infinitejoy commited on
Commit
f786895
1 Parent(s): b8db9e4

End of training

Browse files
all_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "eval_loss": 0.25481978058815,
4
+ "eval_runtime": 138.1788,
5
+ "eval_samples": 2395,
6
+ "eval_samples_per_second": 17.333,
7
+ "eval_steps_per_second": 17.333,
8
+ "eval_wer": 0.26876047421016697,
9
+ "train_loss": 1.2786970950635663,
10
+ "train_runtime": 28590.9558,
11
+ "train_samples": 5216,
12
+ "train_samples_per_second": 18.244,
13
+ "train_steps_per_second": 0.57
14
+ }
eval_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "eval_loss": 0.25481978058815,
4
+ "eval_runtime": 138.1788,
5
+ "eval_samples": 2395,
6
+ "eval_samples_per_second": 17.333,
7
+ "eval_steps_per_second": 17.333,
8
+ "eval_wer": 0.26876047421016697
9
+ }
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:859b16edfdd3a00d61d5683b15146f5a1ff1ca6c8f81395c09f1626420dda009
3
  size 1262091761
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc60fdbac076debd2748b610156dc4bf5ce1f4144de7afa9a8bf7fa14d1cf111
3
  size 1262091761
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "train_loss": 1.2786970950635663,
4
+ "train_runtime": 28590.9558,
5
+ "train_samples": 5216,
6
+ "train_samples_per_second": 18.244,
7
+ "train_steps_per_second": 0.57
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,1075 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 100.0,
5
+ "global_step": 16300,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.61,
12
+ "learning_rate": 3.3949999999999997e-06,
13
+ "loss": 21.3924,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 1.23,
18
+ "learning_rate": 6.895e-06,
19
+ "loss": 10.9359,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 1.84,
24
+ "learning_rate": 1.0394999999999998e-05,
25
+ "loss": 5.904,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 2.45,
30
+ "learning_rate": 1.3895e-05,
31
+ "loss": 4.7637,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 3.07,
36
+ "learning_rate": 1.7395e-05,
37
+ "loss": 3.9441,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 3.68,
42
+ "learning_rate": 2.0894999999999996e-05,
43
+ "loss": 3.4134,
44
+ "step": 600
45
+ },
46
+ {
47
+ "epoch": 4.29,
48
+ "learning_rate": 2.4394999999999996e-05,
49
+ "loss": 3.2058,
50
+ "step": 700
51
+ },
52
+ {
53
+ "epoch": 4.91,
54
+ "learning_rate": 2.7895e-05,
55
+ "loss": 3.1361,
56
+ "step": 800
57
+ },
58
+ {
59
+ "epoch": 5.52,
60
+ "learning_rate": 3.1395e-05,
61
+ "loss": 3.0826,
62
+ "step": 900
63
+ },
64
+ {
65
+ "epoch": 6.13,
66
+ "learning_rate": 3.4895e-05,
67
+ "loss": 3.0202,
68
+ "step": 1000
69
+ },
70
+ {
71
+ "epoch": 6.75,
72
+ "learning_rate": 3.8394999999999994e-05,
73
+ "loss": 2.9628,
74
+ "step": 1100
75
+ },
76
+ {
77
+ "epoch": 7.36,
78
+ "learning_rate": 4.1895e-05,
79
+ "loss": 2.9213,
80
+ "step": 1200
81
+ },
82
+ {
83
+ "epoch": 7.98,
84
+ "learning_rate": 4.5394999999999995e-05,
85
+ "loss": 2.6817,
86
+ "step": 1300
87
+ },
88
+ {
89
+ "epoch": 8.59,
90
+ "learning_rate": 4.8895e-05,
91
+ "loss": 2.166,
92
+ "step": 1400
93
+ },
94
+ {
95
+ "epoch": 9.2,
96
+ "learning_rate": 5.2395e-05,
97
+ "loss": 1.7136,
98
+ "step": 1500
99
+ },
100
+ {
101
+ "epoch": 9.82,
102
+ "learning_rate": 5.589499999999999e-05,
103
+ "loss": 1.5318,
104
+ "step": 1600
105
+ },
106
+ {
107
+ "epoch": 10.43,
108
+ "learning_rate": 5.9394999999999996e-05,
109
+ "loss": 1.4485,
110
+ "step": 1700
111
+ },
112
+ {
113
+ "epoch": 11.04,
114
+ "learning_rate": 6.289499999999999e-05,
115
+ "loss": 1.3967,
116
+ "step": 1800
117
+ },
118
+ {
119
+ "epoch": 11.66,
120
+ "learning_rate": 6.639499999999999e-05,
121
+ "loss": 1.3518,
122
+ "step": 1900
123
+ },
124
+ {
125
+ "epoch": 12.27,
126
+ "learning_rate": 6.9895e-05,
127
+ "loss": 1.3161,
128
+ "step": 2000
129
+ },
130
+ {
131
+ "epoch": 12.27,
132
+ "eval_loss": 0.419926255941391,
133
+ "eval_runtime": 134.9817,
134
+ "eval_samples_per_second": 17.743,
135
+ "eval_steps_per_second": 17.743,
136
+ "eval_wer": 0.4796722736018869,
137
+ "step": 2000
138
+ },
139
+ {
140
+ "epoch": 12.88,
141
+ "learning_rate": 6.952517482517481e-05,
142
+ "loss": 1.3043,
143
+ "step": 2100
144
+ },
145
+ {
146
+ "epoch": 13.5,
147
+ "learning_rate": 6.903566433566433e-05,
148
+ "loss": 1.2899,
149
+ "step": 2200
150
+ },
151
+ {
152
+ "epoch": 14.11,
153
+ "learning_rate": 6.854615384615384e-05,
154
+ "loss": 1.2446,
155
+ "step": 2300
156
+ },
157
+ {
158
+ "epoch": 14.72,
159
+ "learning_rate": 6.805664335664335e-05,
160
+ "loss": 1.243,
161
+ "step": 2400
162
+ },
163
+ {
164
+ "epoch": 15.34,
165
+ "learning_rate": 6.756713286713286e-05,
166
+ "loss": 1.2123,
167
+ "step": 2500
168
+ },
169
+ {
170
+ "epoch": 15.95,
171
+ "learning_rate": 6.707762237762238e-05,
172
+ "loss": 1.2192,
173
+ "step": 2600
174
+ },
175
+ {
176
+ "epoch": 16.56,
177
+ "learning_rate": 6.658811188811187e-05,
178
+ "loss": 1.1792,
179
+ "step": 2700
180
+ },
181
+ {
182
+ "epoch": 17.18,
183
+ "learning_rate": 6.609860139860139e-05,
184
+ "loss": 1.1835,
185
+ "step": 2800
186
+ },
187
+ {
188
+ "epoch": 17.79,
189
+ "learning_rate": 6.56090909090909e-05,
190
+ "loss": 1.146,
191
+ "step": 2900
192
+ },
193
+ {
194
+ "epoch": 18.4,
195
+ "learning_rate": 6.511958041958041e-05,
196
+ "loss": 1.1756,
197
+ "step": 3000
198
+ },
199
+ {
200
+ "epoch": 19.02,
201
+ "learning_rate": 6.463006993006992e-05,
202
+ "loss": 1.1423,
203
+ "step": 3100
204
+ },
205
+ {
206
+ "epoch": 19.63,
207
+ "learning_rate": 6.414055944055944e-05,
208
+ "loss": 1.1392,
209
+ "step": 3200
210
+ },
211
+ {
212
+ "epoch": 20.25,
213
+ "learning_rate": 6.365104895104895e-05,
214
+ "loss": 1.1176,
215
+ "step": 3300
216
+ },
217
+ {
218
+ "epoch": 20.86,
219
+ "learning_rate": 6.316153846153845e-05,
220
+ "loss": 1.1073,
221
+ "step": 3400
222
+ },
223
+ {
224
+ "epoch": 21.47,
225
+ "learning_rate": 6.267202797202796e-05,
226
+ "loss": 1.096,
227
+ "step": 3500
228
+ },
229
+ {
230
+ "epoch": 22.09,
231
+ "learning_rate": 6.218251748251747e-05,
232
+ "loss": 1.0931,
233
+ "step": 3600
234
+ },
235
+ {
236
+ "epoch": 22.7,
237
+ "learning_rate": 6.169300699300698e-05,
238
+ "loss": 1.0872,
239
+ "step": 3700
240
+ },
241
+ {
242
+ "epoch": 23.31,
243
+ "learning_rate": 6.12034965034965e-05,
244
+ "loss": 1.0687,
245
+ "step": 3800
246
+ },
247
+ {
248
+ "epoch": 23.93,
249
+ "learning_rate": 6.071398601398601e-05,
250
+ "loss": 1.0871,
251
+ "step": 3900
252
+ },
253
+ {
254
+ "epoch": 24.54,
255
+ "learning_rate": 6.022447552447552e-05,
256
+ "loss": 1.0643,
257
+ "step": 4000
258
+ },
259
+ {
260
+ "epoch": 24.54,
261
+ "eval_loss": 0.29822662472724915,
262
+ "eval_runtime": 132.5892,
263
+ "eval_samples_per_second": 18.063,
264
+ "eval_steps_per_second": 18.063,
265
+ "eval_wer": 0.3721060145242381,
266
+ "step": 4000
267
+ },
268
+ {
269
+ "epoch": 25.15,
270
+ "learning_rate": 5.973496503496503e-05,
271
+ "loss": 1.0613,
272
+ "step": 4100
273
+ },
274
+ {
275
+ "epoch": 25.77,
276
+ "learning_rate": 5.924545454545454e-05,
277
+ "loss": 1.059,
278
+ "step": 4200
279
+ },
280
+ {
281
+ "epoch": 26.38,
282
+ "learning_rate": 5.876083916083915e-05,
283
+ "loss": 1.0477,
284
+ "step": 4300
285
+ },
286
+ {
287
+ "epoch": 26.99,
288
+ "learning_rate": 5.827132867132866e-05,
289
+ "loss": 1.0558,
290
+ "step": 4400
291
+ },
292
+ {
293
+ "epoch": 27.61,
294
+ "learning_rate": 5.778671328671328e-05,
295
+ "loss": 1.0401,
296
+ "step": 4500
297
+ },
298
+ {
299
+ "epoch": 28.22,
300
+ "learning_rate": 5.729720279720279e-05,
301
+ "loss": 1.0417,
302
+ "step": 4600
303
+ },
304
+ {
305
+ "epoch": 28.83,
306
+ "learning_rate": 5.68076923076923e-05,
307
+ "loss": 1.0258,
308
+ "step": 4700
309
+ },
310
+ {
311
+ "epoch": 29.45,
312
+ "learning_rate": 5.631818181818181e-05,
313
+ "loss": 1.0307,
314
+ "step": 4800
315
+ },
316
+ {
317
+ "epoch": 30.06,
318
+ "learning_rate": 5.5828671328671325e-05,
319
+ "loss": 1.0112,
320
+ "step": 4900
321
+ },
322
+ {
323
+ "epoch": 30.67,
324
+ "learning_rate": 5.5344055944055944e-05,
325
+ "loss": 1.0008,
326
+ "step": 5000
327
+ },
328
+ {
329
+ "epoch": 31.29,
330
+ "learning_rate": 5.485454545454545e-05,
331
+ "loss": 1.0034,
332
+ "step": 5100
333
+ },
334
+ {
335
+ "epoch": 31.9,
336
+ "learning_rate": 5.4365034965034955e-05,
337
+ "loss": 1.0028,
338
+ "step": 5200
339
+ },
340
+ {
341
+ "epoch": 32.52,
342
+ "learning_rate": 5.387552447552447e-05,
343
+ "loss": 0.9832,
344
+ "step": 5300
345
+ },
346
+ {
347
+ "epoch": 33.13,
348
+ "learning_rate": 5.338601398601398e-05,
349
+ "loss": 0.9872,
350
+ "step": 5400
351
+ },
352
+ {
353
+ "epoch": 33.74,
354
+ "learning_rate": 5.289650349650349e-05,
355
+ "loss": 0.998,
356
+ "step": 5500
357
+ },
358
+ {
359
+ "epoch": 34.36,
360
+ "learning_rate": 5.2406993006993005e-05,
361
+ "loss": 0.9826,
362
+ "step": 5600
363
+ },
364
+ {
365
+ "epoch": 34.97,
366
+ "learning_rate": 5.191748251748252e-05,
367
+ "loss": 0.9927,
368
+ "step": 5700
369
+ },
370
+ {
371
+ "epoch": 35.58,
372
+ "learning_rate": 5.142797202797203e-05,
373
+ "loss": 0.9891,
374
+ "step": 5800
375
+ },
376
+ {
377
+ "epoch": 36.2,
378
+ "learning_rate": 5.093846153846153e-05,
379
+ "loss": 0.9817,
380
+ "step": 5900
381
+ },
382
+ {
383
+ "epoch": 36.81,
384
+ "learning_rate": 5.044895104895104e-05,
385
+ "loss": 0.9718,
386
+ "step": 6000
387
+ },
388
+ {
389
+ "epoch": 36.81,
390
+ "eval_loss": 0.27623239159584045,
391
+ "eval_runtime": 136.6558,
392
+ "eval_samples_per_second": 17.526,
393
+ "eval_steps_per_second": 17.526,
394
+ "eval_wer": 0.3333126435354727,
395
+ "step": 6000
396
+ },
397
+ {
398
+ "epoch": 37.42,
399
+ "learning_rate": 4.995944055944055e-05,
400
+ "loss": 0.9633,
401
+ "step": 6100
402
+ },
403
+ {
404
+ "epoch": 38.04,
405
+ "learning_rate": 4.9469930069930065e-05,
406
+ "loss": 0.958,
407
+ "step": 6200
408
+ },
409
+ {
410
+ "epoch": 38.65,
411
+ "learning_rate": 4.898041958041958e-05,
412
+ "loss": 0.9586,
413
+ "step": 6300
414
+ },
415
+ {
416
+ "epoch": 39.26,
417
+ "learning_rate": 4.849580419580419e-05,
418
+ "loss": 0.9523,
419
+ "step": 6400
420
+ },
421
+ {
422
+ "epoch": 39.88,
423
+ "learning_rate": 4.80062937062937e-05,
424
+ "loss": 0.9393,
425
+ "step": 6500
426
+ },
427
+ {
428
+ "epoch": 40.49,
429
+ "learning_rate": 4.7516783216783215e-05,
430
+ "loss": 0.9456,
431
+ "step": 6600
432
+ },
433
+ {
434
+ "epoch": 41.1,
435
+ "learning_rate": 4.702727272727273e-05,
436
+ "loss": 0.9357,
437
+ "step": 6700
438
+ },
439
+ {
440
+ "epoch": 41.72,
441
+ "learning_rate": 4.653776223776223e-05,
442
+ "loss": 0.9249,
443
+ "step": 6800
444
+ },
445
+ {
446
+ "epoch": 42.33,
447
+ "learning_rate": 4.6048251748251745e-05,
448
+ "loss": 0.9207,
449
+ "step": 6900
450
+ },
451
+ {
452
+ "epoch": 42.94,
453
+ "learning_rate": 4.555874125874125e-05,
454
+ "loss": 0.9177,
455
+ "step": 7000
456
+ },
457
+ {
458
+ "epoch": 43.56,
459
+ "learning_rate": 4.506923076923076e-05,
460
+ "loss": 0.8962,
461
+ "step": 7100
462
+ },
463
+ {
464
+ "epoch": 44.17,
465
+ "learning_rate": 4.4579720279720275e-05,
466
+ "loss": 0.9111,
467
+ "step": 7200
468
+ },
469
+ {
470
+ "epoch": 44.79,
471
+ "learning_rate": 4.409020979020979e-05,
472
+ "loss": 0.918,
473
+ "step": 7300
474
+ },
475
+ {
476
+ "epoch": 45.4,
477
+ "learning_rate": 4.36006993006993e-05,
478
+ "loss": 0.9098,
479
+ "step": 7400
480
+ },
481
+ {
482
+ "epoch": 46.01,
483
+ "learning_rate": 4.3111188811188805e-05,
484
+ "loss": 0.8899,
485
+ "step": 7500
486
+ },
487
+ {
488
+ "epoch": 46.63,
489
+ "learning_rate": 4.262167832167832e-05,
490
+ "loss": 0.8959,
491
+ "step": 7600
492
+ },
493
+ {
494
+ "epoch": 47.24,
495
+ "learning_rate": 4.213216783216783e-05,
496
+ "loss": 0.8981,
497
+ "step": 7700
498
+ },
499
+ {
500
+ "epoch": 47.85,
501
+ "learning_rate": 4.1642657342657336e-05,
502
+ "loss": 0.9062,
503
+ "step": 7800
504
+ },
505
+ {
506
+ "epoch": 48.47,
507
+ "learning_rate": 4.115314685314685e-05,
508
+ "loss": 0.8875,
509
+ "step": 7900
510
+ },
511
+ {
512
+ "epoch": 49.08,
513
+ "learning_rate": 4.066363636363636e-05,
514
+ "loss": 0.8772,
515
+ "step": 8000
516
+ },
517
+ {
518
+ "epoch": 49.08,
519
+ "eval_loss": 0.258562296628952,
520
+ "eval_runtime": 132.8917,
521
+ "eval_samples_per_second": 18.022,
522
+ "eval_steps_per_second": 18.022,
523
+ "eval_wer": 0.30507106945565143,
524
+ "step": 8000
525
+ },
526
+ {
527
+ "epoch": 49.69,
528
+ "learning_rate": 4.017412587412587e-05,
529
+ "loss": 0.8811,
530
+ "step": 8100
531
+ },
532
+ {
533
+ "epoch": 50.31,
534
+ "learning_rate": 3.9684615384615385e-05,
535
+ "loss": 0.8742,
536
+ "step": 8200
537
+ },
538
+ {
539
+ "epoch": 50.92,
540
+ "learning_rate": 3.919510489510489e-05,
541
+ "loss": 0.8676,
542
+ "step": 8300
543
+ },
544
+ {
545
+ "epoch": 51.53,
546
+ "learning_rate": 3.87055944055944e-05,
547
+ "loss": 0.849,
548
+ "step": 8400
549
+ },
550
+ {
551
+ "epoch": 52.15,
552
+ "learning_rate": 3.821608391608391e-05,
553
+ "loss": 0.8752,
554
+ "step": 8500
555
+ },
556
+ {
557
+ "epoch": 52.76,
558
+ "learning_rate": 3.772657342657342e-05,
559
+ "loss": 0.8643,
560
+ "step": 8600
561
+ },
562
+ {
563
+ "epoch": 53.37,
564
+ "learning_rate": 3.723706293706293e-05,
565
+ "loss": 0.8682,
566
+ "step": 8700
567
+ },
568
+ {
569
+ "epoch": 53.99,
570
+ "learning_rate": 3.6747552447552445e-05,
571
+ "loss": 0.8516,
572
+ "step": 8800
573
+ },
574
+ {
575
+ "epoch": 54.6,
576
+ "learning_rate": 3.6262937062937065e-05,
577
+ "loss": 0.8413,
578
+ "step": 8900
579
+ },
580
+ {
581
+ "epoch": 55.21,
582
+ "learning_rate": 3.577342657342657e-05,
583
+ "loss": 0.8539,
584
+ "step": 9000
585
+ },
586
+ {
587
+ "epoch": 55.83,
588
+ "learning_rate": 3.5283916083916076e-05,
589
+ "loss": 0.8325,
590
+ "step": 9100
591
+ },
592
+ {
593
+ "epoch": 56.44,
594
+ "learning_rate": 3.4794405594405595e-05,
595
+ "loss": 0.84,
596
+ "step": 9200
597
+ },
598
+ {
599
+ "epoch": 57.06,
600
+ "learning_rate": 3.43048951048951e-05,
601
+ "loss": 0.8442,
602
+ "step": 9300
603
+ },
604
+ {
605
+ "epoch": 57.67,
606
+ "learning_rate": 3.381538461538461e-05,
607
+ "loss": 0.8271,
608
+ "step": 9400
609
+ },
610
+ {
611
+ "epoch": 58.28,
612
+ "learning_rate": 3.3325874125874125e-05,
613
+ "loss": 0.8277,
614
+ "step": 9500
615
+ },
616
+ {
617
+ "epoch": 58.9,
618
+ "learning_rate": 3.283636363636364e-05,
619
+ "loss": 0.8482,
620
+ "step": 9600
621
+ },
622
+ {
623
+ "epoch": 59.51,
624
+ "learning_rate": 3.234685314685314e-05,
625
+ "loss": 0.8216,
626
+ "step": 9700
627
+ },
628
+ {
629
+ "epoch": 60.12,
630
+ "learning_rate": 3.1857342657342655e-05,
631
+ "loss": 0.8225,
632
+ "step": 9800
633
+ },
634
+ {
635
+ "epoch": 60.74,
636
+ "learning_rate": 3.136783216783217e-05,
637
+ "loss": 0.8121,
638
+ "step": 9900
639
+ },
640
+ {
641
+ "epoch": 61.35,
642
+ "learning_rate": 3.087832167832168e-05,
643
+ "loss": 0.8236,
644
+ "step": 10000
645
+ },
646
+ {
647
+ "epoch": 61.35,
648
+ "eval_loss": 0.25750261545181274,
649
+ "eval_runtime": 138.9946,
650
+ "eval_samples_per_second": 17.231,
651
+ "eval_steps_per_second": 17.231,
652
+ "eval_wer": 0.286450251381044,
653
+ "step": 10000
654
+ },
655
+ {
656
+ "epoch": 61.96,
657
+ "learning_rate": 3.0388811188811185e-05,
658
+ "loss": 0.8146,
659
+ "step": 10100
660
+ },
661
+ {
662
+ "epoch": 62.58,
663
+ "learning_rate": 2.9899300699300698e-05,
664
+ "loss": 0.8365,
665
+ "step": 10200
666
+ },
667
+ {
668
+ "epoch": 63.19,
669
+ "learning_rate": 2.9409790209790207e-05,
670
+ "loss": 0.8226,
671
+ "step": 10300
672
+ },
673
+ {
674
+ "epoch": 63.8,
675
+ "learning_rate": 2.8920279720279716e-05,
676
+ "loss": 0.7952,
677
+ "step": 10400
678
+ },
679
+ {
680
+ "epoch": 64.42,
681
+ "learning_rate": 2.8430769230769228e-05,
682
+ "loss": 0.816,
683
+ "step": 10500
684
+ },
685
+ {
686
+ "epoch": 65.03,
687
+ "learning_rate": 2.794125874125874e-05,
688
+ "loss": 0.8067,
689
+ "step": 10600
690
+ },
691
+ {
692
+ "epoch": 65.64,
693
+ "learning_rate": 2.745174825174825e-05,
694
+ "loss": 0.8081,
695
+ "step": 10700
696
+ },
697
+ {
698
+ "epoch": 66.26,
699
+ "learning_rate": 2.6962237762237758e-05,
700
+ "loss": 0.8084,
701
+ "step": 10800
702
+ },
703
+ {
704
+ "epoch": 66.87,
705
+ "learning_rate": 2.647272727272727e-05,
706
+ "loss": 0.7976,
707
+ "step": 10900
708
+ },
709
+ {
710
+ "epoch": 67.48,
711
+ "learning_rate": 2.5983216783216783e-05,
712
+ "loss": 0.7904,
713
+ "step": 11000
714
+ },
715
+ {
716
+ "epoch": 68.1,
717
+ "learning_rate": 2.5493706293706292e-05,
718
+ "loss": 0.809,
719
+ "step": 11100
720
+ },
721
+ {
722
+ "epoch": 68.71,
723
+ "learning_rate": 2.50041958041958e-05,
724
+ "loss": 0.7866,
725
+ "step": 11200
726
+ },
727
+ {
728
+ "epoch": 69.33,
729
+ "learning_rate": 2.4514685314685313e-05,
730
+ "loss": 0.795,
731
+ "step": 11300
732
+ },
733
+ {
734
+ "epoch": 69.94,
735
+ "learning_rate": 2.4025174825174825e-05,
736
+ "loss": 0.786,
737
+ "step": 11400
738
+ },
739
+ {
740
+ "epoch": 70.55,
741
+ "learning_rate": 2.3535664335664334e-05,
742
+ "loss": 0.7829,
743
+ "step": 11500
744
+ },
745
+ {
746
+ "epoch": 71.17,
747
+ "learning_rate": 2.3046153846153843e-05,
748
+ "loss": 0.7913,
749
+ "step": 11600
750
+ },
751
+ {
752
+ "epoch": 71.78,
753
+ "learning_rate": 2.2556643356643356e-05,
754
+ "loss": 0.7659,
755
+ "step": 11700
756
+ },
757
+ {
758
+ "epoch": 72.39,
759
+ "learning_rate": 2.2067132867132865e-05,
760
+ "loss": 0.7859,
761
+ "step": 11800
762
+ },
763
+ {
764
+ "epoch": 73.01,
765
+ "learning_rate": 2.1577622377622374e-05,
766
+ "loss": 0.7646,
767
+ "step": 11900
768
+ },
769
+ {
770
+ "epoch": 73.62,
771
+ "learning_rate": 2.1088111888111886e-05,
772
+ "loss": 0.7745,
773
+ "step": 12000
774
+ },
775
+ {
776
+ "epoch": 73.62,
777
+ "eval_loss": 0.2602725625038147,
778
+ "eval_runtime": 136.3746,
779
+ "eval_samples_per_second": 17.562,
780
+ "eval_steps_per_second": 17.562,
781
+ "eval_wer": 0.2816088386816461,
782
+ "step": 12000
783
+ },
784
+ {
785
+ "epoch": 74.23,
786
+ "learning_rate": 2.0598601398601398e-05,
787
+ "loss": 0.7653,
788
+ "step": 12100
789
+ },
790
+ {
791
+ "epoch": 74.85,
792
+ "learning_rate": 2.0109090909090907e-05,
793
+ "loss": 0.7789,
794
+ "step": 12200
795
+ },
796
+ {
797
+ "epoch": 75.46,
798
+ "learning_rate": 1.9619580419580416e-05,
799
+ "loss": 0.7703,
800
+ "step": 12300
801
+ },
802
+ {
803
+ "epoch": 76.07,
804
+ "learning_rate": 1.9134965034965032e-05,
805
+ "loss": 0.7523,
806
+ "step": 12400
807
+ },
808
+ {
809
+ "epoch": 76.69,
810
+ "learning_rate": 1.8645454545454544e-05,
811
+ "loss": 0.7685,
812
+ "step": 12500
813
+ },
814
+ {
815
+ "epoch": 77.3,
816
+ "learning_rate": 1.8155944055944057e-05,
817
+ "loss": 0.7612,
818
+ "step": 12600
819
+ },
820
+ {
821
+ "epoch": 77.91,
822
+ "learning_rate": 1.7666433566433566e-05,
823
+ "loss": 0.7412,
824
+ "step": 12700
825
+ },
826
+ {
827
+ "epoch": 78.53,
828
+ "learning_rate": 1.7176923076923075e-05,
829
+ "loss": 0.7564,
830
+ "step": 12800
831
+ },
832
+ {
833
+ "epoch": 79.14,
834
+ "learning_rate": 1.6687412587412587e-05,
835
+ "loss": 0.7558,
836
+ "step": 12900
837
+ },
838
+ {
839
+ "epoch": 79.75,
840
+ "learning_rate": 1.6197902097902096e-05,
841
+ "loss": 0.7379,
842
+ "step": 13000
843
+ },
844
+ {
845
+ "epoch": 80.37,
846
+ "learning_rate": 1.5708391608391608e-05,
847
+ "loss": 0.7478,
848
+ "step": 13100
849
+ },
850
+ {
851
+ "epoch": 80.98,
852
+ "learning_rate": 1.5218881118881117e-05,
853
+ "loss": 0.7369,
854
+ "step": 13200
855
+ },
856
+ {
857
+ "epoch": 81.6,
858
+ "learning_rate": 1.4729370629370628e-05,
859
+ "loss": 0.7487,
860
+ "step": 13300
861
+ },
862
+ {
863
+ "epoch": 82.21,
864
+ "learning_rate": 1.4239860139860138e-05,
865
+ "loss": 0.7312,
866
+ "step": 13400
867
+ },
868
+ {
869
+ "epoch": 82.82,
870
+ "learning_rate": 1.3750349650349649e-05,
871
+ "loss": 0.7499,
872
+ "step": 13500
873
+ },
874
+ {
875
+ "epoch": 83.44,
876
+ "learning_rate": 1.326083916083916e-05,
877
+ "loss": 0.7332,
878
+ "step": 13600
879
+ },
880
+ {
881
+ "epoch": 84.05,
882
+ "learning_rate": 1.277132867132867e-05,
883
+ "loss": 0.7389,
884
+ "step": 13700
885
+ },
886
+ {
887
+ "epoch": 84.66,
888
+ "learning_rate": 1.2281818181818181e-05,
889
+ "loss": 0.7348,
890
+ "step": 13800
891
+ },
892
+ {
893
+ "epoch": 85.28,
894
+ "learning_rate": 1.1792307692307692e-05,
895
+ "loss": 0.7241,
896
+ "step": 13900
897
+ },
898
+ {
899
+ "epoch": 85.89,
900
+ "learning_rate": 1.13027972027972e-05,
901
+ "loss": 0.7297,
902
+ "step": 14000
903
+ },
904
+ {
905
+ "epoch": 85.89,
906
+ "eval_loss": 0.25392723083496094,
907
+ "eval_runtime": 135.7135,
908
+ "eval_samples_per_second": 17.647,
909
+ "eval_steps_per_second": 17.647,
910
+ "eval_wer": 0.2727329153994165,
911
+ "step": 14000
912
+ },
913
+ {
914
+ "epoch": 86.5,
915
+ "learning_rate": 1.0813286713286713e-05,
916
+ "loss": 0.727,
917
+ "step": 14100
918
+ },
919
+ {
920
+ "epoch": 87.12,
921
+ "learning_rate": 1.0323776223776222e-05,
922
+ "loss": 0.7309,
923
+ "step": 14200
924
+ },
925
+ {
926
+ "epoch": 87.73,
927
+ "learning_rate": 9.834265734265734e-06,
928
+ "loss": 0.7133,
929
+ "step": 14300
930
+ },
931
+ {
932
+ "epoch": 88.34,
933
+ "learning_rate": 9.344755244755243e-06,
934
+ "loss": 0.7271,
935
+ "step": 14400
936
+ },
937
+ {
938
+ "epoch": 88.96,
939
+ "learning_rate": 8.855244755244755e-06,
940
+ "loss": 0.7264,
941
+ "step": 14500
942
+ },
943
+ {
944
+ "epoch": 89.57,
945
+ "learning_rate": 8.365734265734264e-06,
946
+ "loss": 0.71,
947
+ "step": 14600
948
+ },
949
+ {
950
+ "epoch": 90.18,
951
+ "learning_rate": 7.876223776223775e-06,
952
+ "loss": 0.7277,
953
+ "step": 14700
954
+ },
955
+ {
956
+ "epoch": 90.8,
957
+ "learning_rate": 7.3867132867132865e-06,
958
+ "loss": 0.7109,
959
+ "step": 14800
960
+ },
961
+ {
962
+ "epoch": 91.41,
963
+ "learning_rate": 6.897202797202796e-06,
964
+ "loss": 0.7203,
965
+ "step": 14900
966
+ },
967
+ {
968
+ "epoch": 92.02,
969
+ "learning_rate": 6.407692307692307e-06,
970
+ "loss": 0.6922,
971
+ "step": 15000
972
+ },
973
+ {
974
+ "epoch": 92.64,
975
+ "learning_rate": 5.918181818181818e-06,
976
+ "loss": 0.7343,
977
+ "step": 15100
978
+ },
979
+ {
980
+ "epoch": 93.25,
981
+ "learning_rate": 5.428671328671328e-06,
982
+ "loss": 0.71,
983
+ "step": 15200
984
+ },
985
+ {
986
+ "epoch": 93.87,
987
+ "learning_rate": 4.939160839160839e-06,
988
+ "loss": 0.7136,
989
+ "step": 15300
990
+ },
991
+ {
992
+ "epoch": 94.48,
993
+ "learning_rate": 4.4496503496503495e-06,
994
+ "loss": 0.7185,
995
+ "step": 15400
996
+ },
997
+ {
998
+ "epoch": 95.09,
999
+ "learning_rate": 3.96013986013986e-06,
1000
+ "loss": 0.7178,
1001
+ "step": 15500
1002
+ },
1003
+ {
1004
+ "epoch": 95.71,
1005
+ "learning_rate": 3.4706293706293703e-06,
1006
+ "loss": 0.7125,
1007
+ "step": 15600
1008
+ },
1009
+ {
1010
+ "epoch": 96.32,
1011
+ "learning_rate": 2.981118881118881e-06,
1012
+ "loss": 0.7081,
1013
+ "step": 15700
1014
+ },
1015
+ {
1016
+ "epoch": 96.93,
1017
+ "learning_rate": 2.4916083916083916e-06,
1018
+ "loss": 0.7087,
1019
+ "step": 15800
1020
+ },
1021
+ {
1022
+ "epoch": 97.55,
1023
+ "learning_rate": 2.002097902097902e-06,
1024
+ "loss": 0.7061,
1025
+ "step": 15900
1026
+ },
1027
+ {
1028
+ "epoch": 98.16,
1029
+ "learning_rate": 1.5174825174825173e-06,
1030
+ "loss": 0.7079,
1031
+ "step": 16000
1032
+ },
1033
+ {
1034
+ "epoch": 98.16,
1035
+ "eval_loss": 0.25544750690460205,
1036
+ "eval_runtime": 137.2236,
1037
+ "eval_samples_per_second": 17.453,
1038
+ "eval_steps_per_second": 17.453,
1039
+ "eval_wer": 0.2680777108807647,
1040
+ "step": 16000
1041
+ },
1042
+ {
1043
+ "epoch": 98.77,
1044
+ "learning_rate": 1.027972027972028e-06,
1045
+ "loss": 0.7069,
1046
+ "step": 16100
1047
+ },
1048
+ {
1049
+ "epoch": 99.39,
1050
+ "learning_rate": 5.384615384615384e-07,
1051
+ "loss": 0.7058,
1052
+ "step": 16200
1053
+ },
1054
+ {
1055
+ "epoch": 100.0,
1056
+ "learning_rate": 4.8951048951048945e-08,
1057
+ "loss": 0.7044,
1058
+ "step": 16300
1059
+ },
1060
+ {
1061
+ "epoch": 100.0,
1062
+ "step": 16300,
1063
+ "total_flos": 7.338876659637128e+19,
1064
+ "train_loss": 1.2786970950635663,
1065
+ "train_runtime": 28590.9558,
1066
+ "train_samples_per_second": 18.244,
1067
+ "train_steps_per_second": 0.57
1068
+ }
1069
+ ],
1070
+ "max_steps": 16300,
1071
+ "num_train_epochs": 100,
1072
+ "total_flos": 7.338876659637128e+19,
1073
+ "trial_name": null,
1074
+ "trial_params": null
1075
+ }