thiagobarbosa commited on
Commit
b0c0173
1 Parent(s): 12c4380

End of training

Browse files
Files changed (2) hide show
  1. README.md +9 -6
  2. trainer_state.json +1330 -0
README.md CHANGED
@@ -1,21 +1,24 @@
1
  ---
 
 
2
  license: apache-2.0
3
  base_model: openai/whisper-small
4
  tags:
 
5
  - generated_from_trainer
6
  datasets:
7
- - common_voice_16_0
8
  metrics:
9
  - wer
10
  model-index:
11
- - name: whisper-small-common-voice-16-pt
12
  results:
13
  - task:
14
  name: Automatic Speech Recognition
15
  type: automatic-speech-recognition
16
  dataset:
17
- name: common_voice_16_0
18
- type: common_voice_16_0
19
  config: pt
20
  split: test
21
  args: pt
@@ -28,9 +31,9 @@ model-index:
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
29
  should probably proofread and complete it, then remove this comment. -->
30
 
31
- # whisper-small-common-voice-16-pt
32
 
33
- This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the common_voice_16_0 dataset.
34
  It achieves the following results on the evaluation set:
35
  - Loss: 0.2712
36
  - Wer: 17.3335
 
1
  ---
2
+ language:
3
+ - pt
4
  license: apache-2.0
5
  base_model: openai/whisper-small
6
  tags:
7
+ - hf-asr-leaderboard
8
  - generated_from_trainer
9
  datasets:
10
+ - mozilla-foundation/common_voice_16_0
11
  metrics:
12
  - wer
13
  model-index:
14
+ - name: Whisper small using Common Voice 16 (pt)
15
  results:
16
  - task:
17
  name: Automatic Speech Recognition
18
  type: automatic-speech-recognition
19
  dataset:
20
+ name: Mozilla Common Voices - 16.0 - Portuguese
21
+ type: mozilla-foundation/common_voice_16_0
22
  config: pt
23
  split: test
24
  args: pt
 
31
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
32
  should probably proofread and complete it, then remove this comment. -->
33
 
34
+ # Whisper small using Common Voice 16 (pt)
35
 
36
+ This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Mozilla Common Voices - 16.0 - Portuguese dataset.
37
  It achieves the following results on the evaluation set:
38
  - Loss: 0.2712
39
  - Wer: 17.3335
trainer_state.json ADDED
@@ -0,0 +1,1330 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 17.036199095022624,
3
+ "best_model_checkpoint": "./whisper-small-common-voice-16-pt/checkpoint-4000",
4
+ "epoch": 3.7009622501850483,
5
+ "eval_steps": 500,
6
+ "global_step": 5000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "learning_rate": 4.2000000000000006e-07,
14
+ "loss": 4.9785,
15
+ "step": 25
16
+ },
17
+ {
18
+ "epoch": 0.04,
19
+ "learning_rate": 9.000000000000001e-07,
20
+ "loss": 3.8237,
21
+ "step": 50
22
+ },
23
+ {
24
+ "epoch": 0.06,
25
+ "learning_rate": 1.4000000000000001e-06,
26
+ "loss": 2.8164,
27
+ "step": 75
28
+ },
29
+ {
30
+ "epoch": 0.07,
31
+ "learning_rate": 1.9000000000000002e-06,
32
+ "loss": 1.6381,
33
+ "step": 100
34
+ },
35
+ {
36
+ "epoch": 0.09,
37
+ "learning_rate": 2.4000000000000003e-06,
38
+ "loss": 1.0913,
39
+ "step": 125
40
+ },
41
+ {
42
+ "epoch": 0.11,
43
+ "learning_rate": 2.9e-06,
44
+ "loss": 0.9218,
45
+ "step": 150
46
+ },
47
+ {
48
+ "epoch": 0.13,
49
+ "learning_rate": 3.4000000000000005e-06,
50
+ "loss": 0.8171,
51
+ "step": 175
52
+ },
53
+ {
54
+ "epoch": 0.15,
55
+ "learning_rate": 3.900000000000001e-06,
56
+ "loss": 0.7009,
57
+ "step": 200
58
+ },
59
+ {
60
+ "epoch": 0.17,
61
+ "learning_rate": 4.38e-06,
62
+ "loss": 0.6581,
63
+ "step": 225
64
+ },
65
+ {
66
+ "epoch": 0.19,
67
+ "learning_rate": 4.880000000000001e-06,
68
+ "loss": 0.4877,
69
+ "step": 250
70
+ },
71
+ {
72
+ "epoch": 0.2,
73
+ "learning_rate": 5.380000000000001e-06,
74
+ "loss": 0.3432,
75
+ "step": 275
76
+ },
77
+ {
78
+ "epoch": 0.22,
79
+ "learning_rate": 5.8800000000000005e-06,
80
+ "loss": 0.2772,
81
+ "step": 300
82
+ },
83
+ {
84
+ "epoch": 0.24,
85
+ "learning_rate": 6.380000000000001e-06,
86
+ "loss": 0.3046,
87
+ "step": 325
88
+ },
89
+ {
90
+ "epoch": 0.26,
91
+ "learning_rate": 6.88e-06,
92
+ "loss": 0.3073,
93
+ "step": 350
94
+ },
95
+ {
96
+ "epoch": 0.28,
97
+ "learning_rate": 7.3800000000000005e-06,
98
+ "loss": 0.2705,
99
+ "step": 375
100
+ },
101
+ {
102
+ "epoch": 0.3,
103
+ "learning_rate": 7.88e-06,
104
+ "loss": 0.2777,
105
+ "step": 400
106
+ },
107
+ {
108
+ "epoch": 0.31,
109
+ "learning_rate": 8.380000000000001e-06,
110
+ "loss": 0.2733,
111
+ "step": 425
112
+ },
113
+ {
114
+ "epoch": 0.33,
115
+ "learning_rate": 8.880000000000001e-06,
116
+ "loss": 0.2705,
117
+ "step": 450
118
+ },
119
+ {
120
+ "epoch": 0.35,
121
+ "learning_rate": 9.38e-06,
122
+ "loss": 0.2593,
123
+ "step": 475
124
+ },
125
+ {
126
+ "epoch": 0.37,
127
+ "learning_rate": 9.88e-06,
128
+ "loss": 0.2591,
129
+ "step": 500
130
+ },
131
+ {
132
+ "epoch": 0.37,
133
+ "eval_loss": 0.2776341736316681,
134
+ "eval_runtime": 2710.5305,
135
+ "eval_samples_per_second": 3.468,
136
+ "eval_steps_per_second": 0.433,
137
+ "eval_wer": 20.07272139625081,
138
+ "eval_wer_normalized": 13.67072759666779,
139
+ "step": 500
140
+ },
141
+ {
142
+ "epoch": 0.39,
143
+ "learning_rate": 9.957777777777779e-06,
144
+ "loss": 0.2517,
145
+ "step": 525
146
+ },
147
+ {
148
+ "epoch": 0.41,
149
+ "learning_rate": 9.902222222222223e-06,
150
+ "loss": 0.2477,
151
+ "step": 550
152
+ },
153
+ {
154
+ "epoch": 0.43,
155
+ "learning_rate": 9.846666666666668e-06,
156
+ "loss": 0.2491,
157
+ "step": 575
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "learning_rate": 9.791111111111112e-06,
162
+ "loss": 0.2574,
163
+ "step": 600
164
+ },
165
+ {
166
+ "epoch": 0.46,
167
+ "learning_rate": 9.735555555555556e-06,
168
+ "loss": 0.2692,
169
+ "step": 625
170
+ },
171
+ {
172
+ "epoch": 0.48,
173
+ "learning_rate": 9.68e-06,
174
+ "loss": 0.2729,
175
+ "step": 650
176
+ },
177
+ {
178
+ "epoch": 0.5,
179
+ "learning_rate": 9.624444444444445e-06,
180
+ "loss": 0.2401,
181
+ "step": 675
182
+ },
183
+ {
184
+ "epoch": 0.52,
185
+ "learning_rate": 9.56888888888889e-06,
186
+ "loss": 0.2345,
187
+ "step": 700
188
+ },
189
+ {
190
+ "epoch": 0.54,
191
+ "learning_rate": 9.513333333333334e-06,
192
+ "loss": 0.2319,
193
+ "step": 725
194
+ },
195
+ {
196
+ "epoch": 0.56,
197
+ "learning_rate": 9.457777777777778e-06,
198
+ "loss": 0.283,
199
+ "step": 750
200
+ },
201
+ {
202
+ "epoch": 0.57,
203
+ "learning_rate": 9.402222222222222e-06,
204
+ "loss": 0.2324,
205
+ "step": 775
206
+ },
207
+ {
208
+ "epoch": 0.59,
209
+ "learning_rate": 9.346666666666666e-06,
210
+ "loss": 0.2459,
211
+ "step": 800
212
+ },
213
+ {
214
+ "epoch": 0.61,
215
+ "learning_rate": 9.291111111111112e-06,
216
+ "loss": 0.2373,
217
+ "step": 825
218
+ },
219
+ {
220
+ "epoch": 0.63,
221
+ "learning_rate": 9.235555555555556e-06,
222
+ "loss": 0.2413,
223
+ "step": 850
224
+ },
225
+ {
226
+ "epoch": 0.65,
227
+ "learning_rate": 9.180000000000002e-06,
228
+ "loss": 0.2633,
229
+ "step": 875
230
+ },
231
+ {
232
+ "epoch": 0.67,
233
+ "learning_rate": 9.124444444444444e-06,
234
+ "loss": 0.23,
235
+ "step": 900
236
+ },
237
+ {
238
+ "epoch": 0.68,
239
+ "learning_rate": 9.06888888888889e-06,
240
+ "loss": 0.2402,
241
+ "step": 925
242
+ },
243
+ {
244
+ "epoch": 0.7,
245
+ "learning_rate": 9.013333333333334e-06,
246
+ "loss": 0.2298,
247
+ "step": 950
248
+ },
249
+ {
250
+ "epoch": 0.72,
251
+ "learning_rate": 8.957777777777778e-06,
252
+ "loss": 0.205,
253
+ "step": 975
254
+ },
255
+ {
256
+ "epoch": 0.74,
257
+ "learning_rate": 8.902222222222224e-06,
258
+ "loss": 0.2626,
259
+ "step": 1000
260
+ },
261
+ {
262
+ "epoch": 0.74,
263
+ "eval_loss": 0.2599042057991028,
264
+ "eval_runtime": 2724.1583,
265
+ "eval_samples_per_second": 3.45,
266
+ "eval_steps_per_second": 0.431,
267
+ "eval_wer": 19.400452488687783,
268
+ "eval_wer_normalized": 13.333654355467809,
269
+ "step": 1000
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "learning_rate": 8.846666666666668e-06,
274
+ "loss": 0.2616,
275
+ "step": 1025
276
+ },
277
+ {
278
+ "epoch": 0.78,
279
+ "learning_rate": 8.791111111111112e-06,
280
+ "loss": 0.2532,
281
+ "step": 1050
282
+ },
283
+ {
284
+ "epoch": 0.8,
285
+ "learning_rate": 8.735555555555556e-06,
286
+ "loss": 0.2405,
287
+ "step": 1075
288
+ },
289
+ {
290
+ "epoch": 0.81,
291
+ "learning_rate": 8.68e-06,
292
+ "loss": 0.2162,
293
+ "step": 1100
294
+ },
295
+ {
296
+ "epoch": 0.83,
297
+ "learning_rate": 8.624444444444446e-06,
298
+ "loss": 0.2431,
299
+ "step": 1125
300
+ },
301
+ {
302
+ "epoch": 0.85,
303
+ "learning_rate": 8.56888888888889e-06,
304
+ "loss": 0.2352,
305
+ "step": 1150
306
+ },
307
+ {
308
+ "epoch": 0.87,
309
+ "learning_rate": 8.513333333333335e-06,
310
+ "loss": 0.2682,
311
+ "step": 1175
312
+ },
313
+ {
314
+ "epoch": 0.89,
315
+ "learning_rate": 8.457777777777778e-06,
316
+ "loss": 0.2161,
317
+ "step": 1200
318
+ },
319
+ {
320
+ "epoch": 0.91,
321
+ "learning_rate": 8.402222222222223e-06,
322
+ "loss": 0.2299,
323
+ "step": 1225
324
+ },
325
+ {
326
+ "epoch": 0.93,
327
+ "learning_rate": 8.346666666666668e-06,
328
+ "loss": 0.246,
329
+ "step": 1250
330
+ },
331
+ {
332
+ "epoch": 0.94,
333
+ "learning_rate": 8.291111111111112e-06,
334
+ "loss": 0.2518,
335
+ "step": 1275
336
+ },
337
+ {
338
+ "epoch": 0.96,
339
+ "learning_rate": 8.235555555555557e-06,
340
+ "loss": 0.2467,
341
+ "step": 1300
342
+ },
343
+ {
344
+ "epoch": 0.98,
345
+ "learning_rate": 8.18e-06,
346
+ "loss": 0.2147,
347
+ "step": 1325
348
+ },
349
+ {
350
+ "epoch": 1.0,
351
+ "learning_rate": 8.124444444444445e-06,
352
+ "loss": 0.2452,
353
+ "step": 1350
354
+ },
355
+ {
356
+ "epoch": 1.02,
357
+ "learning_rate": 8.06888888888889e-06,
358
+ "loss": 0.1274,
359
+ "step": 1375
360
+ },
361
+ {
362
+ "epoch": 1.04,
363
+ "learning_rate": 8.013333333333333e-06,
364
+ "loss": 0.1139,
365
+ "step": 1400
366
+ },
367
+ {
368
+ "epoch": 1.05,
369
+ "learning_rate": 7.957777777777779e-06,
370
+ "loss": 0.1065,
371
+ "step": 1425
372
+ },
373
+ {
374
+ "epoch": 1.07,
375
+ "learning_rate": 7.902222222222223e-06,
376
+ "loss": 0.1072,
377
+ "step": 1450
378
+ },
379
+ {
380
+ "epoch": 1.09,
381
+ "learning_rate": 7.846666666666667e-06,
382
+ "loss": 0.1226,
383
+ "step": 1475
384
+ },
385
+ {
386
+ "epoch": 1.11,
387
+ "learning_rate": 7.791111111111111e-06,
388
+ "loss": 0.1131,
389
+ "step": 1500
390
+ },
391
+ {
392
+ "epoch": 1.11,
393
+ "eval_loss": 0.2515513598918915,
394
+ "eval_runtime": 2716.9869,
395
+ "eval_samples_per_second": 3.459,
396
+ "eval_steps_per_second": 0.432,
397
+ "eval_wer": 18.04137039431157,
398
+ "eval_wer_normalized": 12.133031572526926,
399
+ "step": 1500
400
+ },
401
+ {
402
+ "epoch": 1.13,
403
+ "learning_rate": 7.735555555555557e-06,
404
+ "loss": 0.1119,
405
+ "step": 1525
406
+ },
407
+ {
408
+ "epoch": 1.15,
409
+ "learning_rate": 7.680000000000001e-06,
410
+ "loss": 0.1057,
411
+ "step": 1550
412
+ },
413
+ {
414
+ "epoch": 1.17,
415
+ "learning_rate": 7.624444444444445e-06,
416
+ "loss": 0.1145,
417
+ "step": 1575
418
+ },
419
+ {
420
+ "epoch": 1.18,
421
+ "learning_rate": 7.56888888888889e-06,
422
+ "loss": 0.1256,
423
+ "step": 1600
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "learning_rate": 7.513333333333334e-06,
428
+ "loss": 0.1126,
429
+ "step": 1625
430
+ },
431
+ {
432
+ "epoch": 1.22,
433
+ "learning_rate": 7.457777777777778e-06,
434
+ "loss": 0.1097,
435
+ "step": 1650
436
+ },
437
+ {
438
+ "epoch": 1.24,
439
+ "learning_rate": 7.402222222222223e-06,
440
+ "loss": 0.1052,
441
+ "step": 1675
442
+ },
443
+ {
444
+ "epoch": 1.26,
445
+ "learning_rate": 7.346666666666668e-06,
446
+ "loss": 0.1293,
447
+ "step": 1700
448
+ },
449
+ {
450
+ "epoch": 1.28,
451
+ "learning_rate": 7.291111111111112e-06,
452
+ "loss": 0.1116,
453
+ "step": 1725
454
+ },
455
+ {
456
+ "epoch": 1.3,
457
+ "learning_rate": 7.235555555555556e-06,
458
+ "loss": 0.1174,
459
+ "step": 1750
460
+ },
461
+ {
462
+ "epoch": 1.31,
463
+ "learning_rate": 7.180000000000001e-06,
464
+ "loss": 0.1105,
465
+ "step": 1775
466
+ },
467
+ {
468
+ "epoch": 1.33,
469
+ "learning_rate": 7.124444444444445e-06,
470
+ "loss": 0.1221,
471
+ "step": 1800
472
+ },
473
+ {
474
+ "epoch": 1.35,
475
+ "learning_rate": 7.06888888888889e-06,
476
+ "loss": 0.1139,
477
+ "step": 1825
478
+ },
479
+ {
480
+ "epoch": 1.37,
481
+ "learning_rate": 7.0133333333333345e-06,
482
+ "loss": 0.1201,
483
+ "step": 1850
484
+ },
485
+ {
486
+ "epoch": 1.39,
487
+ "learning_rate": 6.9577777777777785e-06,
488
+ "loss": 0.1211,
489
+ "step": 1875
490
+ },
491
+ {
492
+ "epoch": 1.41,
493
+ "learning_rate": 6.902222222222223e-06,
494
+ "loss": 0.1192,
495
+ "step": 1900
496
+ },
497
+ {
498
+ "epoch": 1.42,
499
+ "learning_rate": 6.846666666666667e-06,
500
+ "loss": 0.1133,
501
+ "step": 1925
502
+ },
503
+ {
504
+ "epoch": 1.44,
505
+ "learning_rate": 6.7911111111111115e-06,
506
+ "loss": 0.1013,
507
+ "step": 1950
508
+ },
509
+ {
510
+ "epoch": 1.46,
511
+ "learning_rate": 6.735555555555556e-06,
512
+ "loss": 0.1074,
513
+ "step": 1975
514
+ },
515
+ {
516
+ "epoch": 1.48,
517
+ "learning_rate": 6.680000000000001e-06,
518
+ "loss": 0.1016,
519
+ "step": 2000
520
+ },
521
+ {
522
+ "epoch": 1.48,
523
+ "eval_loss": 0.24819953739643097,
524
+ "eval_runtime": 2711.2399,
525
+ "eval_samples_per_second": 3.467,
526
+ "eval_steps_per_second": 0.433,
527
+ "eval_wer": 18.35972850678733,
528
+ "eval_wer_normalized": 11.924367185117415,
529
+ "step": 2000
530
+ },
531
+ {
532
+ "epoch": 1.5,
533
+ "learning_rate": 6.6244444444444445e-06,
534
+ "loss": 0.1255,
535
+ "step": 2025
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "learning_rate": 6.568888888888889e-06,
540
+ "loss": 0.1113,
541
+ "step": 2050
542
+ },
543
+ {
544
+ "epoch": 1.54,
545
+ "learning_rate": 6.513333333333333e-06,
546
+ "loss": 0.11,
547
+ "step": 2075
548
+ },
549
+ {
550
+ "epoch": 1.55,
551
+ "learning_rate": 6.457777777777778e-06,
552
+ "loss": 0.1096,
553
+ "step": 2100
554
+ },
555
+ {
556
+ "epoch": 1.57,
557
+ "learning_rate": 6.402222222222223e-06,
558
+ "loss": 0.11,
559
+ "step": 2125
560
+ },
561
+ {
562
+ "epoch": 1.59,
563
+ "learning_rate": 6.346666666666668e-06,
564
+ "loss": 0.12,
565
+ "step": 2150
566
+ },
567
+ {
568
+ "epoch": 1.61,
569
+ "learning_rate": 6.291111111111111e-06,
570
+ "loss": 0.1101,
571
+ "step": 2175
572
+ },
573
+ {
574
+ "epoch": 1.63,
575
+ "learning_rate": 6.235555555555556e-06,
576
+ "loss": 0.0962,
577
+ "step": 2200
578
+ },
579
+ {
580
+ "epoch": 1.65,
581
+ "learning_rate": 6.18e-06,
582
+ "loss": 0.104,
583
+ "step": 2225
584
+ },
585
+ {
586
+ "epoch": 1.67,
587
+ "learning_rate": 6.124444444444445e-06,
588
+ "loss": 0.0933,
589
+ "step": 2250
590
+ },
591
+ {
592
+ "epoch": 1.68,
593
+ "learning_rate": 6.06888888888889e-06,
594
+ "loss": 0.1207,
595
+ "step": 2275
596
+ },
597
+ {
598
+ "epoch": 1.7,
599
+ "learning_rate": 6.013333333333335e-06,
600
+ "loss": 0.111,
601
+ "step": 2300
602
+ },
603
+ {
604
+ "epoch": 1.72,
605
+ "learning_rate": 5.957777777777778e-06,
606
+ "loss": 0.1154,
607
+ "step": 2325
608
+ },
609
+ {
610
+ "epoch": 1.74,
611
+ "learning_rate": 5.902222222222223e-06,
612
+ "loss": 0.1024,
613
+ "step": 2350
614
+ },
615
+ {
616
+ "epoch": 1.76,
617
+ "learning_rate": 5.846666666666667e-06,
618
+ "loss": 0.1036,
619
+ "step": 2375
620
+ },
621
+ {
622
+ "epoch": 1.78,
623
+ "learning_rate": 5.791111111111112e-06,
624
+ "loss": 0.1235,
625
+ "step": 2400
626
+ },
627
+ {
628
+ "epoch": 1.79,
629
+ "learning_rate": 5.735555555555557e-06,
630
+ "loss": 0.1019,
631
+ "step": 2425
632
+ },
633
+ {
634
+ "epoch": 1.81,
635
+ "learning_rate": 5.68e-06,
636
+ "loss": 0.1182,
637
+ "step": 2450
638
+ },
639
+ {
640
+ "epoch": 1.83,
641
+ "learning_rate": 5.624444444444445e-06,
642
+ "loss": 0.1065,
643
+ "step": 2475
644
+ },
645
+ {
646
+ "epoch": 1.85,
647
+ "learning_rate": 5.56888888888889e-06,
648
+ "loss": 0.1094,
649
+ "step": 2500
650
+ },
651
+ {
652
+ "epoch": 1.85,
653
+ "eval_loss": 0.24113476276397705,
654
+ "eval_runtime": 2716.3575,
655
+ "eval_samples_per_second": 3.46,
656
+ "eval_steps_per_second": 0.433,
657
+ "eval_wer": 17.419198448610214,
658
+ "eval_wer_normalized": 11.60173993996886,
659
+ "step": 2500
660
+ },
661
+ {
662
+ "epoch": 1.87,
663
+ "learning_rate": 5.513333333333334e-06,
664
+ "loss": 0.094,
665
+ "step": 2525
666
+ },
667
+ {
668
+ "epoch": 1.89,
669
+ "learning_rate": 5.4577777777777785e-06,
670
+ "loss": 0.1107,
671
+ "step": 2550
672
+ },
673
+ {
674
+ "epoch": 1.91,
675
+ "learning_rate": 5.402222222222223e-06,
676
+ "loss": 0.1221,
677
+ "step": 2575
678
+ },
679
+ {
680
+ "epoch": 1.92,
681
+ "learning_rate": 5.346666666666667e-06,
682
+ "loss": 0.1111,
683
+ "step": 2600
684
+ },
685
+ {
686
+ "epoch": 1.94,
687
+ "learning_rate": 5.2911111111111115e-06,
688
+ "loss": 0.108,
689
+ "step": 2625
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "learning_rate": 5.235555555555556e-06,
694
+ "loss": 0.114,
695
+ "step": 2650
696
+ },
697
+ {
698
+ "epoch": 1.98,
699
+ "learning_rate": 5.18e-06,
700
+ "loss": 0.1039,
701
+ "step": 2675
702
+ },
703
+ {
704
+ "epoch": 2.0,
705
+ "learning_rate": 5.124444444444445e-06,
706
+ "loss": 0.1123,
707
+ "step": 2700
708
+ },
709
+ {
710
+ "epoch": 2.02,
711
+ "learning_rate": 5.06888888888889e-06,
712
+ "loss": 0.0625,
713
+ "step": 2725
714
+ },
715
+ {
716
+ "epoch": 2.04,
717
+ "learning_rate": 5.013333333333333e-06,
718
+ "loss": 0.0488,
719
+ "step": 2750
720
+ },
721
+ {
722
+ "epoch": 2.05,
723
+ "learning_rate": 4.957777777777778e-06,
724
+ "loss": 0.0493,
725
+ "step": 2775
726
+ },
727
+ {
728
+ "epoch": 2.07,
729
+ "learning_rate": 4.902222222222222e-06,
730
+ "loss": 0.0465,
731
+ "step": 2800
732
+ },
733
+ {
734
+ "epoch": 2.09,
735
+ "learning_rate": 4.846666666666667e-06,
736
+ "loss": 0.0451,
737
+ "step": 2825
738
+ },
739
+ {
740
+ "epoch": 2.11,
741
+ "learning_rate": 4.791111111111111e-06,
742
+ "loss": 0.0436,
743
+ "step": 2850
744
+ },
745
+ {
746
+ "epoch": 2.13,
747
+ "learning_rate": 4.735555555555556e-06,
748
+ "loss": 0.0606,
749
+ "step": 2875
750
+ },
751
+ {
752
+ "epoch": 2.15,
753
+ "learning_rate": 4.680000000000001e-06,
754
+ "loss": 0.0505,
755
+ "step": 2900
756
+ },
757
+ {
758
+ "epoch": 2.17,
759
+ "learning_rate": 4.624444444444445e-06,
760
+ "loss": 0.0562,
761
+ "step": 2925
762
+ },
763
+ {
764
+ "epoch": 2.18,
765
+ "learning_rate": 4.568888888888889e-06,
766
+ "loss": 0.051,
767
+ "step": 2950
768
+ },
769
+ {
770
+ "epoch": 2.2,
771
+ "learning_rate": 4.513333333333333e-06,
772
+ "loss": 0.0407,
773
+ "step": 2975
774
+ },
775
+ {
776
+ "epoch": 2.22,
777
+ "learning_rate": 4.457777777777778e-06,
778
+ "loss": 0.0524,
779
+ "step": 3000
780
+ },
781
+ {
782
+ "epoch": 2.22,
783
+ "eval_loss": 0.25118857622146606,
784
+ "eval_runtime": 2728.2785,
785
+ "eval_samples_per_second": 3.445,
786
+ "eval_steps_per_second": 0.431,
787
+ "eval_wer": 17.354557207498384,
788
+ "eval_wer_normalized": 11.463700422144107,
789
+ "step": 3000
790
+ },
791
+ {
792
+ "epoch": 2.24,
793
+ "learning_rate": 4.402222222222223e-06,
794
+ "loss": 0.0478,
795
+ "step": 3025
796
+ },
797
+ {
798
+ "epoch": 2.26,
799
+ "learning_rate": 4.346666666666667e-06,
800
+ "loss": 0.0487,
801
+ "step": 3050
802
+ },
803
+ {
804
+ "epoch": 2.28,
805
+ "learning_rate": 4.291111111111112e-06,
806
+ "loss": 0.046,
807
+ "step": 3075
808
+ },
809
+ {
810
+ "epoch": 2.29,
811
+ "learning_rate": 4.235555555555556e-06,
812
+ "loss": 0.0425,
813
+ "step": 3100
814
+ },
815
+ {
816
+ "epoch": 2.31,
817
+ "learning_rate": 4.18e-06,
818
+ "loss": 0.0428,
819
+ "step": 3125
820
+ },
821
+ {
822
+ "epoch": 2.33,
823
+ "learning_rate": 4.124444444444445e-06,
824
+ "loss": 0.0455,
825
+ "step": 3150
826
+ },
827
+ {
828
+ "epoch": 2.35,
829
+ "learning_rate": 4.0688888888888896e-06,
830
+ "loss": 0.0482,
831
+ "step": 3175
832
+ },
833
+ {
834
+ "epoch": 2.37,
835
+ "learning_rate": 4.013333333333334e-06,
836
+ "loss": 0.0428,
837
+ "step": 3200
838
+ },
839
+ {
840
+ "epoch": 2.39,
841
+ "learning_rate": 3.9577777777777785e-06,
842
+ "loss": 0.0501,
843
+ "step": 3225
844
+ },
845
+ {
846
+ "epoch": 2.41,
847
+ "learning_rate": 3.9022222222222225e-06,
848
+ "loss": 0.0486,
849
+ "step": 3250
850
+ },
851
+ {
852
+ "epoch": 2.42,
853
+ "learning_rate": 3.8466666666666665e-06,
854
+ "loss": 0.0554,
855
+ "step": 3275
856
+ },
857
+ {
858
+ "epoch": 2.44,
859
+ "learning_rate": 3.7911111111111114e-06,
860
+ "loss": 0.0443,
861
+ "step": 3300
862
+ },
863
+ {
864
+ "epoch": 2.46,
865
+ "learning_rate": 3.7355555555555555e-06,
866
+ "loss": 0.0536,
867
+ "step": 3325
868
+ },
869
+ {
870
+ "epoch": 2.48,
871
+ "learning_rate": 3.6800000000000003e-06,
872
+ "loss": 0.0562,
873
+ "step": 3350
874
+ },
875
+ {
876
+ "epoch": 2.5,
877
+ "learning_rate": 3.624444444444445e-06,
878
+ "loss": 0.0541,
879
+ "step": 3375
880
+ },
881
+ {
882
+ "epoch": 2.52,
883
+ "learning_rate": 3.568888888888889e-06,
884
+ "loss": 0.0472,
885
+ "step": 3400
886
+ },
887
+ {
888
+ "epoch": 2.54,
889
+ "learning_rate": 3.5133333333333337e-06,
890
+ "loss": 0.0449,
891
+ "step": 3425
892
+ },
893
+ {
894
+ "epoch": 2.55,
895
+ "learning_rate": 3.457777777777778e-06,
896
+ "loss": 0.0424,
897
+ "step": 3450
898
+ },
899
+ {
900
+ "epoch": 2.57,
901
+ "learning_rate": 3.4022222222222222e-06,
902
+ "loss": 0.0478,
903
+ "step": 3475
904
+ },
905
+ {
906
+ "epoch": 2.59,
907
+ "learning_rate": 3.346666666666667e-06,
908
+ "loss": 0.0433,
909
+ "step": 3500
910
+ },
911
+ {
912
+ "epoch": 2.59,
913
+ "eval_loss": 0.24957023561000824,
914
+ "eval_runtime": 2729.044,
915
+ "eval_samples_per_second": 3.444,
916
+ "eval_steps_per_second": 0.431,
917
+ "eval_wer": 17.089528118939885,
918
+ "eval_wer_normalized": 11.298374022888877,
919
+ "step": 3500
920
+ },
921
+ {
922
+ "epoch": 2.61,
923
+ "learning_rate": 3.2911111111111116e-06,
924
+ "loss": 0.0438,
925
+ "step": 3525
926
+ },
927
+ {
928
+ "epoch": 2.63,
929
+ "learning_rate": 3.2355555555555556e-06,
930
+ "loss": 0.0501,
931
+ "step": 3550
932
+ },
933
+ {
934
+ "epoch": 2.65,
935
+ "learning_rate": 3.1800000000000005e-06,
936
+ "loss": 0.0532,
937
+ "step": 3575
938
+ },
939
+ {
940
+ "epoch": 2.66,
941
+ "learning_rate": 3.124444444444445e-06,
942
+ "loss": 0.0478,
943
+ "step": 3600
944
+ },
945
+ {
946
+ "epoch": 2.68,
947
+ "learning_rate": 3.068888888888889e-06,
948
+ "loss": 0.0477,
949
+ "step": 3625
950
+ },
951
+ {
952
+ "epoch": 2.7,
953
+ "learning_rate": 3.013333333333334e-06,
954
+ "loss": 0.0545,
955
+ "step": 3650
956
+ },
957
+ {
958
+ "epoch": 2.72,
959
+ "learning_rate": 2.957777777777778e-06,
960
+ "loss": 0.0532,
961
+ "step": 3675
962
+ },
963
+ {
964
+ "epoch": 2.74,
965
+ "learning_rate": 2.9022222222222223e-06,
966
+ "loss": 0.0441,
967
+ "step": 3700
968
+ },
969
+ {
970
+ "epoch": 2.76,
971
+ "learning_rate": 2.8466666666666672e-06,
972
+ "loss": 0.0473,
973
+ "step": 3725
974
+ },
975
+ {
976
+ "epoch": 2.78,
977
+ "learning_rate": 2.7911111111111113e-06,
978
+ "loss": 0.0473,
979
+ "step": 3750
980
+ },
981
+ {
982
+ "epoch": 2.79,
983
+ "learning_rate": 2.7355555555555557e-06,
984
+ "loss": 0.049,
985
+ "step": 3775
986
+ },
987
+ {
988
+ "epoch": 2.81,
989
+ "learning_rate": 2.68e-06,
990
+ "loss": 0.044,
991
+ "step": 3800
992
+ },
993
+ {
994
+ "epoch": 2.83,
995
+ "learning_rate": 2.6244444444444446e-06,
996
+ "loss": 0.0484,
997
+ "step": 3825
998
+ },
999
+ {
1000
+ "epoch": 2.85,
1001
+ "learning_rate": 2.568888888888889e-06,
1002
+ "loss": 0.0446,
1003
+ "step": 3850
1004
+ },
1005
+ {
1006
+ "epoch": 2.87,
1007
+ "learning_rate": 2.5133333333333336e-06,
1008
+ "loss": 0.0479,
1009
+ "step": 3875
1010
+ },
1011
+ {
1012
+ "epoch": 2.89,
1013
+ "learning_rate": 2.457777777777778e-06,
1014
+ "loss": 0.045,
1015
+ "step": 3900
1016
+ },
1017
+ {
1018
+ "epoch": 2.91,
1019
+ "learning_rate": 2.4022222222222225e-06,
1020
+ "loss": 0.0535,
1021
+ "step": 3925
1022
+ },
1023
+ {
1024
+ "epoch": 2.92,
1025
+ "learning_rate": 2.346666666666667e-06,
1026
+ "loss": 0.0459,
1027
+ "step": 3950
1028
+ },
1029
+ {
1030
+ "epoch": 2.94,
1031
+ "learning_rate": 2.2911111111111114e-06,
1032
+ "loss": 0.0515,
1033
+ "step": 3975
1034
+ },
1035
+ {
1036
+ "epoch": 2.96,
1037
+ "learning_rate": 2.235555555555556e-06,
1038
+ "loss": 0.0453,
1039
+ "step": 4000
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "eval_loss": 0.24788813292980194,
1044
+ "eval_runtime": 2717.3489,
1045
+ "eval_samples_per_second": 3.459,
1046
+ "eval_steps_per_second": 0.432,
1047
+ "eval_wer": 17.036199095022624,
1048
+ "eval_wer_normalized": 11.267876920113643,
1049
+ "step": 4000
1050
+ },
1051
+ {
1052
+ "epoch": 2.98,
1053
+ "learning_rate": 2.1800000000000003e-06,
1054
+ "loss": 0.0497,
1055
+ "step": 4025
1056
+ },
1057
+ {
1058
+ "epoch": 3.0,
1059
+ "learning_rate": 2.1244444444444443e-06,
1060
+ "loss": 0.0448,
1061
+ "step": 4050
1062
+ },
1063
+ {
1064
+ "epoch": 3.02,
1065
+ "learning_rate": 2.0688888888888892e-06,
1066
+ "loss": 0.0283,
1067
+ "step": 4075
1068
+ },
1069
+ {
1070
+ "epoch": 3.03,
1071
+ "learning_rate": 2.0133333333333337e-06,
1072
+ "loss": 0.0222,
1073
+ "step": 4100
1074
+ },
1075
+ {
1076
+ "epoch": 3.05,
1077
+ "learning_rate": 1.9577777777777777e-06,
1078
+ "loss": 0.0215,
1079
+ "step": 4125
1080
+ },
1081
+ {
1082
+ "epoch": 3.07,
1083
+ "learning_rate": 1.9022222222222222e-06,
1084
+ "loss": 0.0248,
1085
+ "step": 4150
1086
+ },
1087
+ {
1088
+ "epoch": 3.09,
1089
+ "learning_rate": 1.8466666666666668e-06,
1090
+ "loss": 0.0227,
1091
+ "step": 4175
1092
+ },
1093
+ {
1094
+ "epoch": 3.11,
1095
+ "learning_rate": 1.7911111111111113e-06,
1096
+ "loss": 0.0241,
1097
+ "step": 4200
1098
+ },
1099
+ {
1100
+ "epoch": 3.13,
1101
+ "learning_rate": 1.7355555555555555e-06,
1102
+ "loss": 0.0204,
1103
+ "step": 4225
1104
+ },
1105
+ {
1106
+ "epoch": 3.15,
1107
+ "learning_rate": 1.6800000000000002e-06,
1108
+ "loss": 0.0178,
1109
+ "step": 4250
1110
+ },
1111
+ {
1112
+ "epoch": 3.16,
1113
+ "learning_rate": 1.6244444444444447e-06,
1114
+ "loss": 0.0182,
1115
+ "step": 4275
1116
+ },
1117
+ {
1118
+ "epoch": 3.18,
1119
+ "learning_rate": 1.568888888888889e-06,
1120
+ "loss": 0.0217,
1121
+ "step": 4300
1122
+ },
1123
+ {
1124
+ "epoch": 3.2,
1125
+ "learning_rate": 1.5133333333333334e-06,
1126
+ "loss": 0.0222,
1127
+ "step": 4325
1128
+ },
1129
+ {
1130
+ "epoch": 3.22,
1131
+ "learning_rate": 1.457777777777778e-06,
1132
+ "loss": 0.022,
1133
+ "step": 4350
1134
+ },
1135
+ {
1136
+ "epoch": 3.24,
1137
+ "learning_rate": 1.4022222222222223e-06,
1138
+ "loss": 0.018,
1139
+ "step": 4375
1140
+ },
1141
+ {
1142
+ "epoch": 3.26,
1143
+ "learning_rate": 1.3466666666666668e-06,
1144
+ "loss": 0.0211,
1145
+ "step": 4400
1146
+ },
1147
+ {
1148
+ "epoch": 3.28,
1149
+ "learning_rate": 1.2911111111111112e-06,
1150
+ "loss": 0.0215,
1151
+ "step": 4425
1152
+ },
1153
+ {
1154
+ "epoch": 3.29,
1155
+ "learning_rate": 1.2355555555555557e-06,
1156
+ "loss": 0.0197,
1157
+ "step": 4450
1158
+ },
1159
+ {
1160
+ "epoch": 3.31,
1161
+ "learning_rate": 1.1800000000000001e-06,
1162
+ "loss": 0.0227,
1163
+ "step": 4475
1164
+ },
1165
+ {
1166
+ "epoch": 3.33,
1167
+ "learning_rate": 1.1244444444444446e-06,
1168
+ "loss": 0.0201,
1169
+ "step": 4500
1170
+ },
1171
+ {
1172
+ "epoch": 3.33,
1173
+ "eval_loss": 0.269341379404068,
1174
+ "eval_runtime": 2723.6438,
1175
+ "eval_samples_per_second": 3.451,
1176
+ "eval_steps_per_second": 0.431,
1177
+ "eval_wer": 17.66321913380737,
1178
+ "eval_wer_normalized": 11.71088746569076,
1179
+ "step": 4500
1180
+ },
1181
+ {
1182
+ "epoch": 3.35,
1183
+ "learning_rate": 1.068888888888889e-06,
1184
+ "loss": 0.0198,
1185
+ "step": 4525
1186
+ },
1187
+ {
1188
+ "epoch": 3.37,
1189
+ "learning_rate": 1.0133333333333333e-06,
1190
+ "loss": 0.021,
1191
+ "step": 4550
1192
+ },
1193
+ {
1194
+ "epoch": 3.39,
1195
+ "learning_rate": 9.57777777777778e-07,
1196
+ "loss": 0.0221,
1197
+ "step": 4575
1198
+ },
1199
+ {
1200
+ "epoch": 3.4,
1201
+ "learning_rate": 9.022222222222222e-07,
1202
+ "loss": 0.0191,
1203
+ "step": 4600
1204
+ },
1205
+ {
1206
+ "epoch": 3.42,
1207
+ "learning_rate": 8.466666666666668e-07,
1208
+ "loss": 0.0198,
1209
+ "step": 4625
1210
+ },
1211
+ {
1212
+ "epoch": 3.44,
1213
+ "learning_rate": 7.911111111111111e-07,
1214
+ "loss": 0.0191,
1215
+ "step": 4650
1216
+ },
1217
+ {
1218
+ "epoch": 3.46,
1219
+ "learning_rate": 7.355555555555556e-07,
1220
+ "loss": 0.02,
1221
+ "step": 4675
1222
+ },
1223
+ {
1224
+ "epoch": 3.48,
1225
+ "learning_rate": 6.800000000000001e-07,
1226
+ "loss": 0.0253,
1227
+ "step": 4700
1228
+ },
1229
+ {
1230
+ "epoch": 3.5,
1231
+ "learning_rate": 6.244444444444445e-07,
1232
+ "loss": 0.0237,
1233
+ "step": 4725
1234
+ },
1235
+ {
1236
+ "epoch": 3.52,
1237
+ "learning_rate": 5.68888888888889e-07,
1238
+ "loss": 0.0162,
1239
+ "step": 4750
1240
+ },
1241
+ {
1242
+ "epoch": 3.53,
1243
+ "learning_rate": 5.133333333333334e-07,
1244
+ "loss": 0.0191,
1245
+ "step": 4775
1246
+ },
1247
+ {
1248
+ "epoch": 3.55,
1249
+ "learning_rate": 4.5777777777777784e-07,
1250
+ "loss": 0.0221,
1251
+ "step": 4800
1252
+ },
1253
+ {
1254
+ "epoch": 3.57,
1255
+ "learning_rate": 4.0222222222222224e-07,
1256
+ "loss": 0.0225,
1257
+ "step": 4825
1258
+ },
1259
+ {
1260
+ "epoch": 3.59,
1261
+ "learning_rate": 3.466666666666667e-07,
1262
+ "loss": 0.0216,
1263
+ "step": 4850
1264
+ },
1265
+ {
1266
+ "epoch": 3.61,
1267
+ "learning_rate": 2.9111111111111116e-07,
1268
+ "loss": 0.0203,
1269
+ "step": 4875
1270
+ },
1271
+ {
1272
+ "epoch": 3.63,
1273
+ "learning_rate": 2.3555555555555556e-07,
1274
+ "loss": 0.0216,
1275
+ "step": 4900
1276
+ },
1277
+ {
1278
+ "epoch": 3.65,
1279
+ "learning_rate": 1.8e-07,
1280
+ "loss": 0.0202,
1281
+ "step": 4925
1282
+ },
1283
+ {
1284
+ "epoch": 3.66,
1285
+ "learning_rate": 1.2444444444444446e-07,
1286
+ "loss": 0.0185,
1287
+ "step": 4950
1288
+ },
1289
+ {
1290
+ "epoch": 3.68,
1291
+ "learning_rate": 6.888888888888889e-08,
1292
+ "loss": 0.0195,
1293
+ "step": 4975
1294
+ },
1295
+ {
1296
+ "epoch": 3.7,
1297
+ "learning_rate": 1.3333333333333334e-08,
1298
+ "loss": 0.0229,
1299
+ "step": 5000
1300
+ },
1301
+ {
1302
+ "epoch": 3.7,
1303
+ "eval_loss": 0.2711798846721649,
1304
+ "eval_runtime": 2712.8452,
1305
+ "eval_samples_per_second": 3.465,
1306
+ "eval_steps_per_second": 0.433,
1307
+ "eval_wer": 17.33354880413704,
1308
+ "eval_wer_normalized": 11.332081347008877,
1309
+ "step": 5000
1310
+ },
1311
+ {
1312
+ "epoch": 3.7,
1313
+ "step": 5000,
1314
+ "total_flos": 2.307990597599232e+19,
1315
+ "train_loss": 0.1926749886929989,
1316
+ "train_runtime": 37692.9225,
1317
+ "train_samples_per_second": 2.122,
1318
+ "train_steps_per_second": 0.133
1319
+ }
1320
+ ],
1321
+ "logging_steps": 25,
1322
+ "max_steps": 5000,
1323
+ "num_input_tokens_seen": 0,
1324
+ "num_train_epochs": 4,
1325
+ "save_steps": 500,
1326
+ "total_flos": 2.307990597599232e+19,
1327
+ "train_batch_size": 16,
1328
+ "trial_name": null,
1329
+ "trial_params": null
1330
+ }