Leonel-Maia commited on
Commit
01e83a8
·
verified ·
1 Parent(s): ac25f89

End of training

Browse files
Files changed (5) hide show
  1. README.md +16 -4
  2. all_results.json +15 -0
  3. eval_results.json +9 -0
  4. train_results.json +9 -0
  5. trainer_state.json +1095 -0
README.md CHANGED
@@ -4,11 +4,23 @@ license: apache-2.0
4
  base_model: Leonel-Maia/fongbe-whisper-small
5
  tags:
6
  - generated_from_trainer
 
 
7
  metrics:
8
  - wer
9
  model-index:
10
  - name: whisper-small-tf
11
- results: []
 
 
 
 
 
 
 
 
 
 
12
  ---
13
 
14
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -16,10 +28,10 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # whisper-small-tf
18
 
19
- This model is a fine-tuned version of [Leonel-Maia/fongbe-whisper-small](https://huggingface.co/Leonel-Maia/fongbe-whisper-small) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.7468
22
- - Wer: 0.4017
23
 
24
  ## Model description
25
 
 
4
  base_model: Leonel-Maia/fongbe-whisper-small
5
  tags:
6
  - generated_from_trainer
7
+ datasets:
8
+ - Leonel-Maia/ewe_dataset
9
  metrics:
10
  - wer
11
  model-index:
12
  - name: whisper-small-tf
13
+ results:
14
+ - task:
15
+ name: Automatic Speech Recognition
16
+ type: automatic-speech-recognition
17
+ dataset:
18
+ name: Leonel-Maia/ewe_dataset
19
+ type: Leonel-Maia/ewe_dataset
20
+ metrics:
21
+ - name: Wer
22
+ type: wer
23
+ value: 0.3810818307905687
24
  ---
25
 
26
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
28
 
29
  # whisper-small-tf
30
 
31
+ This model is a fine-tuned version of [Leonel-Maia/fongbe-whisper-small](https://huggingface.co/Leonel-Maia/fongbe-whisper-small) on the Leonel-Maia/ewe_dataset dataset.
32
  It achieves the following results on the evaluation set:
33
+ - Loss: 0.4403
34
+ - Wer: 0.3811
35
 
36
  ## Model description
37
 
all_results.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.281442392260335,
3
+ "eval_loss": 0.44026532769203186,
4
+ "eval_runtime": 1221.4109,
5
+ "eval_samples": 1607,
6
+ "eval_samples_per_second": 1.316,
7
+ "eval_steps_per_second": 0.329,
8
+ "eval_wer": 0.3810818307905687,
9
+ "total_flos": 3.2234989215744e+19,
10
+ "train_loss": 0.28376776227780753,
11
+ "train_runtime": 61744.3298,
12
+ "train_samples": 9095,
13
+ "train_samples_per_second": 8.838,
14
+ "train_steps_per_second": 0.276
15
+ }
eval_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.281442392260335,
3
+ "eval_loss": 0.44026532769203186,
4
+ "eval_runtime": 1221.4109,
5
+ "eval_samples": 1607,
6
+ "eval_samples_per_second": 1.316,
7
+ "eval_steps_per_second": 0.329,
8
+ "eval_wer": 0.3810818307905687
9
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.281442392260335,
3
+ "total_flos": 3.2234989215744e+19,
4
+ "train_loss": 0.28376776227780753,
5
+ "train_runtime": 61744.3298,
6
+ "train_samples": 9095,
7
+ "train_samples_per_second": 8.838,
8
+ "train_steps_per_second": 0.276
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1095 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 1000,
3
+ "best_metric": 0.44026532769203186,
4
+ "best_model_checkpoint": "./whisper-small-tf/checkpoint-1000",
5
+ "epoch": 12.281442392260335,
6
+ "eval_steps": 500,
7
+ "global_step": 3500,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.08795074758135445,
14
+ "grad_norm": 25.806161880493164,
15
+ "learning_rate": 5.000000000000001e-07,
16
+ "loss": 4.0027,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.1759014951627089,
21
+ "grad_norm": 14.938480377197266,
22
+ "learning_rate": 1.0000000000000002e-06,
23
+ "loss": 3.1256,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.2638522427440633,
28
+ "grad_norm": 8.672367095947266,
29
+ "learning_rate": 1.5e-06,
30
+ "loss": 2.1951,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.3518029903254178,
35
+ "grad_norm": 6.627525329589844,
36
+ "learning_rate": 2.0000000000000003e-06,
37
+ "loss": 1.5878,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.4397537379067722,
42
+ "grad_norm": 5.021503925323486,
43
+ "learning_rate": 2.5e-06,
44
+ "loss": 1.1926,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.5277044854881267,
49
+ "grad_norm": 4.1880974769592285,
50
+ "learning_rate": 3e-06,
51
+ "loss": 0.9852,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.6156552330694811,
56
+ "grad_norm": 3.801764488220215,
57
+ "learning_rate": 3.5e-06,
58
+ "loss": 0.8812,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.7036059806508356,
63
+ "grad_norm": 4.1236252784729,
64
+ "learning_rate": 4.000000000000001e-06,
65
+ "loss": 0.7937,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.7915567282321899,
70
+ "grad_norm": 3.438084125518799,
71
+ "learning_rate": 4.5e-06,
72
+ "loss": 0.7126,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.8795074758135444,
77
+ "grad_norm": 3.6438441276550293,
78
+ "learning_rate": 5e-06,
79
+ "loss": 0.6662,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.9674582233948988,
84
+ "grad_norm": 3.1786344051361084,
85
+ "learning_rate": 5.500000000000001e-06,
86
+ "loss": 0.6345,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 1.0527704485488127,
91
+ "grad_norm": 3.2465078830718994,
92
+ "learning_rate": 6e-06,
93
+ "loss": 0.566,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 1.140721196130167,
98
+ "grad_norm": 2.9073245525360107,
99
+ "learning_rate": 6.5000000000000004e-06,
100
+ "loss": 0.5746,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 1.2286719437115217,
105
+ "grad_norm": 3.0291011333465576,
106
+ "learning_rate": 7e-06,
107
+ "loss": 0.5508,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 1.316622691292876,
112
+ "grad_norm": 3.245563507080078,
113
+ "learning_rate": 7.500000000000001e-06,
114
+ "loss": 0.5416,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 1.4045734388742304,
119
+ "grad_norm": 3.0863044261932373,
120
+ "learning_rate": 8.000000000000001e-06,
121
+ "loss": 0.5405,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 1.492524186455585,
126
+ "grad_norm": 3.0560436248779297,
127
+ "learning_rate": 8.5e-06,
128
+ "loss": 0.5106,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 1.5804749340369393,
133
+ "grad_norm": 3.383167266845703,
134
+ "learning_rate": 9e-06,
135
+ "loss": 0.5055,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 1.6684256816182939,
140
+ "grad_norm": 2.792325019836426,
141
+ "learning_rate": 9.5e-06,
142
+ "loss": 0.4976,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 1.7563764291996482,
147
+ "grad_norm": 2.915156364440918,
148
+ "learning_rate": 1e-05,
149
+ "loss": 0.4849,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 1.7563764291996482,
154
+ "eval_loss": 0.49483707547187805,
155
+ "eval_runtime": 1285.9954,
156
+ "eval_samples_per_second": 1.25,
157
+ "eval_steps_per_second": 0.313,
158
+ "eval_wer": 0.43881645862228386,
159
+ "step": 500
160
+ },
161
+ {
162
+ "epoch": 1.8443271767810026,
163
+ "grad_norm": 2.707939624786377,
164
+ "learning_rate": 9.984885126964934e-06,
165
+ "loss": 0.4947,
166
+ "step": 525
167
+ },
168
+ {
169
+ "epoch": 1.9322779243623571,
170
+ "grad_norm": 2.6238901615142822,
171
+ "learning_rate": 9.969770253929868e-06,
172
+ "loss": 0.4804,
173
+ "step": 550
174
+ },
175
+ {
176
+ "epoch": 2.0175901495162707,
177
+ "grad_norm": 2.485342025756836,
178
+ "learning_rate": 9.9546553808948e-06,
179
+ "loss": 0.4519,
180
+ "step": 575
181
+ },
182
+ {
183
+ "epoch": 2.1055408970976255,
184
+ "grad_norm": 2.6421515941619873,
185
+ "learning_rate": 9.939540507859735e-06,
186
+ "loss": 0.4097,
187
+ "step": 600
188
+ },
189
+ {
190
+ "epoch": 2.19349164467898,
191
+ "grad_norm": 2.8181490898132324,
192
+ "learning_rate": 9.924425634824668e-06,
193
+ "loss": 0.4164,
194
+ "step": 625
195
+ },
196
+ {
197
+ "epoch": 2.281442392260334,
198
+ "grad_norm": 2.8466315269470215,
199
+ "learning_rate": 9.909310761789602e-06,
200
+ "loss": 0.4005,
201
+ "step": 650
202
+ },
203
+ {
204
+ "epoch": 2.3693931398416885,
205
+ "grad_norm": 2.6883461475372314,
206
+ "learning_rate": 9.894195888754535e-06,
207
+ "loss": 0.3989,
208
+ "step": 675
209
+ },
210
+ {
211
+ "epoch": 2.4573438874230433,
212
+ "grad_norm": 2.4732818603515625,
213
+ "learning_rate": 9.879081015719468e-06,
214
+ "loss": 0.3987,
215
+ "step": 700
216
+ },
217
+ {
218
+ "epoch": 2.5452946350043977,
219
+ "grad_norm": 2.8190526962280273,
220
+ "learning_rate": 9.863966142684402e-06,
221
+ "loss": 0.3872,
222
+ "step": 725
223
+ },
224
+ {
225
+ "epoch": 2.633245382585752,
226
+ "grad_norm": 2.836927890777588,
227
+ "learning_rate": 9.848851269649336e-06,
228
+ "loss": 0.395,
229
+ "step": 750
230
+ },
231
+ {
232
+ "epoch": 2.7211961301671064,
233
+ "grad_norm": 2.63208270072937,
234
+ "learning_rate": 9.833736396614269e-06,
235
+ "loss": 0.3887,
236
+ "step": 775
237
+ },
238
+ {
239
+ "epoch": 2.8091468777484607,
240
+ "grad_norm": 2.601407051086426,
241
+ "learning_rate": 9.818621523579203e-06,
242
+ "loss": 0.3934,
243
+ "step": 800
244
+ },
245
+ {
246
+ "epoch": 2.897097625329815,
247
+ "grad_norm": 2.5038022994995117,
248
+ "learning_rate": 9.803506650544136e-06,
249
+ "loss": 0.3639,
250
+ "step": 825
251
+ },
252
+ {
253
+ "epoch": 2.98504837291117,
254
+ "grad_norm": 2.752920389175415,
255
+ "learning_rate": 9.78839177750907e-06,
256
+ "loss": 0.398,
257
+ "step": 850
258
+ },
259
+ {
260
+ "epoch": 3.0703605980650837,
261
+ "grad_norm": 2.29534912109375,
262
+ "learning_rate": 9.773276904474003e-06,
263
+ "loss": 0.3114,
264
+ "step": 875
265
+ },
266
+ {
267
+ "epoch": 3.158311345646438,
268
+ "grad_norm": 2.2628939151763916,
269
+ "learning_rate": 9.758162031438938e-06,
270
+ "loss": 0.3008,
271
+ "step": 900
272
+ },
273
+ {
274
+ "epoch": 3.2462620932277924,
275
+ "grad_norm": 2.3943662643432617,
276
+ "learning_rate": 9.74304715840387e-06,
277
+ "loss": 0.3324,
278
+ "step": 925
279
+ },
280
+ {
281
+ "epoch": 3.3342128408091467,
282
+ "grad_norm": 2.5597457885742188,
283
+ "learning_rate": 9.727932285368805e-06,
284
+ "loss": 0.3288,
285
+ "step": 950
286
+ },
287
+ {
288
+ "epoch": 3.4221635883905015,
289
+ "grad_norm": 2.2671427726745605,
290
+ "learning_rate": 9.712817412333737e-06,
291
+ "loss": 0.3151,
292
+ "step": 975
293
+ },
294
+ {
295
+ "epoch": 3.510114335971856,
296
+ "grad_norm": 2.1839168071746826,
297
+ "learning_rate": 9.697702539298672e-06,
298
+ "loss": 0.3147,
299
+ "step": 1000
300
+ },
301
+ {
302
+ "epoch": 3.510114335971856,
303
+ "eval_loss": 0.44026532769203186,
304
+ "eval_runtime": 1314.0224,
305
+ "eval_samples_per_second": 1.223,
306
+ "eval_steps_per_second": 0.306,
307
+ "eval_wer": 0.3810818307905687,
308
+ "step": 1000
309
+ },
310
+ {
311
+ "epoch": 3.59806508355321,
312
+ "grad_norm": 2.6418137550354004,
313
+ "learning_rate": 9.682587666263604e-06,
314
+ "loss": 0.3107,
315
+ "step": 1025
316
+ },
317
+ {
318
+ "epoch": 3.6860158311345645,
319
+ "grad_norm": 2.4836411476135254,
320
+ "learning_rate": 9.667472793228537e-06,
321
+ "loss": 0.3178,
322
+ "step": 1050
323
+ },
324
+ {
325
+ "epoch": 3.7739665787159193,
326
+ "grad_norm": 2.551145553588867,
327
+ "learning_rate": 9.652357920193472e-06,
328
+ "loss": 0.3315,
329
+ "step": 1075
330
+ },
331
+ {
332
+ "epoch": 3.8619173262972737,
333
+ "grad_norm": 2.5726044178009033,
334
+ "learning_rate": 9.637243047158404e-06,
335
+ "loss": 0.3247,
336
+ "step": 1100
337
+ },
338
+ {
339
+ "epoch": 3.949868073878628,
340
+ "grad_norm": 2.7298617362976074,
341
+ "learning_rate": 9.622128174123339e-06,
342
+ "loss": 0.3286,
343
+ "step": 1125
344
+ },
345
+ {
346
+ "epoch": 4.035180299032541,
347
+ "grad_norm": 2.5060338973999023,
348
+ "learning_rate": 9.607013301088271e-06,
349
+ "loss": 0.2934,
350
+ "step": 1150
351
+ },
352
+ {
353
+ "epoch": 4.123131046613897,
354
+ "grad_norm": 2.5333456993103027,
355
+ "learning_rate": 9.591898428053204e-06,
356
+ "loss": 0.2386,
357
+ "step": 1175
358
+ },
359
+ {
360
+ "epoch": 4.211081794195251,
361
+ "grad_norm": 2.3387417793273926,
362
+ "learning_rate": 9.576783555018138e-06,
363
+ "loss": 0.2373,
364
+ "step": 1200
365
+ },
366
+ {
367
+ "epoch": 4.299032541776605,
368
+ "grad_norm": 2.955852746963501,
369
+ "learning_rate": 9.561668681983071e-06,
370
+ "loss": 0.2715,
371
+ "step": 1225
372
+ },
373
+ {
374
+ "epoch": 4.38698328935796,
375
+ "grad_norm": 2.3756492137908936,
376
+ "learning_rate": 9.546553808948006e-06,
377
+ "loss": 0.26,
378
+ "step": 1250
379
+ },
380
+ {
381
+ "epoch": 4.474934036939314,
382
+ "grad_norm": 2.6328377723693848,
383
+ "learning_rate": 9.531438935912938e-06,
384
+ "loss": 0.2566,
385
+ "step": 1275
386
+ },
387
+ {
388
+ "epoch": 4.562884784520668,
389
+ "grad_norm": 2.5527310371398926,
390
+ "learning_rate": 9.516324062877873e-06,
391
+ "loss": 0.2472,
392
+ "step": 1300
393
+ },
394
+ {
395
+ "epoch": 4.650835532102023,
396
+ "grad_norm": 2.7296290397644043,
397
+ "learning_rate": 9.501209189842805e-06,
398
+ "loss": 0.2526,
399
+ "step": 1325
400
+ },
401
+ {
402
+ "epoch": 4.738786279683377,
403
+ "grad_norm": 2.5807669162750244,
404
+ "learning_rate": 9.48609431680774e-06,
405
+ "loss": 0.2579,
406
+ "step": 1350
407
+ },
408
+ {
409
+ "epoch": 4.826737027264731,
410
+ "grad_norm": 2.646742105484009,
411
+ "learning_rate": 9.470979443772672e-06,
412
+ "loss": 0.2633,
413
+ "step": 1375
414
+ },
415
+ {
416
+ "epoch": 4.914687774846087,
417
+ "grad_norm": 2.3619892597198486,
418
+ "learning_rate": 9.455864570737607e-06,
419
+ "loss": 0.2478,
420
+ "step": 1400
421
+ },
422
+ {
423
+ "epoch": 5.0,
424
+ "grad_norm": 1.2342629432678223,
425
+ "learning_rate": 9.44074969770254e-06,
426
+ "loss": 0.261,
427
+ "step": 1425
428
+ },
429
+ {
430
+ "epoch": 5.087950747581354,
431
+ "grad_norm": 1.9633418321609497,
432
+ "learning_rate": 9.425634824667474e-06,
433
+ "loss": 0.1903,
434
+ "step": 1450
435
+ },
436
+ {
437
+ "epoch": 5.175901495162709,
438
+ "grad_norm": 2.1683809757232666,
439
+ "learning_rate": 9.410519951632408e-06,
440
+ "loss": 0.1858,
441
+ "step": 1475
442
+ },
443
+ {
444
+ "epoch": 5.263852242744063,
445
+ "grad_norm": 2.2380452156066895,
446
+ "learning_rate": 9.395405078597341e-06,
447
+ "loss": 0.1881,
448
+ "step": 1500
449
+ },
450
+ {
451
+ "epoch": 5.263852242744063,
452
+ "eval_loss": 0.46944713592529297,
453
+ "eval_runtime": 1317.879,
454
+ "eval_samples_per_second": 1.219,
455
+ "eval_steps_per_second": 0.305,
456
+ "eval_wer": 0.3842995839112344,
457
+ "step": 1500
458
+ },
459
+ {
460
+ "epoch": 5.351802990325417,
461
+ "grad_norm": 2.387603998184204,
462
+ "learning_rate": 9.380290205562274e-06,
463
+ "loss": 0.1853,
464
+ "step": 1525
465
+ },
466
+ {
467
+ "epoch": 5.439753737906772,
468
+ "grad_norm": 2.206265926361084,
469
+ "learning_rate": 9.365175332527208e-06,
470
+ "loss": 0.1899,
471
+ "step": 1550
472
+ },
473
+ {
474
+ "epoch": 5.527704485488127,
475
+ "grad_norm": 2.6551010608673096,
476
+ "learning_rate": 9.35006045949214e-06,
477
+ "loss": 0.1897,
478
+ "step": 1575
479
+ },
480
+ {
481
+ "epoch": 5.615655233069481,
482
+ "grad_norm": 2.449110746383667,
483
+ "learning_rate": 9.334945586457075e-06,
484
+ "loss": 0.1899,
485
+ "step": 1600
486
+ },
487
+ {
488
+ "epoch": 5.703605980650836,
489
+ "grad_norm": 2.394658327102661,
490
+ "learning_rate": 9.319830713422008e-06,
491
+ "loss": 0.2011,
492
+ "step": 1625
493
+ },
494
+ {
495
+ "epoch": 5.79155672823219,
496
+ "grad_norm": 2.371216297149658,
497
+ "learning_rate": 9.304715840386942e-06,
498
+ "loss": 0.1863,
499
+ "step": 1650
500
+ },
501
+ {
502
+ "epoch": 5.879507475813544,
503
+ "grad_norm": 2.354520320892334,
504
+ "learning_rate": 9.289600967351875e-06,
505
+ "loss": 0.1965,
506
+ "step": 1675
507
+ },
508
+ {
509
+ "epoch": 5.967458223394899,
510
+ "grad_norm": 2.611523389816284,
511
+ "learning_rate": 9.274486094316808e-06,
512
+ "loss": 0.1893,
513
+ "step": 1700
514
+ },
515
+ {
516
+ "epoch": 6.052770448548813,
517
+ "grad_norm": 2.138335704803467,
518
+ "learning_rate": 9.259371221281742e-06,
519
+ "loss": 0.1508,
520
+ "step": 1725
521
+ },
522
+ {
523
+ "epoch": 6.140721196130167,
524
+ "grad_norm": 2.3280434608459473,
525
+ "learning_rate": 9.244256348246675e-06,
526
+ "loss": 0.1262,
527
+ "step": 1750
528
+ },
529
+ {
530
+ "epoch": 6.228671943711522,
531
+ "grad_norm": 2.500696897506714,
532
+ "learning_rate": 9.22914147521161e-06,
533
+ "loss": 0.1254,
534
+ "step": 1775
535
+ },
536
+ {
537
+ "epoch": 6.316622691292876,
538
+ "grad_norm": 2.3168251514434814,
539
+ "learning_rate": 9.214026602176542e-06,
540
+ "loss": 0.1291,
541
+ "step": 1800
542
+ },
543
+ {
544
+ "epoch": 6.40457343887423,
545
+ "grad_norm": 2.113083839416504,
546
+ "learning_rate": 9.198911729141476e-06,
547
+ "loss": 0.1312,
548
+ "step": 1825
549
+ },
550
+ {
551
+ "epoch": 6.492524186455585,
552
+ "grad_norm": 2.849163293838501,
553
+ "learning_rate": 9.183796856106409e-06,
554
+ "loss": 0.1295,
555
+ "step": 1850
556
+ },
557
+ {
558
+ "epoch": 6.580474934036939,
559
+ "grad_norm": 2.3607892990112305,
560
+ "learning_rate": 9.168681983071342e-06,
561
+ "loss": 0.1346,
562
+ "step": 1875
563
+ },
564
+ {
565
+ "epoch": 6.668425681618293,
566
+ "grad_norm": 2.473125457763672,
567
+ "learning_rate": 9.153567110036276e-06,
568
+ "loss": 0.1267,
569
+ "step": 1900
570
+ },
571
+ {
572
+ "epoch": 6.756376429199648,
573
+ "grad_norm": 2.5916550159454346,
574
+ "learning_rate": 9.138452237001209e-06,
575
+ "loss": 0.1365,
576
+ "step": 1925
577
+ },
578
+ {
579
+ "epoch": 6.844327176781003,
580
+ "grad_norm": 2.339555263519287,
581
+ "learning_rate": 9.123337363966143e-06,
582
+ "loss": 0.1354,
583
+ "step": 1950
584
+ },
585
+ {
586
+ "epoch": 6.932277924362357,
587
+ "grad_norm": 2.2868764400482178,
588
+ "learning_rate": 9.108222490931078e-06,
589
+ "loss": 0.1317,
590
+ "step": 1975
591
+ },
592
+ {
593
+ "epoch": 7.017590149516271,
594
+ "grad_norm": 2.6691622734069824,
595
+ "learning_rate": 9.09310761789601e-06,
596
+ "loss": 0.124,
597
+ "step": 2000
598
+ },
599
+ {
600
+ "epoch": 7.017590149516271,
601
+ "eval_loss": 0.5287390351295471,
602
+ "eval_runtime": 1307.9312,
603
+ "eval_samples_per_second": 1.229,
604
+ "eval_steps_per_second": 0.307,
605
+ "eval_wer": 0.3951733703190014,
606
+ "step": 2000
607
+ },
608
+ {
609
+ "epoch": 7.105540897097625,
610
+ "grad_norm": 1.5589745044708252,
611
+ "learning_rate": 9.077992744860945e-06,
612
+ "loss": 0.0803,
613
+ "step": 2025
614
+ },
615
+ {
616
+ "epoch": 7.193491644678979,
617
+ "grad_norm": 1.9738343954086304,
618
+ "learning_rate": 9.062877871825877e-06,
619
+ "loss": 0.0771,
620
+ "step": 2050
621
+ },
622
+ {
623
+ "epoch": 7.281442392260335,
624
+ "grad_norm": 2.0110316276550293,
625
+ "learning_rate": 9.047762998790812e-06,
626
+ "loss": 0.0782,
627
+ "step": 2075
628
+ },
629
+ {
630
+ "epoch": 7.369393139841689,
631
+ "grad_norm": 2.6129631996154785,
632
+ "learning_rate": 9.032648125755745e-06,
633
+ "loss": 0.0823,
634
+ "step": 2100
635
+ },
636
+ {
637
+ "epoch": 7.457343887423043,
638
+ "grad_norm": 2.0788533687591553,
639
+ "learning_rate": 9.017533252720679e-06,
640
+ "loss": 0.0798,
641
+ "step": 2125
642
+ },
643
+ {
644
+ "epoch": 7.545294635004398,
645
+ "grad_norm": 2.0151889324188232,
646
+ "learning_rate": 9.002418379685612e-06,
647
+ "loss": 0.084,
648
+ "step": 2150
649
+ },
650
+ {
651
+ "epoch": 7.633245382585752,
652
+ "grad_norm": 1.7634187936782837,
653
+ "learning_rate": 8.987303506650546e-06,
654
+ "loss": 0.0811,
655
+ "step": 2175
656
+ },
657
+ {
658
+ "epoch": 7.721196130167106,
659
+ "grad_norm": 2.6911752223968506,
660
+ "learning_rate": 8.972188633615479e-06,
661
+ "loss": 0.0856,
662
+ "step": 2200
663
+ },
664
+ {
665
+ "epoch": 7.809146877748461,
666
+ "grad_norm": 2.2355527877807617,
667
+ "learning_rate": 8.957073760580411e-06,
668
+ "loss": 0.0893,
669
+ "step": 2225
670
+ },
671
+ {
672
+ "epoch": 7.897097625329815,
673
+ "grad_norm": 2.63678240776062,
674
+ "learning_rate": 8.941958887545346e-06,
675
+ "loss": 0.0794,
676
+ "step": 2250
677
+ },
678
+ {
679
+ "epoch": 7.985048372911169,
680
+ "grad_norm": 2.3199474811553955,
681
+ "learning_rate": 8.926844014510278e-06,
682
+ "loss": 0.0834,
683
+ "step": 2275
684
+ },
685
+ {
686
+ "epoch": 8.070360598065083,
687
+ "grad_norm": 1.5594936609268188,
688
+ "learning_rate": 8.911729141475213e-06,
689
+ "loss": 0.0518,
690
+ "step": 2300
691
+ },
692
+ {
693
+ "epoch": 8.158311345646439,
694
+ "grad_norm": 1.5886763334274292,
695
+ "learning_rate": 8.896614268440146e-06,
696
+ "loss": 0.0446,
697
+ "step": 2325
698
+ },
699
+ {
700
+ "epoch": 8.246262093227793,
701
+ "grad_norm": 1.9491994380950928,
702
+ "learning_rate": 8.881499395405078e-06,
703
+ "loss": 0.0459,
704
+ "step": 2350
705
+ },
706
+ {
707
+ "epoch": 8.334212840809148,
708
+ "grad_norm": 2.1820008754730225,
709
+ "learning_rate": 8.866384522370013e-06,
710
+ "loss": 0.0489,
711
+ "step": 2375
712
+ },
713
+ {
714
+ "epoch": 8.422163588390502,
715
+ "grad_norm": 1.6546351909637451,
716
+ "learning_rate": 8.851269649334945e-06,
717
+ "loss": 0.0479,
718
+ "step": 2400
719
+ },
720
+ {
721
+ "epoch": 8.510114335971856,
722
+ "grad_norm": 2.204155921936035,
723
+ "learning_rate": 8.83615477629988e-06,
724
+ "loss": 0.0467,
725
+ "step": 2425
726
+ },
727
+ {
728
+ "epoch": 8.59806508355321,
729
+ "grad_norm": 1.9652348756790161,
730
+ "learning_rate": 8.821039903264812e-06,
731
+ "loss": 0.049,
732
+ "step": 2450
733
+ },
734
+ {
735
+ "epoch": 8.686015831134565,
736
+ "grad_norm": 2.3371407985687256,
737
+ "learning_rate": 8.805925030229747e-06,
738
+ "loss": 0.0472,
739
+ "step": 2475
740
+ },
741
+ {
742
+ "epoch": 8.77396657871592,
743
+ "grad_norm": 1.6923274993896484,
744
+ "learning_rate": 8.79081015719468e-06,
745
+ "loss": 0.05,
746
+ "step": 2500
747
+ },
748
+ {
749
+ "epoch": 8.77396657871592,
750
+ "eval_loss": 0.6124064326286316,
751
+ "eval_runtime": 1240.0461,
752
+ "eval_samples_per_second": 1.296,
753
+ "eval_steps_per_second": 0.324,
754
+ "eval_wer": 0.4023855755894591,
755
+ "step": 2500
756
+ },
757
+ {
758
+ "epoch": 8.861917326297274,
759
+ "grad_norm": 1.9137698411941528,
760
+ "learning_rate": 8.775695284159614e-06,
761
+ "loss": 0.0457,
762
+ "step": 2525
763
+ },
764
+ {
765
+ "epoch": 8.949868073878628,
766
+ "grad_norm": 2.437100410461426,
767
+ "learning_rate": 8.760580411124547e-06,
768
+ "loss": 0.0528,
769
+ "step": 2550
770
+ },
771
+ {
772
+ "epoch": 9.035180299032541,
773
+ "grad_norm": 1.3090524673461914,
774
+ "learning_rate": 8.745465538089481e-06,
775
+ "loss": 0.0365,
776
+ "step": 2575
777
+ },
778
+ {
779
+ "epoch": 9.123131046613896,
780
+ "grad_norm": 1.3101695775985718,
781
+ "learning_rate": 8.730350665054414e-06,
782
+ "loss": 0.0263,
783
+ "step": 2600
784
+ },
785
+ {
786
+ "epoch": 9.21108179419525,
787
+ "grad_norm": 1.5315958261489868,
788
+ "learning_rate": 8.715235792019348e-06,
789
+ "loss": 0.0251,
790
+ "step": 2625
791
+ },
792
+ {
793
+ "epoch": 9.299032541776604,
794
+ "grad_norm": 1.448565125465393,
795
+ "learning_rate": 8.700120918984281e-06,
796
+ "loss": 0.0279,
797
+ "step": 2650
798
+ },
799
+ {
800
+ "epoch": 9.386983289357959,
801
+ "grad_norm": 1.5439013242721558,
802
+ "learning_rate": 8.685006045949215e-06,
803
+ "loss": 0.0272,
804
+ "step": 2675
805
+ },
806
+ {
807
+ "epoch": 9.474934036939313,
808
+ "grad_norm": 1.4557873010635376,
809
+ "learning_rate": 8.669891172914148e-06,
810
+ "loss": 0.0266,
811
+ "step": 2700
812
+ },
813
+ {
814
+ "epoch": 9.56288478452067,
815
+ "grad_norm": 1.4524966478347778,
816
+ "learning_rate": 8.654776299879082e-06,
817
+ "loss": 0.0279,
818
+ "step": 2725
819
+ },
820
+ {
821
+ "epoch": 9.650835532102024,
822
+ "grad_norm": 1.9824354648590088,
823
+ "learning_rate": 8.639661426844015e-06,
824
+ "loss": 0.029,
825
+ "step": 2750
826
+ },
827
+ {
828
+ "epoch": 9.738786279683378,
829
+ "grad_norm": 1.7018193006515503,
830
+ "learning_rate": 8.62454655380895e-06,
831
+ "loss": 0.0279,
832
+ "step": 2775
833
+ },
834
+ {
835
+ "epoch": 9.826737027264732,
836
+ "grad_norm": 1.530937671661377,
837
+ "learning_rate": 8.609431680773882e-06,
838
+ "loss": 0.0297,
839
+ "step": 2800
840
+ },
841
+ {
842
+ "epoch": 9.914687774846087,
843
+ "grad_norm": 1.9330962896347046,
844
+ "learning_rate": 8.594316807738817e-06,
845
+ "loss": 0.0274,
846
+ "step": 2825
847
+ },
848
+ {
849
+ "epoch": 10.0,
850
+ "grad_norm": 0.5909117460250854,
851
+ "learning_rate": 8.57920193470375e-06,
852
+ "loss": 0.0272,
853
+ "step": 2850
854
+ },
855
+ {
856
+ "epoch": 10.087950747581354,
857
+ "grad_norm": 0.8719807267189026,
858
+ "learning_rate": 8.564087061668682e-06,
859
+ "loss": 0.0164,
860
+ "step": 2875
861
+ },
862
+ {
863
+ "epoch": 10.175901495162709,
864
+ "grad_norm": 1.261264681816101,
865
+ "learning_rate": 8.548972188633616e-06,
866
+ "loss": 0.0161,
867
+ "step": 2900
868
+ },
869
+ {
870
+ "epoch": 10.263852242744063,
871
+ "grad_norm": 1.44869863986969,
872
+ "learning_rate": 8.533857315598549e-06,
873
+ "loss": 0.0168,
874
+ "step": 2925
875
+ },
876
+ {
877
+ "epoch": 10.351802990325417,
878
+ "grad_norm": 1.14552903175354,
879
+ "learning_rate": 8.518742442563483e-06,
880
+ "loss": 0.0152,
881
+ "step": 2950
882
+ },
883
+ {
884
+ "epoch": 10.439753737906772,
885
+ "grad_norm": 1.1596709489822388,
886
+ "learning_rate": 8.503627569528416e-06,
887
+ "loss": 0.0159,
888
+ "step": 2975
889
+ },
890
+ {
891
+ "epoch": 10.527704485488126,
892
+ "grad_norm": 1.2538233995437622,
893
+ "learning_rate": 8.48851269649335e-06,
894
+ "loss": 0.0166,
895
+ "step": 3000
896
+ },
897
+ {
898
+ "epoch": 10.527704485488126,
899
+ "eval_loss": 0.6773013472557068,
900
+ "eval_runtime": 1227.1249,
901
+ "eval_samples_per_second": 1.31,
902
+ "eval_steps_per_second": 0.328,
903
+ "eval_wer": 0.39892741562644474,
904
+ "step": 3000
905
+ },
906
+ {
907
+ "epoch": 10.61565523306948,
908
+ "grad_norm": 1.7400544881820679,
909
+ "learning_rate": 8.473397823458283e-06,
910
+ "loss": 0.0163,
911
+ "step": 3025
912
+ },
913
+ {
914
+ "epoch": 10.703605980650835,
915
+ "grad_norm": 1.4166711568832397,
916
+ "learning_rate": 8.458282950423216e-06,
917
+ "loss": 0.0184,
918
+ "step": 3050
919
+ },
920
+ {
921
+ "epoch": 10.79155672823219,
922
+ "grad_norm": 1.4197386503219604,
923
+ "learning_rate": 8.44316807738815e-06,
924
+ "loss": 0.0172,
925
+ "step": 3075
926
+ },
927
+ {
928
+ "epoch": 10.879507475813543,
929
+ "grad_norm": 1.5135834217071533,
930
+ "learning_rate": 8.428053204353083e-06,
931
+ "loss": 0.0188,
932
+ "step": 3100
933
+ },
934
+ {
935
+ "epoch": 10.9674582233949,
936
+ "grad_norm": 1.3978145122528076,
937
+ "learning_rate": 8.412938331318017e-06,
938
+ "loss": 0.018,
939
+ "step": 3125
940
+ },
941
+ {
942
+ "epoch": 11.052770448548813,
943
+ "grad_norm": 0.9367815852165222,
944
+ "learning_rate": 8.39782345828295e-06,
945
+ "loss": 0.0144,
946
+ "step": 3150
947
+ },
948
+ {
949
+ "epoch": 11.140721196130167,
950
+ "grad_norm": 0.6169752478599548,
951
+ "learning_rate": 8.382708585247885e-06,
952
+ "loss": 0.0101,
953
+ "step": 3175
954
+ },
955
+ {
956
+ "epoch": 11.228671943711522,
957
+ "grad_norm": 0.9121926426887512,
958
+ "learning_rate": 8.367593712212819e-06,
959
+ "loss": 0.0128,
960
+ "step": 3200
961
+ },
962
+ {
963
+ "epoch": 11.316622691292876,
964
+ "grad_norm": 1.111951231956482,
965
+ "learning_rate": 8.352478839177752e-06,
966
+ "loss": 0.0119,
967
+ "step": 3225
968
+ },
969
+ {
970
+ "epoch": 11.40457343887423,
971
+ "grad_norm": 1.3314982652664185,
972
+ "learning_rate": 8.337363966142686e-06,
973
+ "loss": 0.0118,
974
+ "step": 3250
975
+ },
976
+ {
977
+ "epoch": 11.492524186455585,
978
+ "grad_norm": 1.206529140472412,
979
+ "learning_rate": 8.322249093107619e-06,
980
+ "loss": 0.0117,
981
+ "step": 3275
982
+ },
983
+ {
984
+ "epoch": 11.580474934036939,
985
+ "grad_norm": 1.4378085136413574,
986
+ "learning_rate": 8.307134220072553e-06,
987
+ "loss": 0.0116,
988
+ "step": 3300
989
+ },
990
+ {
991
+ "epoch": 11.668425681618293,
992
+ "grad_norm": 1.2930444478988647,
993
+ "learning_rate": 8.292019347037486e-06,
994
+ "loss": 0.0119,
995
+ "step": 3325
996
+ },
997
+ {
998
+ "epoch": 11.756376429199648,
999
+ "grad_norm": 1.2823526859283447,
1000
+ "learning_rate": 8.27690447400242e-06,
1001
+ "loss": 0.012,
1002
+ "step": 3350
1003
+ },
1004
+ {
1005
+ "epoch": 11.844327176781002,
1006
+ "grad_norm": 1.088985800743103,
1007
+ "learning_rate": 8.261789600967353e-06,
1008
+ "loss": 0.012,
1009
+ "step": 3375
1010
+ },
1011
+ {
1012
+ "epoch": 11.932277924362356,
1013
+ "grad_norm": 2.2263076305389404,
1014
+ "learning_rate": 8.246674727932286e-06,
1015
+ "loss": 0.0126,
1016
+ "step": 3400
1017
+ },
1018
+ {
1019
+ "epoch": 12.017590149516272,
1020
+ "grad_norm": 1.0628622770309448,
1021
+ "learning_rate": 8.23155985489722e-06,
1022
+ "loss": 0.0106,
1023
+ "step": 3425
1024
+ },
1025
+ {
1026
+ "epoch": 12.105540897097626,
1027
+ "grad_norm": 1.4298282861709595,
1028
+ "learning_rate": 8.216444981862153e-06,
1029
+ "loss": 0.0077,
1030
+ "step": 3450
1031
+ },
1032
+ {
1033
+ "epoch": 12.19349164467898,
1034
+ "grad_norm": 1.0117568969726562,
1035
+ "learning_rate": 8.201330108827087e-06,
1036
+ "loss": 0.009,
1037
+ "step": 3475
1038
+ },
1039
+ {
1040
+ "epoch": 12.281442392260335,
1041
+ "grad_norm": 1.1720714569091797,
1042
+ "learning_rate": 8.18621523579202e-06,
1043
+ "loss": 0.0083,
1044
+ "step": 3500
1045
+ },
1046
+ {
1047
+ "epoch": 12.281442392260335,
1048
+ "eval_loss": 0.7468146085739136,
1049
+ "eval_runtime": 1232.127,
1050
+ "eval_samples_per_second": 1.304,
1051
+ "eval_steps_per_second": 0.326,
1052
+ "eval_wer": 0.40171983356449376,
1053
+ "step": 3500
1054
+ },
1055
+ {
1056
+ "epoch": 12.281442392260335,
1057
+ "step": 3500,
1058
+ "total_flos": 3.2234989215744e+19,
1059
+ "train_loss": 0.28376776227780753,
1060
+ "train_runtime": 61744.3298,
1061
+ "train_samples_per_second": 8.838,
1062
+ "train_steps_per_second": 0.276
1063
+ }
1064
+ ],
1065
+ "logging_steps": 25,
1066
+ "max_steps": 17040,
1067
+ "num_input_tokens_seen": 0,
1068
+ "num_train_epochs": 60,
1069
+ "save_steps": 500,
1070
+ "stateful_callbacks": {
1071
+ "EarlyStoppingCallback": {
1072
+ "args": {
1073
+ "early_stopping_patience": 5,
1074
+ "early_stopping_threshold": 0.0
1075
+ },
1076
+ "attributes": {
1077
+ "early_stopping_patience_counter": 5
1078
+ }
1079
+ },
1080
+ "TrainerControl": {
1081
+ "args": {
1082
+ "should_epoch_stop": false,
1083
+ "should_evaluate": false,
1084
+ "should_log": false,
1085
+ "should_save": true,
1086
+ "should_training_stop": true
1087
+ },
1088
+ "attributes": {}
1089
+ }
1090
+ },
1091
+ "total_flos": 3.2234989215744e+19,
1092
+ "train_batch_size": 4,
1093
+ "trial_name": null,
1094
+ "trial_params": null
1095
+ }