zuazo commited on
Commit
b5d5b4e
1 Parent(s): 5953230

End of training

Browse files
Files changed (5) hide show
  1. README.md +9 -6
  2. all_results.json +12 -0
  3. eval_results.json +8 -0
  4. train_results.json +7 -0
  5. trainer_state.json +1273 -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_13_0
8
  metrics:
9
  - wer
10
  model-index:
11
- - name: openai/whisper-small
12
  results:
13
  - task:
14
  name: Automatic Speech Recognition
15
  type: automatic-speech-recognition
16
  dataset:
17
- name: common_voice_13_0
18
- type: common_voice_13_0
19
  config: gl
20
  split: test
21
  args: gl
@@ -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
- # openai/whisper-small
32
 
33
- This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the common_voice_13_0 dataset.
34
  It achieves the following results on the evaluation set:
35
  - Loss: 0.3555
36
  - Wer: 10.9875
 
1
  ---
2
+ language:
3
+ - gl
4
  license: apache-2.0
5
  base_model: openai/whisper-small
6
  tags:
7
+ - whisper-event
8
  - generated_from_trainer
9
  datasets:
10
+ - mozilla-foundation/common_voice_13_0
11
  metrics:
12
  - wer
13
  model-index:
14
+ - name: Whisper Small Galician
15
  results:
16
  - task:
17
  name: Automatic Speech Recognition
18
  type: automatic-speech-recognition
19
  dataset:
20
+ name: mozilla-foundation/common_voice_13_0 gl
21
+ type: mozilla-foundation/common_voice_13_0
22
  config: gl
23
  split: test
24
  args: gl
 
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 Galician
35
 
36
+ This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the mozilla-foundation/common_voice_13_0 gl dataset.
37
  It achieves the following results on the evaluation set:
38
  - Loss: 0.3555
39
  - Wer: 10.9875
all_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 24.04,
3
+ "eval_loss": 0.3554980754852295,
4
+ "eval_runtime": 938.8835,
5
+ "eval_samples_per_second": 6.972,
6
+ "eval_steps_per_second": 0.218,
7
+ "eval_wer": 10.987513796909493,
8
+ "train_loss": 0.05219680506754667,
9
+ "train_runtime": 47055.421,
10
+ "train_samples_per_second": 6.8,
11
+ "train_steps_per_second": 0.106
12
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 24.04,
3
+ "eval_loss": 0.3554980754852295,
4
+ "eval_runtime": 938.8835,
5
+ "eval_samples_per_second": 6.972,
6
+ "eval_steps_per_second": 0.218,
7
+ "eval_wer": 10.987513796909493
8
+ }
train_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 24.04,
3
+ "train_loss": 0.05219680506754667,
4
+ "train_runtime": 47055.421,
5
+ "train_samples_per_second": 6.8,
6
+ "train_steps_per_second": 0.106
7
+ }
trainer_state.json ADDED
@@ -0,0 +1,1273 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 10.987513796909493,
3
+ "best_model_checkpoint": "./checkpoint-5000",
4
+ "epoch": 24.04,
5
+ "eval_steps": 1000,
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": 0.01,
13
+ "learning_rate": 4.2000000000000006e-07,
14
+ "loss": 1.6222,
15
+ "step": 25
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 9.200000000000001e-07,
20
+ "loss": 1.4136,
21
+ "step": 50
22
+ },
23
+ {
24
+ "epoch": 0.01,
25
+ "learning_rate": 1.42e-06,
26
+ "loss": 0.8717,
27
+ "step": 75
28
+ },
29
+ {
30
+ "epoch": 0.02,
31
+ "learning_rate": 1.9200000000000003e-06,
32
+ "loss": 0.593,
33
+ "step": 100
34
+ },
35
+ {
36
+ "epoch": 0.03,
37
+ "learning_rate": 2.42e-06,
38
+ "loss": 0.4696,
39
+ "step": 125
40
+ },
41
+ {
42
+ "epoch": 0.03,
43
+ "learning_rate": 2.92e-06,
44
+ "loss": 0.4241,
45
+ "step": 150
46
+ },
47
+ {
48
+ "epoch": 0.04,
49
+ "learning_rate": 3.4200000000000007e-06,
50
+ "loss": 0.3642,
51
+ "step": 175
52
+ },
53
+ {
54
+ "epoch": 0.04,
55
+ "learning_rate": 3.920000000000001e-06,
56
+ "loss": 0.3191,
57
+ "step": 200
58
+ },
59
+ {
60
+ "epoch": 1.0,
61
+ "learning_rate": 4.42e-06,
62
+ "loss": 0.3119,
63
+ "step": 225
64
+ },
65
+ {
66
+ "epoch": 1.01,
67
+ "learning_rate": 4.92e-06,
68
+ "loss": 0.3326,
69
+ "step": 250
70
+ },
71
+ {
72
+ "epoch": 1.01,
73
+ "learning_rate": 5.420000000000001e-06,
74
+ "loss": 0.2775,
75
+ "step": 275
76
+ },
77
+ {
78
+ "epoch": 1.02,
79
+ "learning_rate": 5.92e-06,
80
+ "loss": 0.2627,
81
+ "step": 300
82
+ },
83
+ {
84
+ "epoch": 1.02,
85
+ "learning_rate": 6.42e-06,
86
+ "loss": 0.2297,
87
+ "step": 325
88
+ },
89
+ {
90
+ "epoch": 1.03,
91
+ "learning_rate": 6.92e-06,
92
+ "loss": 0.2209,
93
+ "step": 350
94
+ },
95
+ {
96
+ "epoch": 1.03,
97
+ "learning_rate": 7.420000000000001e-06,
98
+ "loss": 0.196,
99
+ "step": 375
100
+ },
101
+ {
102
+ "epoch": 1.04,
103
+ "learning_rate": 7.92e-06,
104
+ "loss": 0.1794,
105
+ "step": 400
106
+ },
107
+ {
108
+ "epoch": 2.0,
109
+ "learning_rate": 8.42e-06,
110
+ "loss": 0.1686,
111
+ "step": 425
112
+ },
113
+ {
114
+ "epoch": 2.01,
115
+ "learning_rate": 8.920000000000001e-06,
116
+ "loss": 0.1933,
117
+ "step": 450
118
+ },
119
+ {
120
+ "epoch": 2.02,
121
+ "learning_rate": 9.42e-06,
122
+ "loss": 0.1568,
123
+ "step": 475
124
+ },
125
+ {
126
+ "epoch": 2.02,
127
+ "learning_rate": 9.920000000000002e-06,
128
+ "loss": 0.1471,
129
+ "step": 500
130
+ },
131
+ {
132
+ "epoch": 2.02,
133
+ "learning_rate": 9.953333333333333e-06,
134
+ "loss": 0.1256,
135
+ "step": 525
136
+ },
137
+ {
138
+ "epoch": 2.03,
139
+ "learning_rate": 9.89777777777778e-06,
140
+ "loss": 0.1168,
141
+ "step": 550
142
+ },
143
+ {
144
+ "epoch": 2.04,
145
+ "learning_rate": 9.842222222222223e-06,
146
+ "loss": 0.1041,
147
+ "step": 575
148
+ },
149
+ {
150
+ "epoch": 2.04,
151
+ "learning_rate": 9.786666666666667e-06,
152
+ "loss": 0.0967,
153
+ "step": 600
154
+ },
155
+ {
156
+ "epoch": 3.0,
157
+ "learning_rate": 9.731111111111113e-06,
158
+ "loss": 0.0841,
159
+ "step": 625
160
+ },
161
+ {
162
+ "epoch": 3.01,
163
+ "learning_rate": 9.675555555555555e-06,
164
+ "loss": 0.0993,
165
+ "step": 650
166
+ },
167
+ {
168
+ "epoch": 3.02,
169
+ "learning_rate": 9.620000000000001e-06,
170
+ "loss": 0.0769,
171
+ "step": 675
172
+ },
173
+ {
174
+ "epoch": 3.02,
175
+ "learning_rate": 9.564444444444445e-06,
176
+ "loss": 0.0699,
177
+ "step": 700
178
+ },
179
+ {
180
+ "epoch": 3.02,
181
+ "learning_rate": 9.508888888888889e-06,
182
+ "loss": 0.063,
183
+ "step": 725
184
+ },
185
+ {
186
+ "epoch": 3.03,
187
+ "learning_rate": 9.453333333333335e-06,
188
+ "loss": 0.0532,
189
+ "step": 750
190
+ },
191
+ {
192
+ "epoch": 3.04,
193
+ "learning_rate": 9.397777777777779e-06,
194
+ "loss": 0.0502,
195
+ "step": 775
196
+ },
197
+ {
198
+ "epoch": 3.04,
199
+ "learning_rate": 9.342222222222223e-06,
200
+ "loss": 0.0448,
201
+ "step": 800
202
+ },
203
+ {
204
+ "epoch": 4.0,
205
+ "learning_rate": 9.286666666666667e-06,
206
+ "loss": 0.0387,
207
+ "step": 825
208
+ },
209
+ {
210
+ "epoch": 4.01,
211
+ "learning_rate": 9.231111111111111e-06,
212
+ "loss": 0.0505,
213
+ "step": 850
214
+ },
215
+ {
216
+ "epoch": 4.01,
217
+ "learning_rate": 9.175555555555557e-06,
218
+ "loss": 0.0387,
219
+ "step": 875
220
+ },
221
+ {
222
+ "epoch": 4.02,
223
+ "learning_rate": 9.12e-06,
224
+ "loss": 0.0382,
225
+ "step": 900
226
+ },
227
+ {
228
+ "epoch": 4.03,
229
+ "learning_rate": 9.064444444444447e-06,
230
+ "loss": 0.0268,
231
+ "step": 925
232
+ },
233
+ {
234
+ "epoch": 4.03,
235
+ "learning_rate": 9.008888888888889e-06,
236
+ "loss": 0.0277,
237
+ "step": 950
238
+ },
239
+ {
240
+ "epoch": 4.04,
241
+ "learning_rate": 8.953333333333335e-06,
242
+ "loss": 0.0237,
243
+ "step": 975
244
+ },
245
+ {
246
+ "epoch": 4.04,
247
+ "learning_rate": 8.897777777777779e-06,
248
+ "loss": 0.0214,
249
+ "step": 1000
250
+ },
251
+ {
252
+ "epoch": 4.04,
253
+ "eval_loss": 0.27368560433387756,
254
+ "eval_runtime": 949.2458,
255
+ "eval_samples_per_second": 6.896,
256
+ "eval_steps_per_second": 0.216,
257
+ "eval_wer": 11.539390176600442,
258
+ "step": 1000
259
+ },
260
+ {
261
+ "epoch": 5.0,
262
+ "learning_rate": 8.842222222222223e-06,
263
+ "loss": 0.0191,
264
+ "step": 1025
265
+ },
266
+ {
267
+ "epoch": 5.01,
268
+ "learning_rate": 8.786666666666668e-06,
269
+ "loss": 0.0251,
270
+ "step": 1050
271
+ },
272
+ {
273
+ "epoch": 5.01,
274
+ "learning_rate": 8.73111111111111e-06,
275
+ "loss": 0.0178,
276
+ "step": 1075
277
+ },
278
+ {
279
+ "epoch": 5.02,
280
+ "learning_rate": 8.675555555555556e-06,
281
+ "loss": 0.0209,
282
+ "step": 1100
283
+ },
284
+ {
285
+ "epoch": 5.03,
286
+ "learning_rate": 8.62e-06,
287
+ "loss": 0.0162,
288
+ "step": 1125
289
+ },
290
+ {
291
+ "epoch": 5.03,
292
+ "learning_rate": 8.564444444444445e-06,
293
+ "loss": 0.0139,
294
+ "step": 1150
295
+ },
296
+ {
297
+ "epoch": 5.04,
298
+ "learning_rate": 8.50888888888889e-06,
299
+ "loss": 0.0142,
300
+ "step": 1175
301
+ },
302
+ {
303
+ "epoch": 5.04,
304
+ "learning_rate": 8.453333333333334e-06,
305
+ "loss": 0.0116,
306
+ "step": 1200
307
+ },
308
+ {
309
+ "epoch": 6.0,
310
+ "learning_rate": 8.397777777777778e-06,
311
+ "loss": 0.0103,
312
+ "step": 1225
313
+ },
314
+ {
315
+ "epoch": 6.01,
316
+ "learning_rate": 8.342222222222222e-06,
317
+ "loss": 0.0132,
318
+ "step": 1250
319
+ },
320
+ {
321
+ "epoch": 6.01,
322
+ "learning_rate": 8.286666666666668e-06,
323
+ "loss": 0.0104,
324
+ "step": 1275
325
+ },
326
+ {
327
+ "epoch": 6.02,
328
+ "learning_rate": 8.231111111111112e-06,
329
+ "loss": 0.0103,
330
+ "step": 1300
331
+ },
332
+ {
333
+ "epoch": 6.03,
334
+ "learning_rate": 8.175555555555556e-06,
335
+ "loss": 0.0089,
336
+ "step": 1325
337
+ },
338
+ {
339
+ "epoch": 6.03,
340
+ "learning_rate": 8.120000000000002e-06,
341
+ "loss": 0.0074,
342
+ "step": 1350
343
+ },
344
+ {
345
+ "epoch": 6.04,
346
+ "learning_rate": 8.064444444444444e-06,
347
+ "loss": 0.007,
348
+ "step": 1375
349
+ },
350
+ {
351
+ "epoch": 6.04,
352
+ "learning_rate": 8.00888888888889e-06,
353
+ "loss": 0.0062,
354
+ "step": 1400
355
+ },
356
+ {
357
+ "epoch": 7.0,
358
+ "learning_rate": 7.953333333333334e-06,
359
+ "loss": 0.0065,
360
+ "step": 1425
361
+ },
362
+ {
363
+ "epoch": 7.01,
364
+ "learning_rate": 7.897777777777778e-06,
365
+ "loss": 0.0084,
366
+ "step": 1450
367
+ },
368
+ {
369
+ "epoch": 7.01,
370
+ "learning_rate": 7.842222222222224e-06,
371
+ "loss": 0.0075,
372
+ "step": 1475
373
+ },
374
+ {
375
+ "epoch": 7.02,
376
+ "learning_rate": 7.786666666666666e-06,
377
+ "loss": 0.0065,
378
+ "step": 1500
379
+ },
380
+ {
381
+ "epoch": 7.03,
382
+ "learning_rate": 7.731111111111112e-06,
383
+ "loss": 0.0055,
384
+ "step": 1525
385
+ },
386
+ {
387
+ "epoch": 7.03,
388
+ "learning_rate": 7.675555555555556e-06,
389
+ "loss": 0.0048,
390
+ "step": 1550
391
+ },
392
+ {
393
+ "epoch": 7.04,
394
+ "learning_rate": 7.620000000000001e-06,
395
+ "loss": 0.0052,
396
+ "step": 1575
397
+ },
398
+ {
399
+ "epoch": 7.04,
400
+ "learning_rate": 7.564444444444446e-06,
401
+ "loss": 0.0039,
402
+ "step": 1600
403
+ },
404
+ {
405
+ "epoch": 8.01,
406
+ "learning_rate": 7.50888888888889e-06,
407
+ "loss": 0.0048,
408
+ "step": 1625
409
+ },
410
+ {
411
+ "epoch": 8.01,
412
+ "learning_rate": 7.453333333333334e-06,
413
+ "loss": 0.0054,
414
+ "step": 1650
415
+ },
416
+ {
417
+ "epoch": 8.02,
418
+ "learning_rate": 7.3977777777777786e-06,
419
+ "loss": 0.0045,
420
+ "step": 1675
421
+ },
422
+ {
423
+ "epoch": 8.02,
424
+ "learning_rate": 7.342222222222223e-06,
425
+ "loss": 0.0039,
426
+ "step": 1700
427
+ },
428
+ {
429
+ "epoch": 8.03,
430
+ "learning_rate": 7.2866666666666675e-06,
431
+ "loss": 0.0036,
432
+ "step": 1725
433
+ },
434
+ {
435
+ "epoch": 8.03,
436
+ "learning_rate": 7.231111111111112e-06,
437
+ "loss": 0.0034,
438
+ "step": 1750
439
+ },
440
+ {
441
+ "epoch": 8.04,
442
+ "learning_rate": 7.1755555555555556e-06,
443
+ "loss": 0.0033,
444
+ "step": 1775
445
+ },
446
+ {
447
+ "epoch": 8.04,
448
+ "learning_rate": 7.1200000000000004e-06,
449
+ "loss": 0.0033,
450
+ "step": 1800
451
+ },
452
+ {
453
+ "epoch": 9.01,
454
+ "learning_rate": 7.0644444444444445e-06,
455
+ "loss": 0.0031,
456
+ "step": 1825
457
+ },
458
+ {
459
+ "epoch": 9.01,
460
+ "learning_rate": 7.008888888888889e-06,
461
+ "loss": 0.0036,
462
+ "step": 1850
463
+ },
464
+ {
465
+ "epoch": 9.02,
466
+ "learning_rate": 6.953333333333334e-06,
467
+ "loss": 0.0029,
468
+ "step": 1875
469
+ },
470
+ {
471
+ "epoch": 9.02,
472
+ "learning_rate": 6.897777777777779e-06,
473
+ "loss": 0.0026,
474
+ "step": 1900
475
+ },
476
+ {
477
+ "epoch": 9.03,
478
+ "learning_rate": 6.842222222222222e-06,
479
+ "loss": 0.0026,
480
+ "step": 1925
481
+ },
482
+ {
483
+ "epoch": 9.03,
484
+ "learning_rate": 6.786666666666667e-06,
485
+ "loss": 0.0026,
486
+ "step": 1950
487
+ },
488
+ {
489
+ "epoch": 9.04,
490
+ "learning_rate": 6.731111111111111e-06,
491
+ "loss": 0.0024,
492
+ "step": 1975
493
+ },
494
+ {
495
+ "epoch": 9.04,
496
+ "learning_rate": 6.675555555555556e-06,
497
+ "loss": 0.0024,
498
+ "step": 2000
499
+ },
500
+ {
501
+ "epoch": 9.04,
502
+ "eval_loss": 0.31586742401123047,
503
+ "eval_runtime": 944.7668,
504
+ "eval_samples_per_second": 6.929,
505
+ "eval_steps_per_second": 0.217,
506
+ "eval_wer": 11.05649834437086,
507
+ "step": 2000
508
+ },
509
+ {
510
+ "epoch": 10.01,
511
+ "learning_rate": 6.620000000000001e-06,
512
+ "loss": 0.0024,
513
+ "step": 2025
514
+ },
515
+ {
516
+ "epoch": 10.01,
517
+ "learning_rate": 6.564444444444446e-06,
518
+ "loss": 0.0027,
519
+ "step": 2050
520
+ },
521
+ {
522
+ "epoch": 10.02,
523
+ "learning_rate": 6.508888888888889e-06,
524
+ "loss": 0.0022,
525
+ "step": 2075
526
+ },
527
+ {
528
+ "epoch": 10.02,
529
+ "learning_rate": 6.453333333333334e-06,
530
+ "loss": 0.0021,
531
+ "step": 2100
532
+ },
533
+ {
534
+ "epoch": 10.03,
535
+ "learning_rate": 6.397777777777778e-06,
536
+ "loss": 0.0019,
537
+ "step": 2125
538
+ },
539
+ {
540
+ "epoch": 10.03,
541
+ "learning_rate": 6.342222222222223e-06,
542
+ "loss": 0.0018,
543
+ "step": 2150
544
+ },
545
+ {
546
+ "epoch": 10.04,
547
+ "learning_rate": 6.286666666666668e-06,
548
+ "loss": 0.0019,
549
+ "step": 2175
550
+ },
551
+ {
552
+ "epoch": 10.04,
553
+ "learning_rate": 6.231111111111111e-06,
554
+ "loss": 0.0018,
555
+ "step": 2200
556
+ },
557
+ {
558
+ "epoch": 11.01,
559
+ "learning_rate": 6.175555555555556e-06,
560
+ "loss": 0.0019,
561
+ "step": 2225
562
+ },
563
+ {
564
+ "epoch": 11.01,
565
+ "learning_rate": 6.120000000000001e-06,
566
+ "loss": 0.0018,
567
+ "step": 2250
568
+ },
569
+ {
570
+ "epoch": 11.02,
571
+ "learning_rate": 6.064444444444445e-06,
572
+ "loss": 0.0017,
573
+ "step": 2275
574
+ },
575
+ {
576
+ "epoch": 11.02,
577
+ "learning_rate": 6.00888888888889e-06,
578
+ "loss": 0.0017,
579
+ "step": 2300
580
+ },
581
+ {
582
+ "epoch": 11.03,
583
+ "learning_rate": 5.9533333333333345e-06,
584
+ "loss": 0.0016,
585
+ "step": 2325
586
+ },
587
+ {
588
+ "epoch": 11.03,
589
+ "learning_rate": 5.897777777777778e-06,
590
+ "loss": 0.0016,
591
+ "step": 2350
592
+ },
593
+ {
594
+ "epoch": 11.04,
595
+ "learning_rate": 5.8422222222222226e-06,
596
+ "loss": 0.0015,
597
+ "step": 2375
598
+ },
599
+ {
600
+ "epoch": 11.04,
601
+ "learning_rate": 5.7866666666666674e-06,
602
+ "loss": 0.0015,
603
+ "step": 2400
604
+ },
605
+ {
606
+ "epoch": 12.01,
607
+ "learning_rate": 5.7311111111111115e-06,
608
+ "loss": 0.0015,
609
+ "step": 2425
610
+ },
611
+ {
612
+ "epoch": 12.01,
613
+ "learning_rate": 5.675555555555556e-06,
614
+ "loss": 0.0015,
615
+ "step": 2450
616
+ },
617
+ {
618
+ "epoch": 12.02,
619
+ "learning_rate": 5.620000000000001e-06,
620
+ "loss": 0.0014,
621
+ "step": 2475
622
+ },
623
+ {
624
+ "epoch": 12.02,
625
+ "learning_rate": 5.5644444444444444e-06,
626
+ "loss": 0.0014,
627
+ "step": 2500
628
+ },
629
+ {
630
+ "epoch": 12.03,
631
+ "learning_rate": 5.508888888888889e-06,
632
+ "loss": 0.0014,
633
+ "step": 2525
634
+ },
635
+ {
636
+ "epoch": 12.03,
637
+ "learning_rate": 5.453333333333334e-06,
638
+ "loss": 0.0013,
639
+ "step": 2550
640
+ },
641
+ {
642
+ "epoch": 12.04,
643
+ "learning_rate": 5.397777777777778e-06,
644
+ "loss": 0.0013,
645
+ "step": 2575
646
+ },
647
+ {
648
+ "epoch": 12.04,
649
+ "learning_rate": 5.342222222222223e-06,
650
+ "loss": 0.0013,
651
+ "step": 2600
652
+ },
653
+ {
654
+ "epoch": 13.01,
655
+ "learning_rate": 5.286666666666666e-06,
656
+ "loss": 0.0013,
657
+ "step": 2625
658
+ },
659
+ {
660
+ "epoch": 13.01,
661
+ "learning_rate": 5.231111111111111e-06,
662
+ "loss": 0.0013,
663
+ "step": 2650
664
+ },
665
+ {
666
+ "epoch": 13.02,
667
+ "learning_rate": 5.175555555555556e-06,
668
+ "loss": 0.0013,
669
+ "step": 2675
670
+ },
671
+ {
672
+ "epoch": 13.02,
673
+ "learning_rate": 5.12e-06,
674
+ "loss": 0.0013,
675
+ "step": 2700
676
+ },
677
+ {
678
+ "epoch": 13.03,
679
+ "learning_rate": 5.064444444444445e-06,
680
+ "loss": 0.0012,
681
+ "step": 2725
682
+ },
683
+ {
684
+ "epoch": 13.03,
685
+ "learning_rate": 5.00888888888889e-06,
686
+ "loss": 0.0012,
687
+ "step": 2750
688
+ },
689
+ {
690
+ "epoch": 13.04,
691
+ "learning_rate": 4.953333333333334e-06,
692
+ "loss": 0.0012,
693
+ "step": 2775
694
+ },
695
+ {
696
+ "epoch": 13.04,
697
+ "learning_rate": 4.897777777777778e-06,
698
+ "loss": 0.0011,
699
+ "step": 2800
700
+ },
701
+ {
702
+ "epoch": 14.01,
703
+ "learning_rate": 4.842222222222223e-06,
704
+ "loss": 0.0011,
705
+ "step": 2825
706
+ },
707
+ {
708
+ "epoch": 14.01,
709
+ "learning_rate": 4.786666666666667e-06,
710
+ "loss": 0.0012,
711
+ "step": 2850
712
+ },
713
+ {
714
+ "epoch": 14.02,
715
+ "learning_rate": 4.731111111111112e-06,
716
+ "loss": 0.0011,
717
+ "step": 2875
718
+ },
719
+ {
720
+ "epoch": 14.02,
721
+ "learning_rate": 4.675555555555556e-06,
722
+ "loss": 0.0011,
723
+ "step": 2900
724
+ },
725
+ {
726
+ "epoch": 14.03,
727
+ "learning_rate": 4.620000000000001e-06,
728
+ "loss": 0.0011,
729
+ "step": 2925
730
+ },
731
+ {
732
+ "epoch": 14.03,
733
+ "learning_rate": 4.564444444444445e-06,
734
+ "loss": 0.0011,
735
+ "step": 2950
736
+ },
737
+ {
738
+ "epoch": 14.04,
739
+ "learning_rate": 4.50888888888889e-06,
740
+ "loss": 0.0011,
741
+ "step": 2975
742
+ },
743
+ {
744
+ "epoch": 14.04,
745
+ "learning_rate": 4.453333333333334e-06,
746
+ "loss": 0.001,
747
+ "step": 3000
748
+ },
749
+ {
750
+ "epoch": 14.04,
751
+ "eval_loss": 0.3369614779949188,
752
+ "eval_runtime": 1058.7998,
753
+ "eval_samples_per_second": 6.182,
754
+ "eval_steps_per_second": 0.194,
755
+ "eval_wer": 10.994412251655628,
756
+ "step": 3000
757
+ },
758
+ {
759
+ "epoch": 15.01,
760
+ "learning_rate": 4.397777777777778e-06,
761
+ "loss": 0.001,
762
+ "step": 3025
763
+ },
764
+ {
765
+ "epoch": 15.01,
766
+ "learning_rate": 4.3422222222222225e-06,
767
+ "loss": 0.001,
768
+ "step": 3050
769
+ },
770
+ {
771
+ "epoch": 15.02,
772
+ "learning_rate": 4.2866666666666666e-06,
773
+ "loss": 0.001,
774
+ "step": 3075
775
+ },
776
+ {
777
+ "epoch": 15.02,
778
+ "learning_rate": 4.2311111111111114e-06,
779
+ "loss": 0.001,
780
+ "step": 3100
781
+ },
782
+ {
783
+ "epoch": 15.03,
784
+ "learning_rate": 4.175555555555556e-06,
785
+ "loss": 0.001,
786
+ "step": 3125
787
+ },
788
+ {
789
+ "epoch": 15.03,
790
+ "learning_rate": 4.12e-06,
791
+ "loss": 0.001,
792
+ "step": 3150
793
+ },
794
+ {
795
+ "epoch": 15.04,
796
+ "learning_rate": 4.064444444444444e-06,
797
+ "loss": 0.001,
798
+ "step": 3175
799
+ },
800
+ {
801
+ "epoch": 15.04,
802
+ "learning_rate": 4.008888888888889e-06,
803
+ "loss": 0.0009,
804
+ "step": 3200
805
+ },
806
+ {
807
+ "epoch": 16.0,
808
+ "learning_rate": 3.953333333333333e-06,
809
+ "loss": 0.0009,
810
+ "step": 3225
811
+ },
812
+ {
813
+ "epoch": 16.01,
814
+ "learning_rate": 3.897777777777778e-06,
815
+ "loss": 0.001,
816
+ "step": 3250
817
+ },
818
+ {
819
+ "epoch": 16.02,
820
+ "learning_rate": 3.842222222222223e-06,
821
+ "loss": 0.0009,
822
+ "step": 3275
823
+ },
824
+ {
825
+ "epoch": 16.02,
826
+ "learning_rate": 3.7866666666666667e-06,
827
+ "loss": 0.0009,
828
+ "step": 3300
829
+ },
830
+ {
831
+ "epoch": 16.02,
832
+ "learning_rate": 3.7311111111111116e-06,
833
+ "loss": 0.0009,
834
+ "step": 3325
835
+ },
836
+ {
837
+ "epoch": 16.03,
838
+ "learning_rate": 3.675555555555556e-06,
839
+ "loss": 0.0009,
840
+ "step": 3350
841
+ },
842
+ {
843
+ "epoch": 16.04,
844
+ "learning_rate": 3.62e-06,
845
+ "loss": 0.0009,
846
+ "step": 3375
847
+ },
848
+ {
849
+ "epoch": 16.04,
850
+ "learning_rate": 3.564444444444445e-06,
851
+ "loss": 0.0009,
852
+ "step": 3400
853
+ },
854
+ {
855
+ "epoch": 17.0,
856
+ "learning_rate": 3.508888888888889e-06,
857
+ "loss": 0.0009,
858
+ "step": 3425
859
+ },
860
+ {
861
+ "epoch": 17.01,
862
+ "learning_rate": 3.4533333333333334e-06,
863
+ "loss": 0.0009,
864
+ "step": 3450
865
+ },
866
+ {
867
+ "epoch": 17.02,
868
+ "learning_rate": 3.3977777777777783e-06,
869
+ "loss": 0.0009,
870
+ "step": 3475
871
+ },
872
+ {
873
+ "epoch": 17.02,
874
+ "learning_rate": 3.3422222222222224e-06,
875
+ "loss": 0.0009,
876
+ "step": 3500
877
+ },
878
+ {
879
+ "epoch": 17.02,
880
+ "learning_rate": 3.286666666666667e-06,
881
+ "loss": 0.0008,
882
+ "step": 3525
883
+ },
884
+ {
885
+ "epoch": 17.03,
886
+ "learning_rate": 3.2311111111111117e-06,
887
+ "loss": 0.0008,
888
+ "step": 3550
889
+ },
890
+ {
891
+ "epoch": 17.04,
892
+ "learning_rate": 3.1755555555555557e-06,
893
+ "loss": 0.0008,
894
+ "step": 3575
895
+ },
896
+ {
897
+ "epoch": 17.04,
898
+ "learning_rate": 3.12e-06,
899
+ "loss": 0.0008,
900
+ "step": 3600
901
+ },
902
+ {
903
+ "epoch": 18.0,
904
+ "learning_rate": 3.064444444444445e-06,
905
+ "loss": 0.0008,
906
+ "step": 3625
907
+ },
908
+ {
909
+ "epoch": 18.01,
910
+ "learning_rate": 3.008888888888889e-06,
911
+ "loss": 0.0008,
912
+ "step": 3650
913
+ },
914
+ {
915
+ "epoch": 18.02,
916
+ "learning_rate": 2.9533333333333336e-06,
917
+ "loss": 0.0008,
918
+ "step": 3675
919
+ },
920
+ {
921
+ "epoch": 18.02,
922
+ "learning_rate": 2.8977777777777785e-06,
923
+ "loss": 0.0008,
924
+ "step": 3700
925
+ },
926
+ {
927
+ "epoch": 18.02,
928
+ "learning_rate": 2.8422222222222225e-06,
929
+ "loss": 0.0008,
930
+ "step": 3725
931
+ },
932
+ {
933
+ "epoch": 18.03,
934
+ "learning_rate": 2.786666666666667e-06,
935
+ "loss": 0.0008,
936
+ "step": 3750
937
+ },
938
+ {
939
+ "epoch": 18.04,
940
+ "learning_rate": 2.7311111111111114e-06,
941
+ "loss": 0.0008,
942
+ "step": 3775
943
+ },
944
+ {
945
+ "epoch": 18.04,
946
+ "learning_rate": 2.675555555555556e-06,
947
+ "loss": 0.0007,
948
+ "step": 3800
949
+ },
950
+ {
951
+ "epoch": 19.0,
952
+ "learning_rate": 2.6200000000000003e-06,
953
+ "loss": 0.0007,
954
+ "step": 3825
955
+ },
956
+ {
957
+ "epoch": 19.01,
958
+ "learning_rate": 2.5644444444444444e-06,
959
+ "loss": 0.0008,
960
+ "step": 3850
961
+ },
962
+ {
963
+ "epoch": 19.02,
964
+ "learning_rate": 2.5088888888888892e-06,
965
+ "loss": 0.0008,
966
+ "step": 3875
967
+ },
968
+ {
969
+ "epoch": 19.02,
970
+ "learning_rate": 2.4533333333333333e-06,
971
+ "loss": 0.0008,
972
+ "step": 3900
973
+ },
974
+ {
975
+ "epoch": 19.02,
976
+ "learning_rate": 2.397777777777778e-06,
977
+ "loss": 0.0007,
978
+ "step": 3925
979
+ },
980
+ {
981
+ "epoch": 19.03,
982
+ "learning_rate": 2.342222222222222e-06,
983
+ "loss": 0.0007,
984
+ "step": 3950
985
+ },
986
+ {
987
+ "epoch": 19.04,
988
+ "learning_rate": 2.2866666666666667e-06,
989
+ "loss": 0.0007,
990
+ "step": 3975
991
+ },
992
+ {
993
+ "epoch": 19.04,
994
+ "learning_rate": 2.2311111111111115e-06,
995
+ "loss": 0.0007,
996
+ "step": 4000
997
+ },
998
+ {
999
+ "epoch": 19.04,
1000
+ "eval_loss": 0.3497324585914612,
1001
+ "eval_runtime": 935.4192,
1002
+ "eval_samples_per_second": 6.998,
1003
+ "eval_steps_per_second": 0.219,
1004
+ "eval_wer": 11.01510761589404,
1005
+ "step": 4000
1006
+ },
1007
+ {
1008
+ "epoch": 20.0,
1009
+ "learning_rate": 2.1755555555555556e-06,
1010
+ "loss": 0.0007,
1011
+ "step": 4025
1012
+ },
1013
+ {
1014
+ "epoch": 20.01,
1015
+ "learning_rate": 2.12e-06,
1016
+ "loss": 0.0007,
1017
+ "step": 4050
1018
+ },
1019
+ {
1020
+ "epoch": 20.02,
1021
+ "learning_rate": 2.064444444444445e-06,
1022
+ "loss": 0.0007,
1023
+ "step": 4075
1024
+ },
1025
+ {
1026
+ "epoch": 20.02,
1027
+ "learning_rate": 2.008888888888889e-06,
1028
+ "loss": 0.0007,
1029
+ "step": 4100
1030
+ },
1031
+ {
1032
+ "epoch": 20.02,
1033
+ "learning_rate": 1.9533333333333334e-06,
1034
+ "loss": 0.0007,
1035
+ "step": 4125
1036
+ },
1037
+ {
1038
+ "epoch": 20.03,
1039
+ "learning_rate": 1.8977777777777779e-06,
1040
+ "loss": 0.0007,
1041
+ "step": 4150
1042
+ },
1043
+ {
1044
+ "epoch": 20.04,
1045
+ "learning_rate": 1.8422222222222225e-06,
1046
+ "loss": 0.0007,
1047
+ "step": 4175
1048
+ },
1049
+ {
1050
+ "epoch": 20.04,
1051
+ "learning_rate": 1.7866666666666668e-06,
1052
+ "loss": 0.0007,
1053
+ "step": 4200
1054
+ },
1055
+ {
1056
+ "epoch": 21.0,
1057
+ "learning_rate": 1.7311111111111112e-06,
1058
+ "loss": 0.0007,
1059
+ "step": 4225
1060
+ },
1061
+ {
1062
+ "epoch": 21.01,
1063
+ "learning_rate": 1.675555555555556e-06,
1064
+ "loss": 0.0007,
1065
+ "step": 4250
1066
+ },
1067
+ {
1068
+ "epoch": 21.02,
1069
+ "learning_rate": 1.6200000000000002e-06,
1070
+ "loss": 0.0007,
1071
+ "step": 4275
1072
+ },
1073
+ {
1074
+ "epoch": 21.02,
1075
+ "learning_rate": 1.5644444444444446e-06,
1076
+ "loss": 0.0007,
1077
+ "step": 4300
1078
+ },
1079
+ {
1080
+ "epoch": 21.02,
1081
+ "learning_rate": 1.5088888888888889e-06,
1082
+ "loss": 0.0007,
1083
+ "step": 4325
1084
+ },
1085
+ {
1086
+ "epoch": 21.03,
1087
+ "learning_rate": 1.4533333333333335e-06,
1088
+ "loss": 0.0006,
1089
+ "step": 4350
1090
+ },
1091
+ {
1092
+ "epoch": 21.04,
1093
+ "learning_rate": 1.397777777777778e-06,
1094
+ "loss": 0.0007,
1095
+ "step": 4375
1096
+ },
1097
+ {
1098
+ "epoch": 21.04,
1099
+ "learning_rate": 1.3422222222222222e-06,
1100
+ "loss": 0.0006,
1101
+ "step": 4400
1102
+ },
1103
+ {
1104
+ "epoch": 22.0,
1105
+ "learning_rate": 1.286666666666667e-06,
1106
+ "loss": 0.0006,
1107
+ "step": 4425
1108
+ },
1109
+ {
1110
+ "epoch": 22.01,
1111
+ "learning_rate": 1.2311111111111112e-06,
1112
+ "loss": 0.0007,
1113
+ "step": 4450
1114
+ },
1115
+ {
1116
+ "epoch": 22.02,
1117
+ "learning_rate": 1.1755555555555556e-06,
1118
+ "loss": 0.0006,
1119
+ "step": 4475
1120
+ },
1121
+ {
1122
+ "epoch": 22.02,
1123
+ "learning_rate": 1.12e-06,
1124
+ "loss": 0.0007,
1125
+ "step": 4500
1126
+ },
1127
+ {
1128
+ "epoch": 22.02,
1129
+ "learning_rate": 1.0644444444444445e-06,
1130
+ "loss": 0.0006,
1131
+ "step": 4525
1132
+ },
1133
+ {
1134
+ "epoch": 22.03,
1135
+ "learning_rate": 1.008888888888889e-06,
1136
+ "loss": 0.0006,
1137
+ "step": 4550
1138
+ },
1139
+ {
1140
+ "epoch": 22.04,
1141
+ "learning_rate": 9.533333333333335e-07,
1142
+ "loss": 0.0006,
1143
+ "step": 4575
1144
+ },
1145
+ {
1146
+ "epoch": 22.04,
1147
+ "learning_rate": 8.977777777777778e-07,
1148
+ "loss": 0.0006,
1149
+ "step": 4600
1150
+ },
1151
+ {
1152
+ "epoch": 23.0,
1153
+ "learning_rate": 8.422222222222224e-07,
1154
+ "loss": 0.0006,
1155
+ "step": 4625
1156
+ },
1157
+ {
1158
+ "epoch": 23.01,
1159
+ "learning_rate": 7.866666666666667e-07,
1160
+ "loss": 0.0006,
1161
+ "step": 4650
1162
+ },
1163
+ {
1164
+ "epoch": 23.02,
1165
+ "learning_rate": 7.311111111111112e-07,
1166
+ "loss": 0.0006,
1167
+ "step": 4675
1168
+ },
1169
+ {
1170
+ "epoch": 23.02,
1171
+ "learning_rate": 6.755555555555555e-07,
1172
+ "loss": 0.0006,
1173
+ "step": 4700
1174
+ },
1175
+ {
1176
+ "epoch": 23.02,
1177
+ "learning_rate": 6.200000000000001e-07,
1178
+ "loss": 0.0006,
1179
+ "step": 4725
1180
+ },
1181
+ {
1182
+ "epoch": 23.03,
1183
+ "learning_rate": 5.644444444444445e-07,
1184
+ "loss": 0.0006,
1185
+ "step": 4750
1186
+ },
1187
+ {
1188
+ "epoch": 23.04,
1189
+ "learning_rate": 5.088888888888889e-07,
1190
+ "loss": 0.0006,
1191
+ "step": 4775
1192
+ },
1193
+ {
1194
+ "epoch": 23.04,
1195
+ "learning_rate": 4.533333333333334e-07,
1196
+ "loss": 0.0006,
1197
+ "step": 4800
1198
+ },
1199
+ {
1200
+ "epoch": 24.0,
1201
+ "learning_rate": 3.9777777777777783e-07,
1202
+ "loss": 0.0006,
1203
+ "step": 4825
1204
+ },
1205
+ {
1206
+ "epoch": 24.01,
1207
+ "learning_rate": 3.422222222222223e-07,
1208
+ "loss": 0.0006,
1209
+ "step": 4850
1210
+ },
1211
+ {
1212
+ "epoch": 24.02,
1213
+ "learning_rate": 2.866666666666667e-07,
1214
+ "loss": 0.0006,
1215
+ "step": 4875
1216
+ },
1217
+ {
1218
+ "epoch": 24.02,
1219
+ "learning_rate": 2.3111111111111112e-07,
1220
+ "loss": 0.0006,
1221
+ "step": 4900
1222
+ },
1223
+ {
1224
+ "epoch": 24.02,
1225
+ "learning_rate": 1.7555555555555558e-07,
1226
+ "loss": 0.0006,
1227
+ "step": 4925
1228
+ },
1229
+ {
1230
+ "epoch": 24.03,
1231
+ "learning_rate": 1.2000000000000002e-07,
1232
+ "loss": 0.0006,
1233
+ "step": 4950
1234
+ },
1235
+ {
1236
+ "epoch": 24.04,
1237
+ "learning_rate": 6.444444444444445e-08,
1238
+ "loss": 0.0006,
1239
+ "step": 4975
1240
+ },
1241
+ {
1242
+ "epoch": 24.04,
1243
+ "learning_rate": 8.88888888888889e-09,
1244
+ "loss": 0.0006,
1245
+ "step": 5000
1246
+ },
1247
+ {
1248
+ "epoch": 24.04,
1249
+ "eval_loss": 0.3554980754852295,
1250
+ "eval_runtime": 939.4371,
1251
+ "eval_samples_per_second": 6.968,
1252
+ "eval_steps_per_second": 0.218,
1253
+ "eval_wer": 10.987513796909493,
1254
+ "step": 5000
1255
+ },
1256
+ {
1257
+ "epoch": 24.04,
1258
+ "step": 5000,
1259
+ "total_flos": 9.2304040292352e+19,
1260
+ "train_loss": 0.05219680506754667,
1261
+ "train_runtime": 47055.421,
1262
+ "train_samples_per_second": 6.8,
1263
+ "train_steps_per_second": 0.106
1264
+ }
1265
+ ],
1266
+ "logging_steps": 25,
1267
+ "max_steps": 5000,
1268
+ "num_train_epochs": 9223372036854775807,
1269
+ "save_steps": 1000,
1270
+ "total_flos": 9.2304040292352e+19,
1271
+ "trial_name": null,
1272
+ "trial_params": null
1273
+ }