Hugging Face's logo

nrshoudi
/
hubert-large-ls960-ft-V2-5

Automatic Speech Recognition
Transformers
Safetensors
hubert
Generated from Trainer
Model card Files
xet
 Community
nrshoudi commited on
Commit
35b68ad
·
1 Parent(s): 496b2a0

End of training

Browse files
Files changed (6) hide show
  1. README.md +81 -0
  2. config.json +76 -0
  3. model.safetensors +3 -0
  4. preprocessor_config.json +10 -0
  5. trainer_state.json +348 -0
  6. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: facebook/hubert-large-ls960-ft
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - wer
8
+ model-index:
9
+ - name: hubert-large-ls960-ft-V2-5
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # hubert-large-ls960-ft-V2-5
17
+
18
+ This model is a fine-tuned version of [facebook/hubert-large-ls960-ft](https://huggingface.co/facebook/hubert-large-ls960-ft) on an unknown dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.5760
21
+ - Wer: 0.1085
22
+ - Per: 0.0892
23
+
24
+ ## Model description
25
+
26
+ More information needed
27
+
28
+ ## Intended uses & limitations
29
+
30
+ More information needed
31
+
32
+ ## Training and evaluation data
33
+
34
+ More information needed
35
+
36
+ ## Training procedure
37
+
38
+ ### Training hyperparameters
39
+
40
+ The following hyperparameters were used during training:
41
+ - learning_rate: 0.0001
42
+ - train_batch_size: 2
43
+ - eval_batch_size: 2
44
+ - seed: 42
45
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
+ - lr_scheduler_type: linear
47
+ - lr_scheduler_warmup_ratio: 0.1
48
+ - num_epochs: 20
49
+
50
+ ### Training results
51
+
52
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Per |
53
+ |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
54
+ | 19.707 | 1.0 | 82 | 3.5254 | 1.0 | 1.0 |
55
+ | 3.4906 | 2.0 | 164 | 3.2483 | 1.0 | 1.0 |
56
+ | 3.233 | 3.0 | 246 | 3.1368 | 1.0 | 1.0 |
57
+ | 3.0468 | 4.0 | 328 | 2.9600 | 1.0 | 1.0 |
58
+ | 2.6751 | 5.0 | 410 | 2.3348 | 1.0 | 1.0 |
59
+ | 2.0881 | 6.0 | 492 | 1.7351 | 0.8568 | 0.8726 |
60
+ | 1.4875 | 7.0 | 574 | 1.2264 | 0.6059 | 0.6134 |
61
+ | 1.0922 | 8.0 | 656 | 0.9666 | 0.4068 | 0.3972 |
62
+ | 0.8148 | 9.0 | 738 | 0.7746 | 0.3249 | 0.3138 |
63
+ | 0.6332 | 10.0 | 820 | 0.6755 | 0.2477 | 0.2313 |
64
+ | 0.4797 | 11.0 | 902 | 0.6262 | 0.1612 | 0.1410 |
65
+ | 0.3807 | 12.0 | 984 | 0.5765 | 0.1384 | 0.1172 |
66
+ | 0.3195 | 13.0 | 1066 | 0.5666 | 0.1191 | 0.0992 |
67
+ | 0.2526 | 14.0 | 1148 | 0.5759 | 0.1165 | 0.0970 |
68
+ | 0.2417 | 15.0 | 1230 | 0.5460 | 0.1138 | 0.0946 |
69
+ | 0.2072 | 16.0 | 1312 | 0.5551 | 0.1095 | 0.0912 |
70
+ | 0.1881 | 17.0 | 1394 | 0.5745 | 0.1102 | 0.0917 |
71
+ | 0.1888 | 18.0 | 1476 | 0.5731 | 0.1094 | 0.0907 |
72
+ | 0.202 | 19.0 | 1558 | 0.5774 | 0.1081 | 0.0893 |
73
+ | 0.1813 | 20.0 | 1640 | 0.5760 | 0.1085 | 0.0892 |
74
+
75
+
76
+ ### Framework versions
77
+
78
+ - Transformers 4.35.2
79
+ - Pytorch 2.1.0+cu121
80
+ - Datasets 2.16.1
81
+ - Tokenizers 0.15.0
config.json ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/hubert-large-ls960-ft",
3
+ "activation_dropout": 0.1,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "HubertForCTC"
7
+ ],
8
+ "attention_dropout": 0.1,
9
+ "bos_token_id": 1,
10
+ "classifier_proj_size": 256,
11
+ "conv_bias": true,
12
+ "conv_dim": [
13
+ 512,
14
+ 512,
15
+ 512,
16
+ 512,
17
+ 512,
18
+ 512,
19
+ 512
20
+ ],
21
+ "conv_kernel": [
22
+ 10,
23
+ 3,
24
+ 3,
25
+ 3,
26
+ 3,
27
+ 2,
28
+ 2
29
+ ],
30
+ "conv_stride": [
31
+ 5,
32
+ 2,
33
+ 2,
34
+ 2,
35
+ 2,
36
+ 2,
37
+ 2
38
+ ],
39
+ "ctc_loss_reduction": "mean",
40
+ "ctc_zero_infinity": false,
41
+ "diversity_loss_weight": 0.1,
42
+ "do_stable_layer_norm": true,
43
+ "eos_token_id": 2,
44
+ "feat_extract_activation": "gelu",
45
+ "feat_extract_dropout": 0.0,
46
+ "feat_extract_norm": "layer",
47
+ "feat_proj_dropout": 0.1,
48
+ "feat_proj_layer_norm": true,
49
+ "final_dropout": 0.1,
50
+ "gradient_checkpointing": false,
51
+ "hidden_act": "gelu",
52
+ "hidden_dropout": 0.1,
53
+ "hidden_dropout_prob": 0.1,
54
+ "hidden_size": 1024,
55
+ "initializer_range": 0.02,
56
+ "intermediate_size": 4096,
57
+ "layer_norm_eps": 1e-05,
58
+ "layerdrop": 0.1,
59
+ "mask_feature_length": 10,
60
+ "mask_feature_min_masks": 0,
61
+ "mask_feature_prob": 0.0,
62
+ "mask_time_length": 10,
63
+ "mask_time_min_masks": 2,
64
+ "mask_time_prob": 0.05,
65
+ "model_type": "hubert",
66
+ "num_attention_heads": 16,
67
+ "num_conv_pos_embedding_groups": 16,
68
+ "num_conv_pos_embeddings": 128,
69
+ "num_feat_extract_layers": 7,
70
+ "num_hidden_layers": 24,
71
+ "pad_token_id": 37,
72
+ "torch_dtype": "float32",
73
+ "transformers_version": "4.35.2",
74
+ "use_weighted_layer_sum": false,
75
+ "vocab_size": 40
76
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d657a31c21aa76a193d8697fbc3e8b71150a4d2ceacf3057fe8acb8c781b616
3
+ size 1261970648
preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
trainer_state.json ADDED
@@ -0,0 +1,348 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.5459895730018616,
3
+ "best_model_checkpoint": "nrshoudi/hubert-large-ls960-ft-V2-5/checkpoint-1230",
4
+ "epoch": 20.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1640,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 1.0,
13
+ "learning_rate": 5e-05,
14
+ "loss": 19.707,
15
+ "step": 82
16
+ },
17
+ {
18
+ "epoch": 1.0,
19
+ "eval_loss": 3.5254127979278564,
20
+ "eval_per": 1.0,
21
+ "eval_runtime": 116.6109,
22
+ "eval_samples_per_second": 6.972,
23
+ "eval_steps_per_second": 3.49,
24
+ "eval_wer": 1.0,
25
+ "step": 82
26
+ },
27
+ {
28
+ "epoch": 2.0,
29
+ "learning_rate": 0.0001,
30
+ "loss": 3.4906,
31
+ "step": 164
32
+ },
33
+ {
34
+ "epoch": 2.0,
35
+ "eval_loss": 3.2483434677124023,
36
+ "eval_per": 1.0,
37
+ "eval_runtime": 115.6174,
38
+ "eval_samples_per_second": 7.032,
39
+ "eval_steps_per_second": 3.52,
40
+ "eval_wer": 1.0,
41
+ "step": 164
42
+ },
43
+ {
44
+ "epoch": 3.0,
45
+ "learning_rate": 9.444444444444444e-05,
46
+ "loss": 3.233,
47
+ "step": 246
48
+ },
49
+ {
50
+ "epoch": 3.0,
51
+ "eval_loss": 3.1367974281311035,
52
+ "eval_per": 1.0,
53
+ "eval_runtime": 116.034,
54
+ "eval_samples_per_second": 7.007,
55
+ "eval_steps_per_second": 3.508,
56
+ "eval_wer": 1.0,
57
+ "step": 246
58
+ },
59
+ {
60
+ "epoch": 4.0,
61
+ "learning_rate": 8.888888888888889e-05,
62
+ "loss": 3.0468,
63
+ "step": 328
64
+ },
65
+ {
66
+ "epoch": 4.0,
67
+ "eval_loss": 2.960031509399414,
68
+ "eval_per": 1.0,
69
+ "eval_runtime": 116.4247,
70
+ "eval_samples_per_second": 6.983,
71
+ "eval_steps_per_second": 3.496,
72
+ "eval_wer": 1.0,
73
+ "step": 328
74
+ },
75
+ {
76
+ "epoch": 5.0,
77
+ "learning_rate": 8.333333333333334e-05,
78
+ "loss": 2.6751,
79
+ "step": 410
80
+ },
81
+ {
82
+ "epoch": 5.0,
83
+ "eval_loss": 2.334815263748169,
84
+ "eval_per": 1.0,
85
+ "eval_runtime": 116.7509,
86
+ "eval_samples_per_second": 6.964,
87
+ "eval_steps_per_second": 3.486,
88
+ "eval_wer": 1.0,
89
+ "step": 410
90
+ },
91
+ {
92
+ "epoch": 6.0,
93
+ "learning_rate": 7.777777777777778e-05,
94
+ "loss": 2.0881,
95
+ "step": 492
96
+ },
97
+ {
98
+ "epoch": 6.0,
99
+ "eval_loss": 1.7351281642913818,
100
+ "eval_per": 0.8726326203712428,
101
+ "eval_runtime": 117.1554,
102
+ "eval_samples_per_second": 6.94,
103
+ "eval_steps_per_second": 3.474,
104
+ "eval_wer": 0.8568428911390042,
105
+ "step": 492
106
+ },
107
+ {
108
+ "epoch": 7.0,
109
+ "learning_rate": 7.222222222222222e-05,
110
+ "loss": 1.4875,
111
+ "step": 574
112
+ },
113
+ {
114
+ "epoch": 7.0,
115
+ "eval_loss": 1.22636878490448,
116
+ "eval_per": 0.613422218034486,
117
+ "eval_runtime": 116.8569,
118
+ "eval_samples_per_second": 6.957,
119
+ "eval_steps_per_second": 3.483,
120
+ "eval_wer": 0.6058513368484905,
121
+ "step": 574
122
+ },
123
+ {
124
+ "epoch": 8.0,
125
+ "learning_rate": 6.666666666666667e-05,
126
+ "loss": 1.0922,
127
+ "step": 656
128
+ },
129
+ {
130
+ "epoch": 8.0,
131
+ "eval_loss": 0.9666064381599426,
132
+ "eval_per": 0.3971779892584566,
133
+ "eval_runtime": 117.3764,
134
+ "eval_samples_per_second": 6.926,
135
+ "eval_steps_per_second": 3.467,
136
+ "eval_wer": 0.4067938966917083,
137
+ "step": 656
138
+ },
139
+ {
140
+ "epoch": 9.0,
141
+ "learning_rate": 6.111111111111112e-05,
142
+ "loss": 0.8148,
143
+ "step": 738
144
+ },
145
+ {
146
+ "epoch": 9.0,
147
+ "eval_loss": 0.7746477127075195,
148
+ "eval_per": 0.31381324790351456,
149
+ "eval_runtime": 117.0529,
150
+ "eval_samples_per_second": 6.946,
151
+ "eval_steps_per_second": 3.477,
152
+ "eval_wer": 0.32485651626149037,
153
+ "step": 738
154
+ },
155
+ {
156
+ "epoch": 10.0,
157
+ "learning_rate": 5.555555555555556e-05,
158
+ "loss": 0.6332,
159
+ "step": 820
160
+ },
161
+ {
162
+ "epoch": 10.0,
163
+ "eval_loss": 0.6754913330078125,
164
+ "eval_per": 0.23132007914821445,
165
+ "eval_runtime": 117.3105,
166
+ "eval_samples_per_second": 6.93,
167
+ "eval_steps_per_second": 3.469,
168
+ "eval_wer": 0.24767859642573842,
169
+ "step": 820
170
+ },
171
+ {
172
+ "epoch": 11.0,
173
+ "learning_rate": 5e-05,
174
+ "loss": 0.4797,
175
+ "step": 902
176
+ },
177
+ {
178
+ "epoch": 11.0,
179
+ "eval_loss": 0.6261923313140869,
180
+ "eval_per": 0.14098275699613683,
181
+ "eval_runtime": 116.5342,
182
+ "eval_samples_per_second": 6.976,
183
+ "eval_steps_per_second": 3.493,
184
+ "eval_wer": 0.16121506229294014,
185
+ "step": 902
186
+ },
187
+ {
188
+ "epoch": 12.0,
189
+ "learning_rate": 4.4444444444444447e-05,
190
+ "loss": 0.3807,
191
+ "step": 984
192
+ },
193
+ {
194
+ "epoch": 12.0,
195
+ "eval_loss": 0.5765030384063721,
196
+ "eval_per": 0.11716762461132574,
197
+ "eval_runtime": 116.5005,
198
+ "eval_samples_per_second": 6.979,
199
+ "eval_steps_per_second": 3.494,
200
+ "eval_wer": 0.13844430964490692,
201
+ "step": 984
202
+ },
203
+ {
204
+ "epoch": 13.0,
205
+ "learning_rate": 3.888888888888889e-05,
206
+ "loss": 0.3195,
207
+ "step": 1066
208
+ },
209
+ {
210
+ "epoch": 13.0,
211
+ "eval_loss": 0.5666270852088928,
212
+ "eval_per": 0.09917082822952983,
213
+ "eval_runtime": 116.7786,
214
+ "eval_samples_per_second": 6.962,
215
+ "eval_steps_per_second": 3.485,
216
+ "eval_wer": 0.11912649899678036,
217
+ "step": 1066
218
+ },
219
+ {
220
+ "epoch": 14.0,
221
+ "learning_rate": 3.3333333333333335e-05,
222
+ "loss": 0.2526,
223
+ "step": 1148
224
+ },
225
+ {
226
+ "epoch": 14.0,
227
+ "eval_loss": 0.5758910775184631,
228
+ "eval_per": 0.09698011872232168,
229
+ "eval_runtime": 116.6256,
230
+ "eval_samples_per_second": 6.971,
231
+ "eval_steps_per_second": 3.49,
232
+ "eval_wer": 0.1165134618076618,
233
+ "step": 1148
234
+ },
235
+ {
236
+ "epoch": 15.0,
237
+ "learning_rate": 2.777777777777778e-05,
238
+ "loss": 0.2417,
239
+ "step": 1230
240
+ },
241
+ {
242
+ "epoch": 15.0,
243
+ "eval_loss": 0.5459895730018616,
244
+ "eval_per": 0.09462451710166776,
245
+ "eval_runtime": 116.6754,
246
+ "eval_samples_per_second": 6.968,
247
+ "eval_steps_per_second": 3.488,
248
+ "eval_wer": 0.113807101861789,
249
+ "step": 1230
250
+ },
251
+ {
252
+ "epoch": 16.0,
253
+ "learning_rate": 2.2222222222222223e-05,
254
+ "loss": 0.2072,
255
+ "step": 1312
256
+ },
257
+ {
258
+ "epoch": 16.0,
259
+ "eval_loss": 0.5550991892814636,
260
+ "eval_per": 0.09116178271930651,
261
+ "eval_runtime": 116.9202,
262
+ "eval_samples_per_second": 6.953,
263
+ "eval_steps_per_second": 3.481,
264
+ "eval_wer": 0.10946759367271709,
265
+ "step": 1312
266
+ },
267
+ {
268
+ "epoch": 17.0,
269
+ "learning_rate": 1.6666666666666667e-05,
270
+ "loss": 0.1881,
271
+ "step": 1394
272
+ },
273
+ {
274
+ "epoch": 17.0,
275
+ "eval_loss": 0.5744515657424927,
276
+ "eval_per": 0.09172712710826345,
277
+ "eval_runtime": 116.943,
278
+ "eval_samples_per_second": 6.952,
279
+ "eval_steps_per_second": 3.48,
280
+ "eval_wer": 0.11016751434837385,
281
+ "step": 1394
282
+ },
283
+ {
284
+ "epoch": 18.0,
285
+ "learning_rate": 1.1111111111111112e-05,
286
+ "loss": 0.1888,
287
+ "step": 1476
288
+ },
289
+ {
290
+ "epoch": 18.0,
291
+ "eval_loss": 0.5731486678123474,
292
+ "eval_per": 0.09069066239517573,
293
+ "eval_runtime": 118.5572,
294
+ "eval_samples_per_second": 6.857,
295
+ "eval_steps_per_second": 3.433,
296
+ "eval_wer": 0.10942093229433998,
297
+ "step": 1476
298
+ },
299
+ {
300
+ "epoch": 19.0,
301
+ "learning_rate": 5.555555555555556e-06,
302
+ "loss": 0.202,
303
+ "step": 1558
304
+ },
305
+ {
306
+ "epoch": 19.0,
307
+ "eval_loss": 0.5773624777793884,
308
+ "eval_per": 0.08927730142278338,
309
+ "eval_runtime": 116.8858,
310
+ "eval_samples_per_second": 6.956,
311
+ "eval_steps_per_second": 3.482,
312
+ "eval_wer": 0.10806775232140357,
313
+ "step": 1558
314
+ },
315
+ {
316
+ "epoch": 20.0,
317
+ "learning_rate": 0.0,
318
+ "loss": 0.1813,
319
+ "step": 1640
320
+ },
321
+ {
322
+ "epoch": 20.0,
323
+ "eval_loss": 0.5759608745574951,
324
+ "eval_per": 0.08915952134175069,
325
+ "eval_runtime": 117.5758,
326
+ "eval_samples_per_second": 6.915,
327
+ "eval_steps_per_second": 3.462,
328
+ "eval_wer": 0.10848770472679763,
329
+ "step": 1640
330
+ },
331
+ {
332
+ "epoch": 20.0,
333
+ "step": 1640,
334
+ "total_flos": 5.333828756305947e+17,
335
+ "train_loss": 2.0454907219584397,
336
+ "train_runtime": 3434.6913,
337
+ "train_samples_per_second": 0.949,
338
+ "train_steps_per_second": 0.477
339
+ }
340
+ ],
341
+ "logging_steps": 500,
342
+ "max_steps": 1640,
343
+ "num_train_epochs": 20,
344
+ "save_steps": 500,
345
+ "total_flos": 5.333828756305947e+17,
346
+ "trial_name": null,
347
+ "trial_params": null
348
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ebbd01c12a3d5ca4e1c0e7ab12d7d3b1506e8e466fa26cdb4eae4c79cab62ae2
3
+ size 4600