415,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 9.68,
24
+ "learning_rate": 3.690441176470588e-05,
25
+ "loss": 4.4273,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 12.9,
30
+ "learning_rate": 4.9036764705882346e-05,
31
+ "loss": 3.5184,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 12.9,
36
+ "eval_cer": 1.0,
37
+ "eval_loss": 3.4210262298583984,
38
+ "eval_runtime": 96.9515,
39
+ "eval_samples_per_second": 18.628,
40
+ "eval_steps_per_second": 0.299,
41
+ "eval_wer": 1.0,
42
+ "step": 400
43
+ },
44
+ {
45
+ "epoch": 16.13,
46
+ "learning_rate": 6.116911764705881e-05,
47
+ "loss": 3.3554,
48
+ "step": 500
49
+ },
50
+ {
51
+ "epoch": 19.35,
52
+ "learning_rate": 7.330147058823529e-05,
53
+ "loss": 3.2617,
54
+ "step": 600
55
+ },
56
+ {
57
+ "epoch": 22.58,
58
+ "learning_rate": 7.5e-05,
59
+ "loss": 3.0887,
60
+ "step": 700
61
+ },
62
+ {
63
+ "epoch": 25.81,
64
+ "learning_rate": 7.5e-05,
65
+ "loss": 2.3797,
66
+ "step": 800
67
+ },
68
+ {
69
+ "epoch": 25.81,
70
+ "eval_cer": 0.2583848190644307,
71
+ "eval_loss": 1.1068178415298462,
72
+ "eval_runtime": 99.1553,
73
+ "eval_samples_per_second": 18.214,
74
+ "eval_steps_per_second": 0.292,
75
+ "eval_wer": 0.8389036136789708,
76
+ "step": 800
77
+ },
78
+ {
79
+ "epoch": 29.03,
80
+ "learning_rate": 7.5e-05,
81
+ "loss": 1.8558,
82
+ "step": 900
83
+ },
84
+ {
85
+ "epoch": 32.26,
86
+ "learning_rate": 7.5e-05,
87
+ "loss": 1.6732,
88
+ "step": 1000
89
+ },
90
+ {
91
+ "epoch": 35.48,
92
+ "learning_rate": 7.5e-05,
93
+ "loss": 1.5624,
94
+ "step": 1100
95
+ },
96
+ {
97
+ "epoch": 38.71,
98
+ "learning_rate": 7.5e-05,
99
+ "loss": 1.5022,
100
+ "step": 1200
101
+ },
102
+ {
103
+ "epoch": 38.71,
104
+ "eval_cer": 0.1516846387044783,
105
+ "eval_loss": 0.5277877449989319,
106
+ "eval_runtime": 102.5824,
107
+ "eval_samples_per_second": 17.605,
108
+ "eval_steps_per_second": 0.283,
109
+ "eval_wer": 0.6279656086366281,
110
+ "step": 1200
111
+ },
112
+ {
113
+ "epoch": 41.94,
114
+ "learning_rate": 7.5e-05,
115
+ "loss": 1.4358,
116
+ "step": 1300
117
+ },
118
+ {
119
+ "epoch": 45.16,
120
+ "learning_rate": 7.5e-05,
121
+ "loss": 1.388,
122
+ "step": 1400
123
+ },
124
+ {
125
+ "epoch": 48.39,
126
+ "learning_rate": 7.5e-05,
127
+ "loss": 1.3493,
128
+ "step": 1500
129
+ },
130
+ {
131
+ "epoch": 51.61,
132
+ "learning_rate": 7.5e-05,
133
+ "loss": 1.3181,
134
+ "step": 1600
135
+ },
136
+ {
137
+ "epoch": 51.61,
138
+ "eval_cer": 0.12968648067846042,
139
+ "eval_loss": 0.4253957271575928,
140
+ "eval_runtime": 98.985,
141
+ "eval_samples_per_second": 18.245,
142
+ "eval_steps_per_second": 0.293,
143
+ "eval_wer": 0.5587303639537139,
144
+ "step": 1600
145
+ },
146
+ {
147
+ "epoch": 54.84,
148
+ "learning_rate": 7.5e-05,
149
+ "loss": 1.2814,
150
+ "step": 1700
151
+ },
152
+ {
153
+ "epoch": 58.06,
154
+ "learning_rate": 7.5e-05,
155
+ "loss": 1.2547,
156
+ "step": 1800
157
+ },
158
+ {
159
+ "epoch": 61.29,
160
+ "learning_rate": 7.5e-05,
161
+ "loss": 1.2244,
162
+ "step": 1900
163
+ },
164
+ {
165
+ "epoch": 64.52,
166
+ "learning_rate": 7.5e-05,
167
+ "loss": 1.2037,
168
+ "step": 2000
169
+ },
170
+ {
171
+ "epoch": 64.52,
172
+ "eval_cer": 0.11762730726428489,
173
+ "eval_loss": 0.3835846781730652,
174
+ "eval_runtime": 99.4181,
175
+ "eval_samples_per_second": 18.166,
176
+ "eval_steps_per_second": 0.292,
177
+ "eval_wer": 0.5142543150817764,
178
+ "step": 2000
179
+ },
180
+ {
181
+ "epoch": 67.74,
182
+ "learning_rate": 7.5e-05,
183
+ "loss": 1.1751,
184
+ "step": 2100
185
+ },
186
+ {
187
+ "epoch": 70.97,
188
+ "learning_rate": 7.5e-05,
189
+ "loss": 1.1642,
190
+ "step": 2200
191
+ },
192
+ {
193
+ "epoch": 74.19,
194
+ "learning_rate": 7.5e-05,
195
+ "loss": 1.1397,
196
+ "step": 2300
197
+ },
198
+ {
199
+ "epoch": 77.42,
200
+ "learning_rate": 7.5e-05,
201
+ "loss": 1.1245,
202
+ "step": 2400
203
+ },
204
+ {
205
+ "epoch": 77.42,
206
+ "eval_cer": 0.11111324302544226,
207
+ "eval_loss": 0.36429011821746826,
208
+ "eval_runtime": 98.0055,
209
+ "eval_samples_per_second": 18.428,
210
+ "eval_steps_per_second": 0.296,
211
+ "eval_wer": 0.48710323873553557,
212
+ "step": 2400
213
+ },
214
+ {
215
+ "epoch": 80.65,
216
+ "learning_rate": 7.5e-05,
217
+ "loss": 1.1057,
218
+ "step": 2500
219
+ },
220
+ {
221
+ "epoch": 83.87,
222
+ "learning_rate": 7.5e-05,
223
+ "loss": 1.0852,
224
+ "step": 2600
225
+ },
226
+ {
227
+ "epoch": 87.1,
228
+ "learning_rate": 7.5e-05,
229
+ "loss": 1.073,
230
+ "step": 2700
231
+ },
232
+ {
233
+ "epoch": 90.32,
234
+ "learning_rate": 7.5e-05,
235
+ "loss": 1.0582,
236
+ "step": 2800
237
+ },
238
+ {
239
+ "epoch": 90.32,
240
+ "eval_cer": 0.10623009324993285,
241
+ "eval_loss": 0.3561805486679077,
242
+ "eval_runtime": 94.8015,
243
+ "eval_samples_per_second": 19.05,
244
+ "eval_steps_per_second": 0.306,
245
+ "eval_wer": 0.46758032193419097,
246
+ "step": 2800
247
+ },
248
+ {
249
+ "epoch": 93.55,
250
+ "learning_rate": 7.5e-05,
251
+ "loss": 1.0376,
252
+ "step": 2900
253
+ },
254
+ {
255
+ "epoch": 96.77,
256
+ "learning_rate": 7.5e-05,
257
+ "loss": 1.0341,
258
+ "step": 3000
259
+ },
260
+ {
261
+ "epoch": 100.0,
262
+ "learning_rate": 7.416258570029382e-05,
263
+ "loss": 1.017,
264
+ "step": 3100
265
+ },
266
+ {
267
+ "epoch": 103.23,
268
+ "learning_rate": 7.183643486777668e-05,
269
+ "loss": 1.0027,
270
+ "step": 3200
271
+ },
272
+ {
273
+ "epoch": 103.23,
274
+ "eval_cer": 0.10575041252542308,
275
+ "eval_loss": 0.3529968559741974,
276
+ "eval_runtime": 98.9163,
277
+ "eval_samples_per_second": 18.258,
278
+ "eval_steps_per_second": 0.293,
279
+ "eval_wer": 0.4625379791841748,
280
+ "step": 3200
281
+ },
282
+ {
283
+ "epoch": 106.45,
284
+ "learning_rate": 6.951028403525955e-05,
285
+ "loss": 0.9864,
286
+ "step": 3300
287
+ },
288
+ {
289
+ "epoch": 109.68,
290
+ "learning_rate": 6.71841332027424e-05,
291
+ "loss": 0.9769,
292
+ "step": 3400
293
+ },
294
+ {
295
+ "epoch": 112.9,
296
+ "learning_rate": 6.485798237022525e-05,
297
+ "loss": 0.9587,
298
+ "step": 3500
299
+ },
300
+ {
301
+ "epoch": 116.13,
302
+ "learning_rate": 6.253183153770813e-05,
303
+ "loss": 0.9382,
304
+ "step": 3600
305
+ },
306
+ {
307
+ "epoch": 116.13,
308
+ "eval_cer": 0.10022449057907057,
309
+ "eval_loss": 0.3388434052467346,
310
+ "eval_runtime": 98.6206,
311
+ "eval_samples_per_second": 18.313,
312
+ "eval_steps_per_second": 0.294,
313
+ "eval_wer": 0.44424332536039823,
314
+ "step": 3600
315
+ },
316
+ {
317
+ "epoch": 119.35,
318
+ "learning_rate": 6.020568070519098e-05,
319
+ "loss": 0.9296,
320
+ "step": 3700
321
+ },
322
+ {
323
+ "epoch": 122.58,
324
+ "learning_rate": 5.787952987267385e-05,
325
+ "loss": 0.9145,
326
+ "step": 3800
327
+ },
328
+ {
329
+ "epoch": 125.81,
330
+ "learning_rate": 5.5553379040156705e-05,
331
+ "loss": 0.9043,
332
+ "step": 3900
333
+ },
334
+ {
335
+ "epoch": 129.03,
336
+ "learning_rate": 5.3227228207639564e-05,
337
+ "loss": 0.8915,
338
+ "step": 4000
339
+ },
340
+ {
341
+ "epoch": 129.03,
342
+ "eval_cer": 0.10004221190375685,
343
+ "eval_loss": 0.3429908752441406,
344
+ "eval_runtime": 98.3419,
345
+ "eval_samples_per_second": 18.364,
346
+ "eval_steps_per_second": 0.295,
347
+ "eval_wer": 0.44269183528347017,
348
+ "step": 4000
349
+ },
350
+ {
351
+ "epoch": 132.26,
352
+ "learning_rate": 5.090107737512243e-05,
353
+ "loss": 0.8768,
354
+ "step": 4100
355
+ },
356
+ {
357
+ "epoch": 135.48,
358
+ "learning_rate": 4.857492654260528e-05,
359
+ "loss": 0.8697,
360
+ "step": 4200
361
+ },
362
+ {
363
+ "epoch": 138.71,
364
+ "learning_rate": 4.6248775710088145e-05,
365
+ "loss": 0.8604,
366
+ "step": 4300
367
+ },
368
+ {
369
+ "epoch": 141.94,
370
+ "learning_rate": 4.394588638589617e-05,
371
+ "loss": 0.853,
372
+ "step": 4400
373
+ },
374
+ {
375
+ "epoch": 141.94,
376
+ "eval_cer": 0.09997505660232549,
377
+ "eval_loss": 0.35362082719802856,
378
+ "eval_runtime": 98.9414,
379
+ "eval_samples_per_second": 18.253,
380
+ "eval_steps_per_second": 0.293,
381
+ "eval_wer": 0.437455556273838,
382
+ "step": 4400
383
+ },
384
+ {
385
+ "epoch": 145.16,
386
+ "learning_rate": 4.161973555337904e-05,
387
+ "loss": 0.8447,
388
+ "step": 4500
389
+ },
390
+ {
391
+ "epoch": 148.39,
392
+ "learning_rate": 3.9293584720861896e-05,
393
+ "loss": 0.8309,
394
+ "step": 4600
395
+ },
396
+ {
397
+ "epoch": 151.61,
398
+ "learning_rate": 3.6967433888344755e-05,
399
+ "loss": 0.827,
400
+ "step": 4700
401
+ },
402
+ {
403
+ "epoch": 154.84,
404
+ "learning_rate": 3.464128305582762e-05,
405
+ "loss": 0.8127,
406
+ "step": 4800
407
+ },
408
+ {
409
+ "epoch": 154.84,
410
+ "eval_cer": 0.09857438888675697,
411
+ "eval_loss": 0.3511004149913788,
412
+ "eval_runtime": 97.5708,
413
+ "eval_samples_per_second": 18.51,
414
+ "eval_steps_per_second": 0.297,
415
+ "eval_wer": 0.4343525761199819,
416
+ "step": 4800
417
+ },
418
+ {
419
+ "epoch": 158.06,
420
+ "learning_rate": 3.231513222331048e-05,
421
+ "loss": 0.8123,
422
+ "step": 4900
423
+ },
424
+ {
425
+ "epoch": 161.29,
426
+ "learning_rate": 2.9988981390793336e-05,
427
+ "loss": 0.8063,
428
+ "step": 5000
429
+ },
430
+ {
431
+ "epoch": 164.52,
432
+ "learning_rate": 2.76628305582762e-05,
433
+ "loss": 0.8046,
434
+ "step": 5100
435
+ },
436
+ {
437
+ "epoch": 167.74,
438
+ "learning_rate": 2.5336679725759056e-05,
439
+ "loss": 0.7861,
440
+ "step": 5200
441
+ },
442
+ {
443
+ "epoch": 167.74,
444
+ "eval_cer": 0.09927472274454123,
445
+ "eval_loss": 0.3595349192619324,
446
+ "eval_runtime": 97.9334,
447
+ "eval_samples_per_second": 18.441,
448
+ "eval_steps_per_second": 0.296,
449
+ "eval_wer": 0.43719697459435,
450
+ "step": 5200
451
+ },
452
+ {
453
+ "epoch": 170.97,
454
+ "learning_rate": 2.301052889324192e-05,
455
+ "loss": 0.7904,
456
+ "step": 5300
457
+ },
458
+ {
459
+ "epoch": 174.19,
460
+ "learning_rate": 2.068437806072478e-05,
461
+ "loss": 0.775,
462
+ "step": 5400
463
+ },
464
+ {
465
+ "epoch": 177.42,
466
+ "learning_rate": 1.8358227228207637e-05,
467
+ "loss": 0.7737,
468
+ "step": 5500
469
+ },
470
+ {
471
+ "epoch": 180.65,
472
+ "learning_rate": 1.6032076395690502e-05,
473
+ "loss": 0.7619,
474
+ "step": 5600
475
+ },
476
+ {
477
+ "epoch": 180.65,
478
+ "eval_cer": 0.0985456080432864,
479
+ "eval_loss": 0.3628048300743103,
480
+ "eval_runtime": 97.525,
481
+ "eval_samples_per_second": 18.518,
482
+ "eval_steps_per_second": 0.297,
483
+ "eval_wer": 0.43157282306548583,
484
+ "step": 5600
485
+ },
486
+ {
487
+ "epoch": 183.87,
488
+ "learning_rate": 1.3705925563173359e-05,
489
+ "loss": 0.7654,
490
+ "step": 5700
491
+ },
492
+ {
493
+ "epoch": 187.1,
494
+ "learning_rate": 1.1379774730656215e-05,
495
+ "loss": 0.7586,
496
+ "step": 5800
497
+ },
498
+ {
499
+ "epoch": 190.32,
500
+ "learning_rate": 9.053623898139082e-06,
501
+ "loss": 0.7537,
502
+ "step": 5900
503
+ },
504
+ {
505
+ "epoch": 193.55,
506
+ "learning_rate": 6.727473065621939e-06,
507
+ "loss": 0.7537,
508
+ "step": 6000
509
+ },
510
+ {
511
+ "epoch": 193.55,
512
+ "eval_cer": 0.09426685598065927,
513
+ "eval_loss": 0.3633102476596832,
514
+ "eval_runtime": 94.1997,
515
+ "eval_samples_per_second": 19.172,
516
+ "eval_steps_per_second": 0.308,
517
+ "eval_wer": 0.41735083069364537,
518
+ "step": 6000
519
+ },
520
+ {
521
+ "epoch": 196.77,
522
+ "learning_rate": 4.4245837414299736e-06,
523
+ "loss": 0.7486,
524
+ "step": 6100
525
+ },
526
+ {
527
+ "epoch": 197.55,
528
+ "step": 6124,
529
+ "total_flos": 1.4975736611439493e+20,
530
+ "train_loss": 1.5647427196365646,
531
+ "train_runtime": 50885.6899,
532
+ "train_samples_per_second": 15.405,
533
+ "train_steps_per_second": 0.12
534
+ }
535
+ ],
536
+ "max_steps": 6124,
537
+ "num_train_epochs": 198,
538
+ "total_flos": 1.4975736611439493e+20,
539
+ "trial_name": null,
540
+ "trial_params": null
541
+ }