nefasto commited on
Commit
264b714
1 Parent(s): 5c5cbe7

End of training

Browse files
all_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.06,
3
+ "eval_loss": 0.25342342257499695,
4
+ "eval_runtime": 5321.0317,
5
+ "eval_samples_per_second": 2.82,
6
+ "eval_steps_per_second": 0.088,
7
+ "eval_wer": 12.303981501169467,
8
+ "train_loss": 0.1507516227165858,
9
+ "train_runtime": 84190.0984,
10
+ "train_samples_per_second": 4.561,
11
+ "train_steps_per_second": 0.071
12
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.06,
3
+ "eval_loss": 0.25342342257499695,
4
+ "eval_runtime": 5321.0317,
5
+ "eval_samples_per_second": 2.82,
6
+ "eval_steps_per_second": 0.088,
7
+ "eval_wer": 12.303981501169467
8
+ }
log CHANGED
@@ -33761,4 +33761,42 @@ remote: LFS file scan complete.
33761
  To https://huggingface.co/nefasto/whisper-small-it
33762
  1875cbe..950168c main -> main
33763
 
33764
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
33761
  To https://huggingface.co/nefasto/whisper-small-it
33762
  1875cbe..950168c main -> main
33763
 
33764
+
33765
+ To https://huggingface.co/nefasto/whisper-small-it
33766
+ 950168c..5c5cbe7 main -> main
33767
+
33768
+ 12/15/2022 17:21:37 - WARNING - huggingface_hub.repository - To https://huggingface.co/nefasto/whisper-small-it
33769
+ 950168c..5c5cbe7 main -> main
33770
+
33771
+ ***** train metrics *****
33772
+ epoch = 12.06
33773
+ train_loss = 0.1508
33774
+ train_runtime = 23:23:10.09
33775
+ train_samples_per_second = 4.561
33776
+ train_steps_per_second = 0.071
33777
+ 12/15/2022 17:21:40 - INFO - __main__ - *** Evaluate ***
33778
+ [INFO|trainer.py:2955] 2022-12-15 17:21:40,699 >> ***** Running Evaluation *****
33779
+ [INFO|trainer.py:2959] 2022-12-15 17:21:40,700 >> Num examples: Unknown
33780
+ [INFO|trainer.py:2960] 2022-12-15 17:21:40,700 >> Batch size = 32
33781
+
33782
+ [INFO|trainer_utils.py:689] 2022-12-15 17:21:44,618 >> The following columns in the evaluation set don't have a corresponding argument in `WhisperForConditionalGeneration.forward` and have been ignored: input_length. If input_length are not expected by `WhisperForConditionalGeneration.forward`, you can safely ignore this message.
33783
+ 12/15/2022 18:01:30 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [1/20]
33784
+ 12/15/2022 18:01:35 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [2/20]
33785
+ 12/15/2022 18:01:50 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [3/20]
33786
+ 12/15/2022 18:02:05 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [4/20]
33787
+ 12/15/2022 18:02:20 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [5/20]
33788
+ 12/15/2022 18:02:35 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [6/20]
33789
+ 12/15/2022 18:02:50 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [7/20]
33790
+ 12/15/2022 18:03:05 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [8/20]
33791
+ 12/15/2022 18:03:20 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [9/20]
33792
+ 12/15/2022 18:03:35 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [10/20]
33793
+ 12/15/2022 18:03:50 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [11/20]
33794
+ 12/15/2022 18:04:05 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [12/20]
33795
+ 12/15/2022 18:04:20 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [13/20]
33796
+ 12/15/2022 18:04:35 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [14/20]
33797
+ 12/15/2022 18:04:40 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [15/20]
33798
+ 12/15/2022 18:04:45 - WARNING - datasets.download.streaming_download_manager - Got disconnected from remote data host. Retrying in 5sec [16/20]
33799
+ [INFO|trainer.py:2700] 2022-12-15 18:50:21,733 >> Saving model checkpoint to ./
33800
+ [INFO|configuration_utils.py:447] 2022-12-15 18:50:21,734 >> Configuration saved in ./config.json
33801
+ [INFO|modeling_utils.py:1680] 2022-12-15 18:50:22,805 >> Model weights saved in ./pytorch_model.bin
33802
+ [INFO|feature_extraction_utils.py:368] 2022-12-15 18:50:22,823 >> Feature extractor saved in ./preprocessor_config.json
runs/Dec14_17-57-28_129-213-29-60/events.out.tfevents.1671130221.129-213-29-60.71149.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb55b6c1742eb52cbc3317e8b453958a6d2918b6cff3a7cab4232250bf636656
3
+ size 358
train_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.06,
3
+ "train_loss": 0.1507516227165858,
4
+ "train_runtime": 84190.0984,
5
+ "train_samples_per_second": 4.561,
6
+ "train_steps_per_second": 0.071
7
+ }
trainer_state.json ADDED
@@ -0,0 +1,1519 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 12.303981501169467,
3
+ "best_model_checkpoint": "./checkpoint-4000",
4
+ "epoch": 12.062,
5
+ "global_step": 6000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 3.52e-07,
13
+ "loss": 1.0655,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 7.52e-07,
19
+ "loss": 0.9756,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.152e-06,
25
+ "loss": 0.7902,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.02,
30
+ "learning_rate": 1.552e-06,
31
+ "loss": 0.4928,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.02,
36
+ "learning_rate": 1.9519999999999997e-06,
37
+ "loss": 0.3652,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.03,
42
+ "learning_rate": 2.3519999999999997e-06,
43
+ "loss": 0.3222,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.03,
48
+ "learning_rate": 2.7519999999999997e-06,
49
+ "loss": 0.3031,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.03,
54
+ "learning_rate": 3.152e-06,
55
+ "loss": 0.3682,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.04,
60
+ "learning_rate": 3.552e-06,
61
+ "loss": 0.3873,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.04,
66
+ "learning_rate": 3.952e-06,
67
+ "loss": 0.3546,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.05,
72
+ "learning_rate": 4.352e-06,
73
+ "loss": 0.3597,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.05,
78
+ "learning_rate": 4.7519999999999996e-06,
79
+ "loss": 0.3844,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.05,
84
+ "learning_rate": 5.152e-06,
85
+ "loss": 0.3962,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.06,
90
+ "learning_rate": 5.5519999999999995e-06,
91
+ "loss": 0.3593,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.06,
96
+ "learning_rate": 5.952e-06,
97
+ "loss": 0.3424,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.07,
102
+ "learning_rate": 6.352e-06,
103
+ "loss": 0.348,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.07,
108
+ "learning_rate": 6.751999999999999e-06,
109
+ "loss": 0.3214,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.07,
114
+ "learning_rate": 7.1519999999999995e-06,
115
+ "loss": 0.2422,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 1.0,
120
+ "learning_rate": 7.551999999999999e-06,
121
+ "loss": 0.2384,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 1.01,
126
+ "learning_rate": 7.952e-06,
127
+ "loss": 0.2678,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 1.01,
132
+ "learning_rate": 7.967999999999999e-06,
133
+ "loss": 0.2878,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 1.01,
138
+ "learning_rate": 7.931636363636364e-06,
139
+ "loss": 0.2682,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 1.02,
144
+ "learning_rate": 7.895272727272728e-06,
145
+ "loss": 0.2454,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 1.02,
150
+ "learning_rate": 7.85890909090909e-06,
151
+ "loss": 0.2459,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 1.03,
156
+ "learning_rate": 7.822545454545454e-06,
157
+ "loss": 0.2372,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 1.03,
162
+ "learning_rate": 7.786181818181817e-06,
163
+ "loss": 0.2726,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 1.03,
168
+ "learning_rate": 7.749818181818181e-06,
169
+ "loss": 0.1999,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 1.04,
174
+ "learning_rate": 7.713454545454545e-06,
175
+ "loss": 0.2111,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 1.04,
180
+ "learning_rate": 7.677090909090908e-06,
181
+ "loss": 0.2177,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 1.05,
186
+ "learning_rate": 7.640727272727272e-06,
187
+ "loss": 0.2232,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 1.05,
192
+ "learning_rate": 7.6043636363636365e-06,
193
+ "loss": 0.2485,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 1.06,
198
+ "learning_rate": 7.567999999999999e-06,
199
+ "loss": 0.2419,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 1.06,
204
+ "learning_rate": 7.531636363636363e-06,
205
+ "loss": 0.2476,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 1.06,
210
+ "learning_rate": 7.495272727272727e-06,
211
+ "loss": 0.2193,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 1.07,
216
+ "learning_rate": 7.45890909090909e-06,
217
+ "loss": 0.1901,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 1.07,
222
+ "learning_rate": 7.422545454545454e-06,
223
+ "loss": 0.1994,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 1.08,
228
+ "learning_rate": 7.386181818181818e-06,
229
+ "loss": 0.2006,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 2.0,
234
+ "learning_rate": 7.3498181818181814e-06,
235
+ "loss": 0.2218,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 2.01,
240
+ "learning_rate": 7.313454545454545e-06,
241
+ "loss": 0.2398,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 2.01,
246
+ "learning_rate": 7.277090909090909e-06,
247
+ "loss": 0.2737,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 2.01,
252
+ "eval_loss": 0.27280035614967346,
253
+ "eval_runtime": 5409.8244,
254
+ "eval_samples_per_second": 2.773,
255
+ "eval_steps_per_second": 0.087,
256
+ "eval_wer": 13.409658728471188,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 2.01,
261
+ "learning_rate": 7.2407272727272725e-06,
262
+ "loss": 0.2345,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 2.02,
267
+ "learning_rate": 7.204363636363636e-06,
268
+ "loss": 0.2426,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 2.02,
273
+ "learning_rate": 7.168e-06,
274
+ "loss": 0.2022,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 2.03,
279
+ "learning_rate": 7.1316363636363635e-06,
280
+ "loss": 0.2105,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 2.03,
285
+ "learning_rate": 7.095272727272726e-06,
286
+ "loss": 0.1926,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 2.04,
291
+ "learning_rate": 7.058909090909091e-06,
292
+ "loss": 0.1891,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 2.04,
297
+ "learning_rate": 7.0225454545454545e-06,
298
+ "loss": 0.1971,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 2.04,
303
+ "learning_rate": 6.986181818181818e-06,
304
+ "loss": 0.2074,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 2.05,
309
+ "learning_rate": 6.949818181818181e-06,
310
+ "loss": 0.2283,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 2.05,
315
+ "learning_rate": 6.913454545454545e-06,
316
+ "loss": 0.2194,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 2.06,
321
+ "learning_rate": 6.877090909090909e-06,
322
+ "loss": 0.1934,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 2.06,
327
+ "learning_rate": 6.840727272727273e-06,
328
+ "loss": 0.1857,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 2.06,
333
+ "learning_rate": 6.804363636363636e-06,
334
+ "loss": 0.1681,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 2.07,
339
+ "learning_rate": 6.767999999999999e-06,
340
+ "loss": 0.1534,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 2.07,
345
+ "learning_rate": 6.731636363636363e-06,
346
+ "loss": 0.1522,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 2.08,
351
+ "learning_rate": 6.695272727272728e-06,
352
+ "loss": 0.1874,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 3.0,
357
+ "learning_rate": 6.658909090909091e-06,
358
+ "loss": 0.2014,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 3.01,
363
+ "learning_rate": 6.622545454545454e-06,
364
+ "loss": 0.1817,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 3.01,
369
+ "learning_rate": 6.586181818181818e-06,
370
+ "loss": 0.1792,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 3.02,
375
+ "learning_rate": 6.5498181818181815e-06,
376
+ "loss": 0.1578,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 3.02,
381
+ "learning_rate": 6.513454545454545e-06,
382
+ "loss": 0.1522,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 3.02,
387
+ "learning_rate": 6.477090909090909e-06,
388
+ "loss": 0.1457,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 3.03,
393
+ "learning_rate": 6.4407272727272725e-06,
394
+ "loss": 0.1482,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 3.03,
399
+ "learning_rate": 6.404363636363636e-06,
400
+ "loss": 0.1669,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 3.04,
405
+ "learning_rate": 6.368e-06,
406
+ "loss": 0.2161,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 3.04,
411
+ "learning_rate": 6.331636363636363e-06,
412
+ "loss": 0.1928,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 3.04,
417
+ "learning_rate": 6.295272727272727e-06,
418
+ "loss": 0.1923,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 3.05,
423
+ "learning_rate": 6.258909090909091e-06,
424
+ "loss": 0.2077,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 3.05,
429
+ "learning_rate": 6.2225454545454546e-06,
430
+ "loss": 0.2226,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 3.06,
435
+ "learning_rate": 6.186181818181817e-06,
436
+ "loss": 0.2089,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 3.06,
441
+ "learning_rate": 6.149818181818181e-06,
442
+ "loss": 0.1904,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 3.07,
447
+ "learning_rate": 6.113454545454546e-06,
448
+ "loss": 0.1825,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 3.07,
453
+ "learning_rate": 6.077090909090909e-06,
454
+ "loss": 0.2006,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 3.07,
459
+ "learning_rate": 6.040727272727273e-06,
460
+ "loss": 0.1328,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 3.08,
465
+ "learning_rate": 6.004363636363636e-06,
466
+ "loss": 0.1086,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 4.0,
471
+ "learning_rate": 5.9679999999999994e-06,
472
+ "loss": 0.1554,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 4.01,
477
+ "learning_rate": 5.931636363636364e-06,
478
+ "loss": 0.1507,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 4.01,
483
+ "learning_rate": 5.895272727272728e-06,
484
+ "loss": 0.1845,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 4.02,
489
+ "learning_rate": 5.8589090909090905e-06,
490
+ "loss": 0.1394,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 4.02,
495
+ "learning_rate": 5.822545454545454e-06,
496
+ "loss": 0.1536,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 4.02,
501
+ "eval_loss": 0.2611413598060608,
502
+ "eval_runtime": 5136.3747,
503
+ "eval_samples_per_second": 2.921,
504
+ "eval_steps_per_second": 0.091,
505
+ "eval_wer": 12.989714012332554,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 4.02,
510
+ "learning_rate": 5.786181818181818e-06,
511
+ "loss": 0.1309,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 4.03,
516
+ "learning_rate": 5.7498181818181815e-06,
517
+ "loss": 0.1297,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 4.03,
522
+ "learning_rate": 5.713454545454545e-06,
523
+ "loss": 0.1217,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 4.04,
528
+ "learning_rate": 5.677090909090909e-06,
529
+ "loss": 0.1264,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 4.04,
534
+ "learning_rate": 5.6407272727272725e-06,
535
+ "loss": 0.1166,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 4.05,
540
+ "learning_rate": 5.604363636363636e-06,
541
+ "loss": 0.1529,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 4.05,
546
+ "learning_rate": 5.567999999999999e-06,
547
+ "loss": 0.1509,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 4.05,
552
+ "learning_rate": 5.5316363636363636e-06,
553
+ "loss": 0.1333,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 4.06,
558
+ "learning_rate": 5.495272727272727e-06,
559
+ "loss": 0.1253,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 4.06,
564
+ "learning_rate": 5.458909090909091e-06,
565
+ "loss": 0.1116,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 4.07,
570
+ "learning_rate": 5.422545454545454e-06,
571
+ "loss": 0.1084,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 4.07,
576
+ "learning_rate": 5.386181818181817e-06,
577
+ "loss": 0.0988,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 4.07,
582
+ "learning_rate": 5.349818181818182e-06,
583
+ "loss": 0.1122,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 5.0,
588
+ "learning_rate": 5.313454545454546e-06,
589
+ "loss": 0.1112,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 5.0,
594
+ "learning_rate": 5.277090909090909e-06,
595
+ "loss": 0.1192,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 5.01,
600
+ "learning_rate": 5.240727272727272e-06,
601
+ "loss": 0.1304,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 5.01,
606
+ "learning_rate": 5.204363636363636e-06,
607
+ "loss": 0.1241,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 5.02,
612
+ "learning_rate": 5.168e-06,
613
+ "loss": 0.1102,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 5.02,
618
+ "learning_rate": 5.131636363636364e-06,
619
+ "loss": 0.1181,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 5.03,
624
+ "learning_rate": 5.095272727272727e-06,
625
+ "loss": 0.1136,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 5.03,
630
+ "learning_rate": 5.0589090909090905e-06,
631
+ "loss": 0.1396,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 5.03,
636
+ "learning_rate": 5.022545454545454e-06,
637
+ "loss": 0.0885,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 5.04,
642
+ "learning_rate": 4.986181818181818e-06,
643
+ "loss": 0.1082,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 5.04,
648
+ "learning_rate": 4.9498181818181815e-06,
649
+ "loss": 0.1056,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 5.05,
654
+ "learning_rate": 4.913454545454545e-06,
655
+ "loss": 0.1102,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 5.05,
660
+ "learning_rate": 4.877090909090909e-06,
661
+ "loss": 0.123,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 5.05,
666
+ "learning_rate": 4.8407272727272726e-06,
667
+ "loss": 0.1202,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 5.06,
672
+ "learning_rate": 4.804363636363635e-06,
673
+ "loss": 0.1317,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 5.06,
678
+ "learning_rate": 4.768e-06,
679
+ "loss": 0.1099,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 5.07,
684
+ "learning_rate": 4.731636363636364e-06,
685
+ "loss": 0.0966,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 5.07,
690
+ "learning_rate": 4.695272727272727e-06,
691
+ "loss": 0.0976,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 5.08,
696
+ "learning_rate": 4.658909090909091e-06,
697
+ "loss": 0.1015,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 6.0,
702
+ "learning_rate": 4.622545454545454e-06,
703
+ "loss": 0.0906,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 6.01,
708
+ "learning_rate": 4.586181818181818e-06,
709
+ "loss": 0.0913,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 6.01,
714
+ "learning_rate": 4.549818181818182e-06,
715
+ "loss": 0.0917,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 6.01,
720
+ "learning_rate": 4.513454545454546e-06,
721
+ "loss": 0.0857,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 6.02,
726
+ "learning_rate": 4.4770909090909085e-06,
727
+ "loss": 0.0799,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 6.02,
732
+ "learning_rate": 4.440727272727272e-06,
733
+ "loss": 0.0794,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 6.03,
738
+ "learning_rate": 4.404363636363637e-06,
739
+ "loss": 0.0874,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 6.03,
744
+ "learning_rate": 4.368e-06,
745
+ "loss": 0.0905,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 6.03,
750
+ "eval_loss": 0.26857462525367737,
751
+ "eval_runtime": 5158.1008,
752
+ "eval_samples_per_second": 2.909,
753
+ "eval_steps_per_second": 0.091,
754
+ "eval_wer": 12.927253880501807,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 6.04,
759
+ "learning_rate": 4.331636363636363e-06,
760
+ "loss": 0.0671,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 6.04,
765
+ "learning_rate": 4.295272727272727e-06,
766
+ "loss": 0.0776,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 6.04,
771
+ "learning_rate": 4.2589090909090905e-06,
772
+ "loss": 0.0743,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 6.05,
777
+ "learning_rate": 4.222545454545454e-06,
778
+ "loss": 0.0817,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 6.05,
783
+ "learning_rate": 4.186181818181818e-06,
784
+ "loss": 0.088,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 6.06,
789
+ "learning_rate": 4.1498181818181816e-06,
790
+ "loss": 0.0901,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 6.06,
795
+ "learning_rate": 4.113454545454545e-06,
796
+ "loss": 0.094,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 6.06,
801
+ "learning_rate": 4.077090909090909e-06,
802
+ "loss": 0.0753,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 6.07,
807
+ "learning_rate": 4.040727272727273e-06,
808
+ "loss": 0.0676,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 6.07,
813
+ "learning_rate": 4.004363636363636e-06,
814
+ "loss": 0.0714,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 6.08,
819
+ "learning_rate": 3.968e-06,
820
+ "loss": 0.0729,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 7.0,
825
+ "learning_rate": 3.931636363636364e-06,
826
+ "loss": 0.1062,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 7.01,
831
+ "learning_rate": 3.895272727272727e-06,
832
+ "loss": 0.0978,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 7.01,
837
+ "learning_rate": 3.858909090909091e-06,
838
+ "loss": 0.0966,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 7.02,
843
+ "learning_rate": 3.822545454545455e-06,
844
+ "loss": 0.0864,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 7.02,
849
+ "learning_rate": 3.7861818181818183e-06,
850
+ "loss": 0.0805,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 7.02,
855
+ "learning_rate": 3.7498181818181816e-06,
856
+ "loss": 0.0788,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 7.03,
861
+ "learning_rate": 3.7134545454545453e-06,
862
+ "loss": 0.0825,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 7.03,
867
+ "learning_rate": 3.677090909090909e-06,
868
+ "loss": 0.0943,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 7.04,
873
+ "learning_rate": 3.6407272727272726e-06,
874
+ "loss": 0.1333,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 7.04,
879
+ "learning_rate": 3.604363636363636e-06,
880
+ "loss": 0.1137,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 7.04,
885
+ "learning_rate": 3.568e-06,
886
+ "loss": 0.1184,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 7.05,
891
+ "learning_rate": 3.5316363636363632e-06,
892
+ "loss": 0.1253,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 7.05,
897
+ "learning_rate": 3.4952727272727273e-06,
898
+ "loss": 0.1369,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 7.06,
903
+ "learning_rate": 3.4589090909090906e-06,
904
+ "loss": 0.1325,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 7.06,
909
+ "learning_rate": 3.4225454545454543e-06,
910
+ "loss": 0.1182,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 7.07,
915
+ "learning_rate": 3.3861818181818184e-06,
916
+ "loss": 0.112,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 7.07,
921
+ "learning_rate": 3.3498181818181816e-06,
922
+ "loss": 0.1322,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 7.07,
927
+ "learning_rate": 3.3134545454545457e-06,
928
+ "loss": 0.0821,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 7.08,
933
+ "learning_rate": 3.277090909090909e-06,
934
+ "loss": 0.0599,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 8.0,
939
+ "learning_rate": 3.2407272727272726e-06,
940
+ "loss": 0.1378,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 8.01,
945
+ "learning_rate": 3.2043636363636363e-06,
946
+ "loss": 0.1425,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 8.01,
951
+ "learning_rate": 3.168e-06,
952
+ "loss": 0.1534,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 8.02,
957
+ "learning_rate": 3.1316363636363632e-06,
958
+ "loss": 0.139,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 8.02,
963
+ "learning_rate": 3.0952727272727273e-06,
964
+ "loss": 0.1286,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 8.02,
969
+ "learning_rate": 3.0589090909090906e-06,
970
+ "loss": 0.1312,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 8.03,
975
+ "learning_rate": 3.0225454545454547e-06,
976
+ "loss": 0.1612,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 8.03,
981
+ "learning_rate": 2.986181818181818e-06,
982
+ "loss": 0.127,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 8.04,
987
+ "learning_rate": 2.9498181818181816e-06,
988
+ "loss": 0.1268,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 8.04,
993
+ "learning_rate": 2.9134545454545453e-06,
994
+ "loss": 0.1301,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 8.04,
999
+ "eval_loss": 0.25342342257499695,
1000
+ "eval_runtime": 5076.796,
1001
+ "eval_samples_per_second": 2.955,
1002
+ "eval_steps_per_second": 0.092,
1003
+ "eval_wer": 12.303981501169467,
1004
+ "step": 4000
1005
+ },
1006
+ {
1007
+ "epoch": 8.05,
1008
+ "learning_rate": 2.877090909090909e-06,
1009
+ "loss": 0.133,
1010
+ "step": 4025
1011
+ },
1012
+ {
1013
+ "epoch": 8.05,
1014
+ "learning_rate": 2.8407272727272722e-06,
1015
+ "loss": 0.1366,
1016
+ "step": 4050
1017
+ },
1018
+ {
1019
+ "epoch": 8.05,
1020
+ "learning_rate": 2.8043636363636363e-06,
1021
+ "loss": 0.1338,
1022
+ "step": 4075
1023
+ },
1024
+ {
1025
+ "epoch": 8.06,
1026
+ "learning_rate": 2.7679999999999996e-06,
1027
+ "loss": 0.1319,
1028
+ "step": 4100
1029
+ },
1030
+ {
1031
+ "epoch": 8.06,
1032
+ "learning_rate": 2.7316363636363637e-06,
1033
+ "loss": 0.1254,
1034
+ "step": 4125
1035
+ },
1036
+ {
1037
+ "epoch": 8.07,
1038
+ "learning_rate": 2.6952727272727274e-06,
1039
+ "loss": 0.1219,
1040
+ "step": 4150
1041
+ },
1042
+ {
1043
+ "epoch": 8.07,
1044
+ "learning_rate": 2.6589090909090906e-06,
1045
+ "loss": 0.1427,
1046
+ "step": 4175
1047
+ },
1048
+ {
1049
+ "epoch": 8.07,
1050
+ "learning_rate": 2.6225454545454547e-06,
1051
+ "loss": 0.1301,
1052
+ "step": 4200
1053
+ },
1054
+ {
1055
+ "epoch": 9.0,
1056
+ "learning_rate": 2.586181818181818e-06,
1057
+ "loss": 0.1234,
1058
+ "step": 4225
1059
+ },
1060
+ {
1061
+ "epoch": 9.0,
1062
+ "learning_rate": 2.549818181818182e-06,
1063
+ "loss": 0.0937,
1064
+ "step": 4250
1065
+ },
1066
+ {
1067
+ "epoch": 9.01,
1068
+ "learning_rate": 2.5134545454545453e-06,
1069
+ "loss": 0.0843,
1070
+ "step": 4275
1071
+ },
1072
+ {
1073
+ "epoch": 9.01,
1074
+ "learning_rate": 2.477090909090909e-06,
1075
+ "loss": 0.1116,
1076
+ "step": 4300
1077
+ },
1078
+ {
1079
+ "epoch": 9.02,
1080
+ "learning_rate": 2.4407272727272727e-06,
1081
+ "loss": 0.0894,
1082
+ "step": 4325
1083
+ },
1084
+ {
1085
+ "epoch": 9.02,
1086
+ "learning_rate": 2.4043636363636364e-06,
1087
+ "loss": 0.0813,
1088
+ "step": 4350
1089
+ },
1090
+ {
1091
+ "epoch": 9.03,
1092
+ "learning_rate": 2.3679999999999996e-06,
1093
+ "loss": 0.0782,
1094
+ "step": 4375
1095
+ },
1096
+ {
1097
+ "epoch": 9.03,
1098
+ "learning_rate": 2.3316363636363637e-06,
1099
+ "loss": 0.0706,
1100
+ "step": 4400
1101
+ },
1102
+ {
1103
+ "epoch": 9.03,
1104
+ "learning_rate": 2.295272727272727e-06,
1105
+ "loss": 0.0674,
1106
+ "step": 4425
1107
+ },
1108
+ {
1109
+ "epoch": 9.04,
1110
+ "learning_rate": 2.258909090909091e-06,
1111
+ "loss": 0.0769,
1112
+ "step": 4450
1113
+ },
1114
+ {
1115
+ "epoch": 9.04,
1116
+ "learning_rate": 2.2225454545454543e-06,
1117
+ "loss": 0.0688,
1118
+ "step": 4475
1119
+ },
1120
+ {
1121
+ "epoch": 9.05,
1122
+ "learning_rate": 2.186181818181818e-06,
1123
+ "loss": 0.0904,
1124
+ "step": 4500
1125
+ },
1126
+ {
1127
+ "epoch": 9.05,
1128
+ "learning_rate": 2.1498181818181817e-06,
1129
+ "loss": 0.0896,
1130
+ "step": 4525
1131
+ },
1132
+ {
1133
+ "epoch": 9.05,
1134
+ "learning_rate": 2.1134545454545453e-06,
1135
+ "loss": 0.0699,
1136
+ "step": 4550
1137
+ },
1138
+ {
1139
+ "epoch": 9.06,
1140
+ "learning_rate": 2.0770909090909086e-06,
1141
+ "loss": 0.071,
1142
+ "step": 4575
1143
+ },
1144
+ {
1145
+ "epoch": 9.06,
1146
+ "learning_rate": 2.0407272727272727e-06,
1147
+ "loss": 0.0629,
1148
+ "step": 4600
1149
+ },
1150
+ {
1151
+ "epoch": 9.07,
1152
+ "learning_rate": 2.0043636363636364e-06,
1153
+ "loss": 0.0588,
1154
+ "step": 4625
1155
+ },
1156
+ {
1157
+ "epoch": 9.07,
1158
+ "learning_rate": 1.968e-06,
1159
+ "loss": 0.0579,
1160
+ "step": 4650
1161
+ },
1162
+ {
1163
+ "epoch": 9.08,
1164
+ "learning_rate": 1.9316363636363633e-06,
1165
+ "loss": 0.0671,
1166
+ "step": 4675
1167
+ },
1168
+ {
1169
+ "epoch": 10.0,
1170
+ "learning_rate": 1.8952727272727272e-06,
1171
+ "loss": 0.0735,
1172
+ "step": 4700
1173
+ },
1174
+ {
1175
+ "epoch": 10.01,
1176
+ "learning_rate": 1.8589090909090909e-06,
1177
+ "loss": 0.0715,
1178
+ "step": 4725
1179
+ },
1180
+ {
1181
+ "epoch": 10.01,
1182
+ "learning_rate": 1.8225454545454543e-06,
1183
+ "loss": 0.0705,
1184
+ "step": 4750
1185
+ },
1186
+ {
1187
+ "epoch": 10.01,
1188
+ "learning_rate": 1.786181818181818e-06,
1189
+ "loss": 0.0586,
1190
+ "step": 4775
1191
+ },
1192
+ {
1193
+ "epoch": 10.02,
1194
+ "learning_rate": 1.7498181818181817e-06,
1195
+ "loss": 0.0556,
1196
+ "step": 4800
1197
+ },
1198
+ {
1199
+ "epoch": 10.02,
1200
+ "learning_rate": 1.7134545454545456e-06,
1201
+ "loss": 0.0543,
1202
+ "step": 4825
1203
+ },
1204
+ {
1205
+ "epoch": 10.03,
1206
+ "learning_rate": 1.677090909090909e-06,
1207
+ "loss": 0.053,
1208
+ "step": 4850
1209
+ },
1210
+ {
1211
+ "epoch": 10.03,
1212
+ "learning_rate": 1.6407272727272727e-06,
1213
+ "loss": 0.0578,
1214
+ "step": 4875
1215
+ },
1216
+ {
1217
+ "epoch": 10.04,
1218
+ "learning_rate": 1.6043636363636364e-06,
1219
+ "loss": 0.0944,
1220
+ "step": 4900
1221
+ },
1222
+ {
1223
+ "epoch": 10.04,
1224
+ "learning_rate": 1.568e-06,
1225
+ "loss": 0.0829,
1226
+ "step": 4925
1227
+ },
1228
+ {
1229
+ "epoch": 10.04,
1230
+ "learning_rate": 1.5316363636363635e-06,
1231
+ "loss": 0.0856,
1232
+ "step": 4950
1233
+ },
1234
+ {
1235
+ "epoch": 10.05,
1236
+ "learning_rate": 1.4952727272727272e-06,
1237
+ "loss": 0.0906,
1238
+ "step": 4975
1239
+ },
1240
+ {
1241
+ "epoch": 10.05,
1242
+ "learning_rate": 1.4589090909090909e-06,
1243
+ "loss": 0.096,
1244
+ "step": 5000
1245
+ },
1246
+ {
1247
+ "epoch": 10.05,
1248
+ "eval_loss": 0.27268335223197937,
1249
+ "eval_runtime": 5022.2596,
1250
+ "eval_samples_per_second": 2.987,
1251
+ "eval_steps_per_second": 0.093,
1252
+ "eval_wer": 12.612959812885391,
1253
+ "step": 5000
1254
+ },
1255
+ {
1256
+ "epoch": 10.06,
1257
+ "learning_rate": 1.4225454545454546e-06,
1258
+ "loss": 0.1049,
1259
+ "step": 5025
1260
+ },
1261
+ {
1262
+ "epoch": 10.06,
1263
+ "learning_rate": 1.386181818181818e-06,
1264
+ "loss": 0.09,
1265
+ "step": 5050
1266
+ },
1267
+ {
1268
+ "epoch": 10.06,
1269
+ "learning_rate": 1.3498181818181817e-06,
1270
+ "loss": 0.0809,
1271
+ "step": 5075
1272
+ },
1273
+ {
1274
+ "epoch": 10.07,
1275
+ "learning_rate": 1.3134545454545454e-06,
1276
+ "loss": 0.1007,
1277
+ "step": 5100
1278
+ },
1279
+ {
1280
+ "epoch": 10.07,
1281
+ "learning_rate": 1.277090909090909e-06,
1282
+ "loss": 0.0755,
1283
+ "step": 5125
1284
+ },
1285
+ {
1286
+ "epoch": 10.08,
1287
+ "learning_rate": 1.2407272727272725e-06,
1288
+ "loss": 0.0448,
1289
+ "step": 5150
1290
+ },
1291
+ {
1292
+ "epoch": 11.0,
1293
+ "learning_rate": 1.2043636363636362e-06,
1294
+ "loss": 0.0836,
1295
+ "step": 5175
1296
+ },
1297
+ {
1298
+ "epoch": 11.01,
1299
+ "learning_rate": 1.1679999999999999e-06,
1300
+ "loss": 0.1083,
1301
+ "step": 5200
1302
+ },
1303
+ {
1304
+ "epoch": 11.01,
1305
+ "learning_rate": 1.1316363636363635e-06,
1306
+ "loss": 0.1112,
1307
+ "step": 5225
1308
+ },
1309
+ {
1310
+ "epoch": 11.02,
1311
+ "learning_rate": 1.0952727272727272e-06,
1312
+ "loss": 0.1122,
1313
+ "step": 5250
1314
+ },
1315
+ {
1316
+ "epoch": 11.02,
1317
+ "learning_rate": 1.0589090909090907e-06,
1318
+ "loss": 0.0921,
1319
+ "step": 5275
1320
+ },
1321
+ {
1322
+ "epoch": 11.02,
1323
+ "learning_rate": 1.0225454545454546e-06,
1324
+ "loss": 0.0977,
1325
+ "step": 5300
1326
+ },
1327
+ {
1328
+ "epoch": 11.03,
1329
+ "learning_rate": 9.86181818181818e-07,
1330
+ "loss": 0.1235,
1331
+ "step": 5325
1332
+ },
1333
+ {
1334
+ "epoch": 11.03,
1335
+ "learning_rate": 9.498181818181817e-07,
1336
+ "loss": 0.1045,
1337
+ "step": 5350
1338
+ },
1339
+ {
1340
+ "epoch": 11.04,
1341
+ "learning_rate": 9.134545454545454e-07,
1342
+ "loss": 0.0944,
1343
+ "step": 5375
1344
+ },
1345
+ {
1346
+ "epoch": 11.04,
1347
+ "learning_rate": 8.770909090909091e-07,
1348
+ "loss": 0.0946,
1349
+ "step": 5400
1350
+ },
1351
+ {
1352
+ "epoch": 11.04,
1353
+ "learning_rate": 8.407272727272727e-07,
1354
+ "loss": 0.1045,
1355
+ "step": 5425
1356
+ },
1357
+ {
1358
+ "epoch": 11.05,
1359
+ "learning_rate": 8.043636363636363e-07,
1360
+ "loss": 0.1078,
1361
+ "step": 5450
1362
+ },
1363
+ {
1364
+ "epoch": 11.05,
1365
+ "learning_rate": 7.68e-07,
1366
+ "loss": 0.1053,
1367
+ "step": 5475
1368
+ },
1369
+ {
1370
+ "epoch": 11.06,
1371
+ "learning_rate": 7.316363636363636e-07,
1372
+ "loss": 0.1013,
1373
+ "step": 5500
1374
+ },
1375
+ {
1376
+ "epoch": 11.06,
1377
+ "learning_rate": 6.952727272727272e-07,
1378
+ "loss": 0.098,
1379
+ "step": 5525
1380
+ },
1381
+ {
1382
+ "epoch": 11.07,
1383
+ "learning_rate": 6.589090909090908e-07,
1384
+ "loss": 0.0974,
1385
+ "step": 5550
1386
+ },
1387
+ {
1388
+ "epoch": 11.07,
1389
+ "learning_rate": 6.225454545454545e-07,
1390
+ "loss": 0.1084,
1391
+ "step": 5575
1392
+ },
1393
+ {
1394
+ "epoch": 11.07,
1395
+ "learning_rate": 5.861818181818181e-07,
1396
+ "loss": 0.1079,
1397
+ "step": 5600
1398
+ },
1399
+ {
1400
+ "epoch": 11.08,
1401
+ "learning_rate": 5.498181818181818e-07,
1402
+ "loss": 0.0986,
1403
+ "step": 5625
1404
+ },
1405
+ {
1406
+ "epoch": 12.0,
1407
+ "learning_rate": 5.134545454545455e-07,
1408
+ "loss": 0.0563,
1409
+ "step": 5650
1410
+ },
1411
+ {
1412
+ "epoch": 12.01,
1413
+ "learning_rate": 4.770909090909091e-07,
1414
+ "loss": 0.0572,
1415
+ "step": 5675
1416
+ },
1417
+ {
1418
+ "epoch": 12.01,
1419
+ "learning_rate": 4.407272727272727e-07,
1420
+ "loss": 0.0561,
1421
+ "step": 5700
1422
+ },
1423
+ {
1424
+ "epoch": 12.02,
1425
+ "learning_rate": 4.0436363636363634e-07,
1426
+ "loss": 0.0505,
1427
+ "step": 5725
1428
+ },
1429
+ {
1430
+ "epoch": 12.02,
1431
+ "learning_rate": 3.6799999999999996e-07,
1432
+ "loss": 0.0479,
1433
+ "step": 5750
1434
+ },
1435
+ {
1436
+ "epoch": 12.02,
1437
+ "learning_rate": 3.316363636363636e-07,
1438
+ "loss": 0.0522,
1439
+ "step": 5775
1440
+ },
1441
+ {
1442
+ "epoch": 12.03,
1443
+ "learning_rate": 2.9527272727272726e-07,
1444
+ "loss": 0.0617,
1445
+ "step": 5800
1446
+ },
1447
+ {
1448
+ "epoch": 12.03,
1449
+ "learning_rate": 2.5890909090909094e-07,
1450
+ "loss": 0.0468,
1451
+ "step": 5825
1452
+ },
1453
+ {
1454
+ "epoch": 12.04,
1455
+ "learning_rate": 2.2254545454545453e-07,
1456
+ "loss": 0.0468,
1457
+ "step": 5850
1458
+ },
1459
+ {
1460
+ "epoch": 12.04,
1461
+ "learning_rate": 1.8618181818181816e-07,
1462
+ "loss": 0.0472,
1463
+ "step": 5875
1464
+ },
1465
+ {
1466
+ "epoch": 12.05,
1467
+ "learning_rate": 1.498181818181818e-07,
1468
+ "loss": 0.0495,
1469
+ "step": 5900
1470
+ },
1471
+ {
1472
+ "epoch": 12.05,
1473
+ "learning_rate": 1.1345454545454546e-07,
1474
+ "loss": 0.0546,
1475
+ "step": 5925
1476
+ },
1477
+ {
1478
+ "epoch": 12.05,
1479
+ "learning_rate": 7.709090909090908e-08,
1480
+ "loss": 0.0579,
1481
+ "step": 5950
1482
+ },
1483
+ {
1484
+ "epoch": 12.06,
1485
+ "learning_rate": 4.072727272727273e-08,
1486
+ "loss": 0.0623,
1487
+ "step": 5975
1488
+ },
1489
+ {
1490
+ "epoch": 12.06,
1491
+ "learning_rate": 4.363636363636364e-09,
1492
+ "loss": 0.0604,
1493
+ "step": 6000
1494
+ },
1495
+ {
1496
+ "epoch": 12.06,
1497
+ "eval_loss": 0.26977524161338806,
1498
+ "eval_runtime": 5061.9846,
1499
+ "eval_samples_per_second": 2.964,
1500
+ "eval_steps_per_second": 0.093,
1501
+ "eval_wer": 12.502657877950243,
1502
+ "step": 6000
1503
+ },
1504
+ {
1505
+ "epoch": 12.06,
1506
+ "step": 6000,
1507
+ "total_flos": 1.1074060717719552e+20,
1508
+ "train_loss": 0.1507516227165858,
1509
+ "train_runtime": 84190.0984,
1510
+ "train_samples_per_second": 4.561,
1511
+ "train_steps_per_second": 0.071
1512
+ }
1513
+ ],
1514
+ "max_steps": 6000,
1515
+ "num_train_epochs": 9223372036854775807,
1516
+ "total_flos": 1.1074060717719552e+20,
1517
+ "trial_name": null,
1518
+ "trial_params": null
1519
+ }