PereLluis13 commited on
Commit
98963dd
1 Parent(s): b9f77f2
.ipynb_checkpoints/README-checkpoint.md DELETED
@@ -1,158 +0,0 @@
1
- ---
2
- language:
3
- - ca
4
- license: apache-2.0
5
- tags:
6
- - automatic-speech-recognition
7
- - mozilla-foundation/common_voice_8_0
8
- - collectivat/tv3_parla
9
- - projecte-aina/parlament_parla
10
- - generated_from_trainer
11
- - robust-speech-event
12
- datasets:
13
- - mozilla-foundation/common_voice_8_0
14
- - collectivat/tv3_parla
15
- - projecte-aina/parlament_parla
16
- model-index:
17
- - name: wav2vec2-xls-r-300m-ca
18
- results:
19
- - task:
20
- name: Speech Recognition
21
- type: automatic-speech-recognition
22
- dataset:
23
- name: mozilla-foundation/common_voice_8_0 ca
24
- type: mozilla-foundation/common_voice_8_0
25
- args: ca
26
- metrics:
27
- - name: Test WER
28
- type: wer
29
- value: 0.15636874077301
30
- - name: Test CER
31
- type: cer
32
- value: 0.04086725403909639
33
- - task:
34
- name: Speech Recognition
35
- type: automatic-speech-recognition
36
- dataset:
37
- name: projecte-aina/parlament_parla ca
38
- type: projecte-aina/parlament_parla
39
- args: clean
40
- metrics:
41
- - name: Test WER
42
- type: wer
43
- value: 0.09940385143350199
44
- - name: Test CER
45
- type: cer
46
- value: 0.026906712890009454
47
- - task:
48
- name: Speech Recognition
49
- type: automatic-speech-recognition
50
- dataset:
51
- name: collectivat/tv3_parla ca
52
- type: collectivat/tv3_parla
53
- args: ca
54
- metrics:
55
- - name: Test WER
56
- type: wer
57
- value: 0.27349193517342263
58
- - name: Test CER
59
- type: cer
60
- value: 0.11571091827304163
61
- ---
62
-
63
- <!-- This model card has been generated automatically according to the information the Trainer had access to. You
64
- should probably proofread and complete it, then remove this comment. -->
65
-
66
- # wav2vec2-xls-r-300m-ca
67
-
68
- This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - CA dataset.
69
- It achieves the following results on the evaluation set:
70
- - Loss: 0.2549
71
- - Wer: 0.1573
72
-
73
- ## Model description
74
-
75
- More information needed
76
-
77
- ## Intended uses & limitations
78
-
79
- More information needed
80
-
81
- ## Training and evaluation data
82
-
83
- More information needed
84
-
85
- ## Training procedure
86
-
87
- ### Training hyperparameters
88
-
89
- The following hyperparameters were used during training:
90
- - learning_rate: 7.5e-05
91
- - train_batch_size: 32
92
- - eval_batch_size: 32
93
- - seed: 42
94
- - gradient_accumulation_steps: 4
95
- - total_train_batch_size: 128
96
- - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
97
- - lr_scheduler_type: linear
98
- - lr_scheduler_warmup_steps: 2000
99
- - num_epochs: 12.0
100
- - mixed_precision_training: Native AMP
101
-
102
- ### Training results
103
-
104
- | Training Loss | Epoch | Step | Validation Loss | Wer |
105
- |:-------------:|:-----:|:-----:|:---------------:|:------:|
106
- | 6.2099 | 0.09 | 500 | 3.4125 | 1.0 |
107
- | 2.9961 | 0.18 | 1000 | 2.9224 | 1.0 |
108
- | 2.2147 | 0.26 | 1500 | 0.6521 | 0.5568 |
109
- | 1.3017 | 0.35 | 2000 | 0.3153 | 0.2761 |
110
- | 1.1196 | 0.44 | 2500 | 0.2444 | 0.2367 |
111
- | 1.0712 | 0.53 | 3000 | 0.2324 | 0.2132 |
112
- | 1.052 | 0.62 | 3500 | 0.2173 | 0.2032 |
113
- | 1.2813 | 2.13 | 4000 | 0.3326 | 0.2099 |
114
- | 1.2365 | 2.4 | 4500 | 0.3224 | 0.2003 |
115
- | 1.2193 | 2.66 | 5000 | 0.3198 | 0.1957 |
116
- | 1.2072 | 2.93 | 5500 | 0.3063 | 0.1933 |
117
- | 1.213 | 3.2 | 6000 | 0.3051 | 0.1980 |
118
- | 1.2074 | 3.46 | 6500 | 0.3012 | 0.1879 |
119
- | 1.1918 | 3.73 | 7000 | 0.2947 | 0.1829 |
120
- | 1.1893 | 4.0 | 7500 | 0.2895 | 0.1807 |
121
- | 1.1751 | 4.26 | 8000 | 0.2878 | 0.1776 |
122
- | 1.1628 | 4.53 | 8500 | 0.2835 | 0.1731 |
123
- | 1.1577 | 4.79 | 9000 | 0.2816 | 0.1761 |
124
- | 1.1448 | 5.06 | 9500 | 0.2757 | 0.1740 |
125
- | 1.1407 | 5.33 | 10000 | 0.2768 | 0.1798 |
126
- | 1.1401 | 5.59 | 10500 | 0.2780 | 0.1816 |
127
- | 1.1333 | 5.86 | 11000 | 0.2748 | 0.1750 |
128
- | 1.1571 | 6.13 | 11500 | 0.2808 | 0.1708 |
129
- | 1.1505 | 6.39 | 12000 | 0.2726 | 0.1692 |
130
- | 1.1519 | 6.66 | 12500 | 0.2749 | 0.1654 |
131
- | 1.136 | 6.93 | 13000 | 0.2765 | 0.1643 |
132
- | 1.1326 | 7.19 | 13500 | 0.2706 | 0.1668 |
133
- | 1.1342 | 7.46 | 14000 | 0.2665 | 0.1638 |
134
- | 1.1286 | 7.72 | 14500 | 0.2669 | 0.1636 |
135
- | 1.1243 | 7.99 | 15000 | 0.2619 | 0.1623 |
136
- | 1.1173 | 8.26 | 15500 | 0.2652 | 0.1604 |
137
- | 1.1129 | 8.52 | 16000 | 0.2610 | 0.1598 |
138
- | 1.1091 | 8.79 | 16500 | 0.2608 | 0.1584 |
139
- | 1.1053 | 9.06 | 17000 | 0.2633 | 0.1664 |
140
- | 1.1004 | 9.32 | 17500 | 0.2594 | 0.1662 |
141
- | 1.0995 | 9.59 | 18000 | 0.2623 | 0.1569 |
142
- | 1.0964 | 9.86 | 18500 | 0.2624 | 0.1597 |
143
- | 1.09 | 10.12 | 19000 | 0.2577 | 0.1578 |
144
- | 1.089 | 10.39 | 19500 | 0.2574 | 0.1531 |
145
- | 1.0864 | 10.66 | 20000 | 0.2556 | 0.1546 |
146
- | 1.0806 | 10.92 | 20500 | 0.2548 | 0.1583 |
147
- | 1.0842 | 11.19 | 21000 | 0.2550 | 0.1542 |
148
- | 1.0805 | 11.45 | 21500 | 0.2561 | 0.1524 |
149
- | 1.0722 | 11.72 | 22000 | 0.2540 | 0.1566 |
150
- | 1.0763 | 11.99 | 22500 | 0.2549 | 0.1572 |
151
-
152
-
153
- ### Framework versions
154
-
155
- - Transformers 4.16.0.dev0
156
- - Pytorch 1.10.1+cu102
157
- - Datasets 1.18.1
158
- - Tokenizers 0.11.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.ipynb_checkpoints/all_results-checkpoint.json DELETED
@@ -1,14 +0,0 @@
1
- {
2
- "epoch": 12.0,
3
- "eval_loss": 0.25491979718208313,
4
- "eval_runtime": 392.0567,
5
- "eval_samples": 4297,
6
- "eval_samples_per_second": 10.96,
7
- "eval_steps_per_second": 0.344,
8
- "eval_wer": 0.15725760362438562,
9
- "train_loss": 0.5676147035501541,
10
- "train_runtime": 172546.67,
11
- "train_samples": 240334,
12
- "train_samples_per_second": 16.714,
13
- "train_steps_per_second": 0.131
14
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.ipynb_checkpoints/eval_results-checkpoint.json DELETED
@@ -1,9 +0,0 @@
1
- {
2
- "epoch": 12.0,
3
- "eval_loss": 0.25491979718208313,
4
- "eval_runtime": 392.0567,
5
- "eval_samples": 4297,
6
- "eval_samples_per_second": 10.96,
7
- "eval_steps_per_second": 0.344,
8
- "eval_wer": 0.15725760362438562
9
- }
 
 
 
 
 
 
 
 
 
 
.ipynb_checkpoints/run-checkpoint.sh DELETED
@@ -1,39 +0,0 @@
1
- python run_speech_recognition_ctc.py \
2
- --dataset_name "mozilla-foundation/common_voice_8_0" "collectivat/tv3_parla" "projecte-aina/parlament_parla" \
3
- --dataset_config_name "ca" "ca" "clean" \
4
- --model_name_or_path="facebook/wav2vec2-xls-r-300m" \
5
- --train_split_name "train+validation" "train" "train+validation" \
6
- --eval_split_name "test" "test" "test" \
7
- --audio_column_name "audio" "audio" "audio" \
8
- --output_dir="wav2vec2-xls-r-300m-ca" \
9
- --overwrite_output_dir \
10
- --num_train_epochs="10" \
11
- --per_device_train_batch_size="32" \
12
- --per_device_eval_batch_size="32" \
13
- --gradient_accumulation_steps="4" \
14
- --learning_rate="7.5e-5" \
15
- --warmup_steps="2000" \
16
- --length_column_name="input_length" \
17
- --evaluation_strategy="steps" \
18
- --text_column_name "sentence" "text" "sentence" \
19
- --chars_to_ignore [ , ? . ! \; \: \" “ % ” � — … – ] \
20
- --save_steps="500" \
21
- --eval_steps="500" \
22
- --logging_steps="500" \
23
- --layerdrop="0.0" \
24
- --activation_dropout="0.1" \
25
- --save_total_limit="3" \
26
- --freeze_feature_encoder \
27
- --feat_proj_dropout="0.0" \
28
- --mask_time_prob="0.75" \
29
- --preprocessing_num_workers="12" \
30
- --mask_time_length="10" \
31
- --mask_feature_prob="0.25" \
32
- --mask_feature_length="64" \
33
- --gradient_checkpointing \
34
- --use_auth_token \
35
- --fp16 \
36
- --group_by_length \
37
- --do_train --do_eval \
38
- --push_to_hub
39
- #&> train.log
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.ipynb_checkpoints/train_results-checkpoint.json DELETED
@@ -1,8 +0,0 @@
1
- {
2
- "epoch": 12.0,
3
- "train_loss": 0.5676147035501541,
4
- "train_runtime": 172546.67,
5
- "train_samples": 240334,
6
- "train_samples_per_second": 16.714,
7
- "train_steps_per_second": 0.131
8
- }
 
 
 
 
 
 
 
 
 
.ipynb_checkpoints/trainer_state-checkpoint.json DELETED
@@ -1,700 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 11.999600585807482,
5
- "global_step": 22524,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.09,
12
- "learning_rate": 1.8712499999999997e-05,
13
- "loss": 6.2099,
14
- "step": 500
15
- },
16
- {
17
- "epoch": 0.09,
18
- "eval_loss": 3.41253662109375,
19
- "eval_runtime": 1358.074,
20
- "eval_samples_per_second": 14.868,
21
- "eval_steps_per_second": 0.465,
22
- "eval_wer": 1.0,
23
- "step": 500
24
- },
25
- {
26
- "epoch": 0.18,
27
- "learning_rate": 3.7462499999999996e-05,
28
- "loss": 2.9961,
29
- "step": 1000
30
- },
31
- {
32
- "epoch": 0.18,
33
- "eval_loss": 2.9223873615264893,
34
- "eval_runtime": 1361.8836,
35
- "eval_samples_per_second": 14.827,
36
- "eval_steps_per_second": 0.463,
37
- "eval_wer": 1.0,
38
- "step": 1000
39
- },
40
- {
41
- "epoch": 0.26,
42
- "learning_rate": 5.62125e-05,
43
- "loss": 2.2147,
44
- "step": 1500
45
- },
46
- {
47
- "epoch": 0.26,
48
- "eval_loss": 0.6520791053771973,
49
- "eval_runtime": 1355.9385,
50
- "eval_samples_per_second": 14.892,
51
- "eval_steps_per_second": 0.465,
52
- "eval_wer": 0.5567824255052013,
53
- "step": 1500
54
- },
55
- {
56
- "epoch": 0.35,
57
- "learning_rate": 7.49625e-05,
58
- "loss": 1.3017,
59
- "step": 2000
60
- },
61
- {
62
- "epoch": 0.35,
63
- "eval_loss": 0.31525641679763794,
64
- "eval_runtime": 1371.0684,
65
- "eval_samples_per_second": 14.727,
66
- "eval_steps_per_second": 0.46,
67
- "eval_wer": 0.27607269814249236,
68
- "step": 2000
69
- },
70
- {
71
- "epoch": 0.44,
72
- "learning_rate": 7.431581352833637e-05,
73
- "loss": 1.1196,
74
- "step": 2500
75
- },
76
- {
77
- "epoch": 0.44,
78
- "eval_loss": 0.24444133043289185,
79
- "eval_runtime": 1363.4869,
80
- "eval_samples_per_second": 14.809,
81
- "eval_steps_per_second": 0.463,
82
- "eval_wer": 0.23673497683411066,
83
- "step": 2500
84
- },
85
- {
86
- "epoch": 0.53,
87
- "learning_rate": 7.363162705667275e-05,
88
- "loss": 1.0712,
89
- "step": 3000
90
- },
91
- {
92
- "epoch": 0.53,
93
- "eval_loss": 0.23236490786075592,
94
- "eval_runtime": 1368.4487,
95
- "eval_samples_per_second": 14.755,
96
- "eval_steps_per_second": 0.461,
97
- "eval_wer": 0.21321861436015543,
98
- "step": 3000
99
- },
100
- {
101
- "epoch": 0.62,
102
- "learning_rate": 7.294606946983546e-05,
103
- "loss": 1.052,
104
- "step": 3500
105
- },
106
- {
107
- "epoch": 0.62,
108
- "eval_loss": 0.21734564006328583,
109
- "eval_runtime": 1358.6503,
110
- "eval_samples_per_second": 14.862,
111
- "eval_steps_per_second": 0.464,
112
- "eval_wer": 0.20317428528513493,
113
- "step": 3500
114
- },
115
- {
116
- "epoch": 2.13,
117
- "learning_rate": 3.375103277334068e-05,
118
- "loss": 1.2813,
119
- "step": 4000
120
- },
121
- {
122
- "epoch": 2.13,
123
- "eval_loss": 0.332599401473999,
124
- "eval_runtime": 393.989,
125
- "eval_samples_per_second": 10.906,
126
- "eval_steps_per_second": 0.343,
127
- "eval_wer": 0.20989847979467358,
128
- "step": 4000
129
- },
130
- {
131
- "epoch": 2.4,
132
- "learning_rate": 2.3423299366565682e-05,
133
- "loss": 1.2365,
134
- "step": 4500
135
- },
136
- {
137
- "epoch": 2.4,
138
- "eval_loss": 0.3223731815814972,
139
- "eval_runtime": 391.7507,
140
- "eval_samples_per_second": 10.969,
141
- "eval_steps_per_second": 0.345,
142
- "eval_wer": 0.2003283560376986,
143
- "step": 4500
144
- },
145
- {
146
- "epoch": 2.66,
147
- "learning_rate": 1.309556595979069e-05,
148
- "loss": 1.2193,
149
- "step": 5000
150
- },
151
- {
152
- "epoch": 2.66,
153
- "eval_loss": 0.31981074810028076,
154
- "eval_runtime": 402.4439,
155
- "eval_samples_per_second": 10.677,
156
- "eval_steps_per_second": 0.335,
157
- "eval_wer": 0.19565239980464894,
158
- "step": 5000
159
- },
160
- {
161
- "epoch": 2.93,
162
- "learning_rate": 2.767832553015698e-06,
163
- "loss": 1.2072,
164
- "step": 5500
165
- },
166
- {
167
- "epoch": 2.93,
168
- "eval_loss": 0.3063456118106842,
169
- "eval_runtime": 396.6525,
170
- "eval_samples_per_second": 10.833,
171
- "eval_steps_per_second": 0.34,
172
- "eval_wer": 0.19327285763271923,
173
- "step": 5500
174
- },
175
- {
176
- "epoch": 3.2,
177
- "learning_rate": 4.263388037141005e-05,
178
- "loss": 1.213,
179
- "step": 6000
180
- },
181
- {
182
- "epoch": 3.2,
183
- "eval_loss": 0.3051116168498993,
184
- "eval_runtime": 371.5477,
185
- "eval_samples_per_second": 11.565,
186
- "eval_steps_per_second": 0.363,
187
- "eval_wer": 0.19797998690732255,
188
- "step": 6000
189
- },
190
- {
191
- "epoch": 3.46,
192
- "learning_rate": 3.8585078816670256e-05,
193
- "loss": 1.2074,
194
- "step": 6500
195
- },
196
- {
197
- "epoch": 3.46,
198
- "eval_loss": 0.3011966645717621,
199
- "eval_runtime": 372.0587,
200
- "eval_samples_per_second": 11.549,
201
- "eval_steps_per_second": 0.363,
202
- "eval_wer": 0.18792148549934018,
203
- "step": 6500
204
- },
205
- {
206
- "epoch": 3.73,
207
- "learning_rate": 3.4536277261930465e-05,
208
- "loss": 1.1918,
209
- "step": 7000
210
- },
211
- {
212
- "epoch": 3.73,
213
- "eval_loss": 0.29471954703330994,
214
- "eval_runtime": 372.8767,
215
- "eval_samples_per_second": 11.524,
216
- "eval_steps_per_second": 0.362,
217
- "eval_wer": 0.18287145276764655,
218
- "step": 7000
219
- },
220
- {
221
- "epoch": 4.0,
222
- "learning_rate": 3.0487475707190668e-05,
223
- "loss": 1.1893,
224
- "step": 7500
225
- },
226
- {
227
- "epoch": 4.0,
228
- "eval_loss": 0.28953900933265686,
229
- "eval_runtime": 372.4558,
230
- "eval_samples_per_second": 11.537,
231
- "eval_steps_per_second": 0.362,
232
- "eval_wer": 0.1807205129004437,
233
- "step": 7500
234
- },
235
- {
236
- "epoch": 4.26,
237
- "learning_rate": 2.643867415245087e-05,
238
- "loss": 1.1751,
239
- "step": 8000
240
- },
241
- {
242
- "epoch": 4.26,
243
- "eval_loss": 0.2877567410469055,
244
- "eval_runtime": 371.9244,
245
- "eval_samples_per_second": 11.553,
246
- "eval_steps_per_second": 0.363,
247
- "eval_wer": 0.17759281773122604,
248
- "step": 8000
249
- },
250
- {
251
- "epoch": 4.53,
252
- "learning_rate": 2.2389872597711073e-05,
253
- "loss": 1.1628,
254
- "step": 8500
255
- },
256
- {
257
- "epoch": 4.53,
258
- "eval_loss": 0.28352275490760803,
259
- "eval_runtime": 373.4388,
260
- "eval_samples_per_second": 11.507,
261
- "eval_steps_per_second": 0.362,
262
- "eval_wer": 0.1730727267059447,
263
- "step": 8500
264
- },
265
- {
266
- "epoch": 4.79,
267
- "learning_rate": 1.834107104297128e-05,
268
- "loss": 1.1577,
269
- "step": 9000
270
- },
271
- {
272
- "epoch": 4.79,
273
- "eval_loss": 0.28156954050064087,
274
- "eval_runtime": 371.5997,
275
- "eval_samples_per_second": 11.564,
276
- "eval_steps_per_second": 0.363,
277
- "eval_wer": 0.17605494768124527,
278
- "step": 9000
279
- },
280
- {
281
- "epoch": 5.06,
282
- "learning_rate": 1.4300367091340963e-05,
283
- "loss": 1.1448,
284
- "step": 9500
285
- },
286
- {
287
- "epoch": 5.06,
288
- "eval_loss": 0.2757455110549927,
289
- "eval_runtime": 373.7148,
290
- "eval_samples_per_second": 11.498,
291
- "eval_steps_per_second": 0.361,
292
- "eval_wer": 0.17397674491100096,
293
- "step": 9500
294
- },
295
- {
296
- "epoch": 5.33,
297
- "learning_rate": 1.0251565536601165e-05,
298
- "loss": 1.1407,
299
- "step": 10000
300
- },
301
- {
302
- "epoch": 5.33,
303
- "eval_loss": 0.2768358588218689,
304
- "eval_runtime": 373.2273,
305
- "eval_samples_per_second": 11.513,
306
- "eval_steps_per_second": 0.362,
307
- "eval_wer": 0.17976453962613131,
308
- "step": 10000
309
- },
310
- {
311
- "epoch": 5.59,
312
- "learning_rate": 6.202763981861369e-06,
313
- "loss": 1.1401,
314
- "step": 10500
315
- },
316
- {
317
- "epoch": 5.59,
318
- "eval_loss": 0.2780420780181885,
319
- "eval_runtime": 375.3753,
320
- "eval_samples_per_second": 11.447,
321
- "eval_steps_per_second": 0.36,
322
- "eval_wer": 0.18160374907779753,
323
- "step": 10500
324
- },
325
- {
326
- "epoch": 5.86,
327
- "learning_rate": 2.153962427121572e-06,
328
- "loss": 1.1333,
329
- "step": 11000
330
- },
331
- {
332
- "epoch": 5.86,
333
- "eval_loss": 0.2748132646083832,
334
- "eval_runtime": 371.6352,
335
- "eval_samples_per_second": 11.562,
336
- "eval_steps_per_second": 0.363,
337
- "eval_wer": 0.17495350021301578,
338
- "step": 11000
339
- },
340
- {
341
- "epoch": 6.13,
342
- "learning_rate": 4.029916195673358e-05,
343
- "loss": 1.1571,
344
- "step": 11500
345
- },
346
- {
347
- "epoch": 6.13,
348
- "eval_loss": 0.2807982861995697,
349
- "eval_runtime": 393.9701,
350
- "eval_samples_per_second": 10.907,
351
- "eval_steps_per_second": 0.343,
352
- "eval_wer": 0.17083865872793208,
353
- "step": 11500
354
- },
355
- {
356
- "epoch": 6.39,
357
- "learning_rate": 3.8472032742155524e-05,
358
- "loss": 1.1505,
359
- "step": 12000
360
- },
361
- {
362
- "epoch": 6.39,
363
- "eval_loss": 0.27258625626564026,
364
- "eval_runtime": 389.5937,
365
- "eval_samples_per_second": 11.029,
366
- "eval_steps_per_second": 0.347,
367
- "eval_wer": 0.16920726955329032,
368
- "step": 12000
369
- },
370
- {
371
- "epoch": 6.66,
372
- "learning_rate": 3.6648557786006625e-05,
373
- "loss": 1.1519,
374
- "step": 12500
375
- },
376
- {
377
- "epoch": 6.66,
378
- "eval_loss": 0.2748955190181732,
379
- "eval_runtime": 387.583,
380
- "eval_samples_per_second": 11.087,
381
- "eval_steps_per_second": 0.348,
382
- "eval_wer": 0.16539376746989204,
383
- "step": 12500
384
- },
385
- {
386
- "epoch": 6.93,
387
- "learning_rate": 3.482142857142857e-05,
388
- "loss": 1.136,
389
- "step": 13000
390
- },
391
- {
392
- "epoch": 6.93,
393
- "eval_loss": 0.276467502117157,
394
- "eval_runtime": 393.6733,
395
- "eval_samples_per_second": 10.915,
396
- "eval_steps_per_second": 0.343,
397
- "eval_wer": 0.16425075594625768,
398
- "step": 13000
399
- },
400
- {
401
- "epoch": 7.19,
402
- "learning_rate": 3.299429935685051e-05,
403
- "loss": 1.1326,
404
- "step": 13500
405
- },
406
- {
407
- "epoch": 7.19,
408
- "eval_loss": 0.27056241035461426,
409
- "eval_runtime": 391.6082,
410
- "eval_samples_per_second": 10.973,
411
- "eval_steps_per_second": 0.345,
412
- "eval_wer": 0.1668277273813606,
413
- "step": 13500
414
- },
415
- {
416
- "epoch": 7.46,
417
- "learning_rate": 3.116717014227246e-05,
418
- "loss": 1.1342,
419
- "step": 14000
420
- },
421
- {
422
- "epoch": 7.46,
423
- "eval_loss": 0.26654428243637085,
424
- "eval_runtime": 388.567,
425
- "eval_samples_per_second": 11.059,
426
- "eval_steps_per_second": 0.347,
427
- "eval_wer": 0.1638247243783576,
428
- "step": 14000
429
- },
430
- {
431
- "epoch": 7.72,
432
- "learning_rate": 2.9340040927694404e-05,
433
- "loss": 1.1286,
434
- "step": 14500
435
- },
436
- {
437
- "epoch": 7.72,
438
- "eval_loss": 0.2668865919113159,
439
- "eval_runtime": 384.4478,
440
- "eval_samples_per_second": 11.177,
441
- "eval_steps_per_second": 0.351,
442
- "eval_wer": 0.1635857310597795,
443
- "step": 14500
444
- },
445
- {
446
- "epoch": 7.99,
447
- "learning_rate": 2.7516565971545504e-05,
448
- "loss": 1.1243,
449
- "step": 15000
450
- },
451
- {
452
- "epoch": 7.99,
453
- "eval_loss": 0.2619013488292694,
454
- "eval_runtime": 392.1143,
455
- "eval_samples_per_second": 10.959,
456
- "eval_steps_per_second": 0.344,
457
- "eval_wer": 0.16234920041148415,
458
- "step": 15000
459
- },
460
- {
461
- "epoch": 8.26,
462
- "learning_rate": 2.5689436756967453e-05,
463
- "loss": 1.1173,
464
- "step": 15500
465
- },
466
- {
467
- "epoch": 8.26,
468
- "eval_loss": 0.2651631236076355,
469
- "eval_runtime": 388.7223,
470
- "eval_samples_per_second": 11.054,
471
- "eval_steps_per_second": 0.347,
472
- "eval_wer": 0.16037490777975208,
473
- "step": 15500
474
- },
475
- {
476
- "epoch": 8.52,
477
- "learning_rate": 2.3862307542389395e-05,
478
- "loss": 1.1129,
479
- "step": 16000
480
- },
481
- {
482
- "epoch": 8.52,
483
- "eval_loss": 0.26095372438430786,
484
- "eval_runtime": 387.1141,
485
- "eval_samples_per_second": 11.1,
486
- "eval_steps_per_second": 0.349,
487
- "eval_wer": 0.1597514469486788,
488
- "step": 16000
489
- },
490
- {
491
- "epoch": 8.79,
492
- "learning_rate": 2.203517832781134e-05,
493
- "loss": 1.1091,
494
- "step": 16500
495
- },
496
- {
497
- "epoch": 8.79,
498
- "eval_loss": 0.2608127295970917,
499
- "eval_runtime": 386.5271,
500
- "eval_samples_per_second": 11.117,
501
- "eval_steps_per_second": 0.349,
502
- "eval_wer": 0.15837983312031756,
503
- "step": 16500
504
- },
505
- {
506
- "epoch": 9.06,
507
- "learning_rate": 2.0208049113233286e-05,
508
- "loss": 1.1053,
509
- "step": 17000
510
- },
511
- {
512
- "epoch": 9.06,
513
- "eval_loss": 0.2632738947868347,
514
- "eval_runtime": 389.2784,
515
- "eval_samples_per_second": 11.038,
516
- "eval_steps_per_second": 0.347,
517
- "eval_wer": 0.1664016958134605,
518
- "step": 17000
519
- },
520
- {
521
- "epoch": 9.32,
522
- "learning_rate": 1.8384574157084387e-05,
523
- "loss": 1.1004,
524
- "step": 17500
525
- },
526
- {
527
- "epoch": 9.32,
528
- "eval_loss": 0.2594461441040039,
529
- "eval_runtime": 388.9349,
530
- "eval_samples_per_second": 11.048,
531
- "eval_steps_per_second": 0.347,
532
- "eval_wer": 0.16618348452258486,
533
- "step": 17500
534
- },
535
- {
536
- "epoch": 9.59,
537
- "learning_rate": 1.6557444942506332e-05,
538
- "loss": 1.0995,
539
- "step": 18000
540
- },
541
- {
542
- "epoch": 9.59,
543
- "eval_loss": 0.26234543323516846,
544
- "eval_runtime": 383.4387,
545
- "eval_samples_per_second": 11.206,
546
- "eval_steps_per_second": 0.352,
547
- "eval_wer": 0.15694587320884898,
548
- "step": 18000
549
- },
550
- {
551
- "epoch": 9.86,
552
- "learning_rate": 1.4730315727928276e-05,
553
- "loss": 1.0964,
554
- "step": 18500
555
- },
556
- {
557
- "epoch": 9.86,
558
- "eval_loss": 0.26244357228279114,
559
- "eval_runtime": 384.9493,
560
- "eval_samples_per_second": 11.163,
561
- "eval_steps_per_second": 0.351,
562
- "eval_wer": 0.15968910086557145,
563
- "step": 18500
564
- },
565
- {
566
- "epoch": 10.12,
567
- "learning_rate": 1.2903186513350224e-05,
568
- "loss": 1.09,
569
- "step": 19000
570
- },
571
- {
572
- "epoch": 10.12,
573
- "eval_loss": 0.2576591968536377,
574
- "eval_runtime": 391.5397,
575
- "eval_samples_per_second": 10.975,
576
- "eval_steps_per_second": 0.345,
577
- "eval_wer": 0.15782910938620281,
578
- "step": 19000
579
- },
580
- {
581
- "epoch": 10.39,
582
- "learning_rate": 1.108336581563048e-05,
583
- "loss": 1.089,
584
- "step": 19500
585
- },
586
- {
587
- "epoch": 10.39,
588
- "eval_loss": 0.25744903087615967,
589
- "eval_runtime": 384.7348,
590
- "eval_samples_per_second": 11.169,
591
- "eval_steps_per_second": 0.351,
592
- "eval_wer": 0.1531219801115995,
593
- "step": 19500
594
- },
595
- {
596
- "epoch": 10.66,
597
- "learning_rate": 9.256236601052426e-06,
598
- "loss": 1.0864,
599
- "step": 20000
600
- },
601
- {
602
- "epoch": 10.66,
603
- "eval_loss": 0.2556162476539612,
604
- "eval_runtime": 391.8515,
605
- "eval_samples_per_second": 10.966,
606
- "eval_steps_per_second": 0.345,
607
- "eval_wer": 0.15464945914772904,
608
- "step": 20000
609
- },
610
- {
611
- "epoch": 10.92,
612
- "learning_rate": 7.429107386474371e-06,
613
- "loss": 1.0806,
614
- "step": 20500
615
- },
616
- {
617
- "epoch": 10.92,
618
- "eval_loss": 0.2548317313194275,
619
- "eval_runtime": 384.3272,
620
- "eval_samples_per_second": 11.181,
621
- "eval_steps_per_second": 0.351,
622
- "eval_wer": 0.1583174870372102,
623
- "step": 20500
624
- },
625
- {
626
- "epoch": 11.19,
627
- "learning_rate": 5.6019781718963155e-06,
628
- "loss": 1.0842,
629
- "step": 21000
630
- },
631
- {
632
- "epoch": 11.19,
633
- "eval_loss": 0.25501224398612976,
634
- "eval_runtime": 383.8764,
635
- "eval_samples_per_second": 11.194,
636
- "eval_steps_per_second": 0.352,
637
- "eval_wer": 0.1542026455521265,
638
- "step": 21000
639
- },
640
- {
641
- "epoch": 11.45,
642
- "learning_rate": 3.774848957318261e-06,
643
- "loss": 1.0805,
644
- "step": 21500
645
- },
646
- {
647
- "epoch": 11.45,
648
- "eval_loss": 0.25607678294181824,
649
- "eval_runtime": 388.7678,
650
- "eval_samples_per_second": 11.053,
651
- "eval_steps_per_second": 0.347,
652
- "eval_wer": 0.15236343610046033,
653
- "step": 21500
654
- },
655
- {
656
- "epoch": 11.72,
657
- "learning_rate": 1.9477197427402063e-06,
658
- "loss": 1.0722,
659
- "step": 22000
660
- },
661
- {
662
- "epoch": 11.72,
663
- "eval_loss": 0.25402259826660156,
664
- "eval_runtime": 385.5053,
665
- "eval_samples_per_second": 11.146,
666
- "eval_steps_per_second": 0.35,
667
- "eval_wer": 0.1565614056963538,
668
- "step": 22000
669
- },
670
- {
671
- "epoch": 11.99,
672
- "learning_rate": 1.2059052816215163e-07,
673
- "loss": 1.0763,
674
- "step": 22500
675
- },
676
- {
677
- "epoch": 11.99,
678
- "eval_loss": 0.2549177408218384,
679
- "eval_runtime": 379.953,
680
- "eval_samples_per_second": 11.309,
681
- "eval_steps_per_second": 0.355,
682
- "eval_wer": 0.15720564855512953,
683
- "step": 22500
684
- },
685
- {
686
- "epoch": 12.0,
687
- "step": 22524,
688
- "total_flos": 6.281601139352125e+20,
689
- "train_loss": 0.5676147035501541,
690
- "train_runtime": 172546.67,
691
- "train_samples_per_second": 16.714,
692
- "train_steps_per_second": 0.131
693
- }
694
- ],
695
- "max_steps": 22524,
696
- "num_train_epochs": 12,
697
- "total_flos": 6.281601139352125e+20,
698
- "trial_name": null,
699
- "trial_params": null
700
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.ipynb_checkpoints/vocab-checkpoint.json DELETED
@@ -1 +0,0 @@
1
- {"#": 1, "'": 2, "-": 3, "a": 4, "b": 5, "c": 6, "d": 7, "e": 8, "f": 9, "g": 10, "h": 11, "i": 12, "j": 13, "k": 14, "l": 15, "m": 16, "n": 17, "o": 18, "p": 19, "q": 20, "r": 21, "s": 22, "t": 23, "u": 24, "v": 25, "w": 26, "x": 27, "y": 28, "z": 29, "·": 30, "à": 31, "ç": 32, "è": 33, "é": 34, "í": 35, "ï": 36, "ò": 37, "ó": 38, "ú": 39, "ü": 40, "ः": 41, "|": 0, "[UNK]": 42, "[PAD]": 43}
 
 
special_tokens_map.json CHANGED
@@ -1 +1 @@
1
- {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}