Hugging Face's logo

nrshoudi
/
hubert_base_arabic_mdd

Automatic Speech Recognition
Transformers
Safetensors
hubert
Generated from Trainer
Inference Endpoints
Model card Files Community
nrshoudi commited on
Commit
8edf60c
•
1 Parent(s): 4b2b61a

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-base-ls960
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - wer
8
+ model-index:
9
+ - name: hubert_base_arabic_mdd
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_base_arabic_mdd
17
+
18
+ This model is a fine-tuned version of [facebook/hubert-base-ls960](https://huggingface.co/facebook/hubert-base-ls960) on an unknown dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 3.2265
21
+ - Wer: 1.0
22
+ - Per: 1.0
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
+ | 6.526 | 1.0 | 1637 | 3.3650 | 1.0 | 1.0 |
55
+ | 3.2555 | 2.0 | 3274 | 3.2755 | 1.0 | 1.0 |
56
+ | 3.2548 | 3.0 | 4911 | 3.2238 | 1.0 | 1.0 |
57
+ | 3.2385 | 4.0 | 6548 | 3.2845 | 1.0 | 1.0 |
58
+ | 3.2358 | 5.0 | 8185 | 3.2271 | 1.0 | 1.0 |
59
+ | 3.237 | 6.0 | 9822 | 3.2473 | 1.0 | 1.0 |
60
+ | 3.2622 | 7.0 | 11459 | 3.2289 | 1.0 | 1.0 |
61
+ | 3.2614 | 8.0 | 13096 | 3.2283 | 1.0 | 1.0 |
62
+ | 3.224 | 9.0 | 14733 | 3.2249 | 1.0 | 1.0 |
63
+ | 3.2221 | 10.0 | 16370 | 3.2335 | 1.0 | 1.0 |
64
+ | 3.222 | 11.0 | 18007 | 3.2357 | 1.0 | 1.0 |
65
+ | 3.2218 | 12.0 | 19644 | 3.2491 | 1.0 | 1.0 |
66
+ | 3.2183 | 13.0 | 21281 | 3.2446 | 1.0 | 1.0 |
67
+ | 3.2181 | 14.0 | 22918 | 3.2416 | 1.0 | 1.0 |
68
+ | 3.2164 | 15.0 | 24555 | 3.2259 | 1.0 | 1.0 |
69
+ | 3.2148 | 16.0 | 26192 | 3.2249 | 1.0 | 1.0 |
70
+ | 3.2139 | 17.0 | 27829 | 3.2327 | 1.0 | 1.0 |
71
+ | 3.2133 | 18.0 | 29466 | 3.2251 | 1.0 | 1.0 |
72
+ | 3.2128 | 19.0 | 31103 | 3.2288 | 1.0 | 1.0 |
73
+ | 3.2113 | 20.0 | 32740 | 3.2265 | 1.0 | 1.0 |
74
+
75
+
76
+ ### Framework versions
77
+
78
+ - Transformers 4.35.2
79
+ - Pytorch 2.1.0+cu118
80
+ - Datasets 2.15.0
81
+ - Tokenizers 0.15.0
config.json ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/hubert-base-ls960",
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": false,
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
+ "do_stable_layer_norm": false,
42
+ "eos_token_id": 2,
43
+ "feat_extract_activation": "gelu",
44
+ "feat_extract_dropout": 0.0,
45
+ "feat_extract_norm": "group",
46
+ "feat_proj_dropout": 0.1,
47
+ "feat_proj_layer_norm": true,
48
+ "final_dropout": 0.1,
49
+ "gradient_checkpointing": false,
50
+ "hidden_act": "gelu",
51
+ "hidden_dropout": 0.1,
52
+ "hidden_dropout_prob": 0.1,
53
+ "hidden_size": 768,
54
+ "initializer_range": 0.02,
55
+ "intermediate_size": 3072,
56
+ "layer_norm_eps": 1e-05,
57
+ "layerdrop": 0.1,
58
+ "mask_feature_length": 10,
59
+ "mask_feature_min_masks": 0,
60
+ "mask_feature_prob": 0.0,
61
+ "mask_time_length": 10,
62
+ "mask_time_min_masks": 2,
63
+ "mask_time_prob": 0.05,
64
+ "model_type": "hubert",
65
+ "num_attention_heads": 12,
66
+ "num_conv_pos_embedding_groups": 16,
67
+ "num_conv_pos_embeddings": 128,
68
+ "num_feat_extract_layers": 7,
69
+ "num_hidden_layers": 12,
70
+ "pad_token_id": 37,
71
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
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:ff6283325d1a1bf35f848cb2d982e77d44d4d48cf78dd77191944ba4405e91a2
3
+ size 377635320
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": false,
9
+ "sampling_rate": 16000
10
+ }
trainer_state.json ADDED
@@ -0,0 +1,348 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 3.2237985134124756,
3
+ "best_model_checkpoint": "nrshoudi/hubert_base_arabic_mdd/checkpoint-4911",
4
+ "epoch": 20.0,
5
+ "eval_steps": 500,
6
+ "global_step": 32740,
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": 6.526,
15
+ "step": 1637
16
+ },
17
+ {
18
+ "epoch": 1.0,
19
+ "eval_loss": 3.364994525909424,
20
+ "eval_per": 1.0,
21
+ "eval_runtime": 79.0541,
22
+ "eval_samples_per_second": 10.284,
23
+ "eval_steps_per_second": 5.148,
24
+ "eval_wer": 1.0,
25
+ "step": 1637
26
+ },
27
+ {
28
+ "epoch": 2.0,
29
+ "learning_rate": 0.0001,
30
+ "loss": 3.2555,
31
+ "step": 3274
32
+ },
33
+ {
34
+ "epoch": 2.0,
35
+ "eval_loss": 3.275482416152954,
36
+ "eval_per": 1.0,
37
+ "eval_runtime": 81.9459,
38
+ "eval_samples_per_second": 9.921,
39
+ "eval_steps_per_second": 4.967,
40
+ "eval_wer": 1.0,
41
+ "step": 3274
42
+ },
43
+ {
44
+ "epoch": 3.0,
45
+ "learning_rate": 9.444444444444444e-05,
46
+ "loss": 3.2548,
47
+ "step": 4911
48
+ },
49
+ {
50
+ "epoch": 3.0,
51
+ "eval_loss": 3.2237985134124756,
52
+ "eval_per": 1.0,
53
+ "eval_runtime": 79.8785,
54
+ "eval_samples_per_second": 10.178,
55
+ "eval_steps_per_second": 5.095,
56
+ "eval_wer": 1.0,
57
+ "step": 4911
58
+ },
59
+ {
60
+ "epoch": 4.0,
61
+ "learning_rate": 8.888888888888889e-05,
62
+ "loss": 3.2385,
63
+ "step": 6548
64
+ },
65
+ {
66
+ "epoch": 4.0,
67
+ "eval_loss": 3.284461259841919,
68
+ "eval_per": 1.0,
69
+ "eval_runtime": 80.0287,
70
+ "eval_samples_per_second": 10.159,
71
+ "eval_steps_per_second": 5.086,
72
+ "eval_wer": 1.0,
73
+ "step": 6548
74
+ },
75
+ {
76
+ "epoch": 5.0,
77
+ "learning_rate": 8.333333333333334e-05,
78
+ "loss": 3.2358,
79
+ "step": 8185
80
+ },
81
+ {
82
+ "epoch": 5.0,
83
+ "eval_loss": 3.227111339569092,
84
+ "eval_per": 1.0,
85
+ "eval_runtime": 79.4783,
86
+ "eval_samples_per_second": 10.229,
87
+ "eval_steps_per_second": 5.121,
88
+ "eval_wer": 1.0,
89
+ "step": 8185
90
+ },
91
+ {
92
+ "epoch": 6.0,
93
+ "learning_rate": 7.777777777777778e-05,
94
+ "loss": 3.237,
95
+ "step": 9822
96
+ },
97
+ {
98
+ "epoch": 6.0,
99
+ "eval_loss": 3.2472596168518066,
100
+ "eval_per": 1.0,
101
+ "eval_runtime": 80.029,
102
+ "eval_samples_per_second": 10.159,
103
+ "eval_steps_per_second": 5.086,
104
+ "eval_wer": 1.0,
105
+ "step": 9822
106
+ },
107
+ {
108
+ "epoch": 7.0,
109
+ "learning_rate": 7.222222222222222e-05,
110
+ "loss": 3.2622,
111
+ "step": 11459
112
+ },
113
+ {
114
+ "epoch": 7.0,
115
+ "eval_loss": 3.2289435863494873,
116
+ "eval_per": 1.0,
117
+ "eval_runtime": 81.2354,
118
+ "eval_samples_per_second": 10.008,
119
+ "eval_steps_per_second": 5.01,
120
+ "eval_wer": 1.0,
121
+ "step": 11459
122
+ },
123
+ {
124
+ "epoch": 8.0,
125
+ "learning_rate": 6.666666666666667e-05,
126
+ "loss": 3.2614,
127
+ "step": 13096
128
+ },
129
+ {
130
+ "epoch": 8.0,
131
+ "eval_loss": 3.2283053398132324,
132
+ "eval_per": 1.0,
133
+ "eval_runtime": 80.9742,
134
+ "eval_samples_per_second": 10.04,
135
+ "eval_steps_per_second": 5.026,
136
+ "eval_wer": 1.0,
137
+ "step": 13096
138
+ },
139
+ {
140
+ "epoch": 9.0,
141
+ "learning_rate": 6.111111111111112e-05,
142
+ "loss": 3.224,
143
+ "step": 14733
144
+ },
145
+ {
146
+ "epoch": 9.0,
147
+ "eval_loss": 3.224926233291626,
148
+ "eval_per": 1.0,
149
+ "eval_runtime": 79.9396,
150
+ "eval_samples_per_second": 10.17,
151
+ "eval_steps_per_second": 5.091,
152
+ "eval_wer": 1.0,
153
+ "step": 14733
154
+ },
155
+ {
156
+ "epoch": 10.0,
157
+ "learning_rate": 5.555555555555556e-05,
158
+ "loss": 3.2221,
159
+ "step": 16370
160
+ },
161
+ {
162
+ "epoch": 10.0,
163
+ "eval_loss": 3.233494520187378,
164
+ "eval_per": 1.0,
165
+ "eval_runtime": 79.7551,
166
+ "eval_samples_per_second": 10.194,
167
+ "eval_steps_per_second": 5.103,
168
+ "eval_wer": 1.0,
169
+ "step": 16370
170
+ },
171
+ {
172
+ "epoch": 11.0,
173
+ "learning_rate": 5e-05,
174
+ "loss": 3.222,
175
+ "step": 18007
176
+ },
177
+ {
178
+ "epoch": 11.0,
179
+ "eval_loss": 3.2356951236724854,
180
+ "eval_per": 1.0,
181
+ "eval_runtime": 80.1958,
182
+ "eval_samples_per_second": 10.138,
183
+ "eval_steps_per_second": 5.075,
184
+ "eval_wer": 1.0,
185
+ "step": 18007
186
+ },
187
+ {
188
+ "epoch": 12.0,
189
+ "learning_rate": 4.4444444444444447e-05,
190
+ "loss": 3.2218,
191
+ "step": 19644
192
+ },
193
+ {
194
+ "epoch": 12.0,
195
+ "eval_loss": 3.24910569190979,
196
+ "eval_per": 1.0,
197
+ "eval_runtime": 81.0704,
198
+ "eval_samples_per_second": 10.028,
199
+ "eval_steps_per_second": 5.02,
200
+ "eval_wer": 1.0,
201
+ "step": 19644
202
+ },
203
+ {
204
+ "epoch": 13.0,
205
+ "learning_rate": 3.888888888888889e-05,
206
+ "loss": 3.2183,
207
+ "step": 21281
208
+ },
209
+ {
210
+ "epoch": 13.0,
211
+ "eval_loss": 3.2446420192718506,
212
+ "eval_per": 1.0,
213
+ "eval_runtime": 81.004,
214
+ "eval_samples_per_second": 10.037,
215
+ "eval_steps_per_second": 5.024,
216
+ "eval_wer": 1.0,
217
+ "step": 21281
218
+ },
219
+ {
220
+ "epoch": 14.0,
221
+ "learning_rate": 3.3333333333333335e-05,
222
+ "loss": 3.2181,
223
+ "step": 22918
224
+ },
225
+ {
226
+ "epoch": 14.0,
227
+ "eval_loss": 3.2415904998779297,
228
+ "eval_per": 1.0,
229
+ "eval_runtime": 83.7321,
230
+ "eval_samples_per_second": 9.71,
231
+ "eval_steps_per_second": 4.861,
232
+ "eval_wer": 1.0,
233
+ "step": 22918
234
+ },
235
+ {
236
+ "epoch": 15.0,
237
+ "learning_rate": 2.777777777777778e-05,
238
+ "loss": 3.2164,
239
+ "step": 24555
240
+ },
241
+ {
242
+ "epoch": 15.0,
243
+ "eval_loss": 3.22587513923645,
244
+ "eval_per": 1.0,
245
+ "eval_runtime": 81.1128,
246
+ "eval_samples_per_second": 10.023,
247
+ "eval_steps_per_second": 5.018,
248
+ "eval_wer": 1.0,
249
+ "step": 24555
250
+ },
251
+ {
252
+ "epoch": 16.0,
253
+ "learning_rate": 2.2222222222222223e-05,
254
+ "loss": 3.2148,
255
+ "step": 26192
256
+ },
257
+ {
258
+ "epoch": 16.0,
259
+ "eval_loss": 3.224907159805298,
260
+ "eval_per": 1.0,
261
+ "eval_runtime": 80.4106,
262
+ "eval_samples_per_second": 10.111,
263
+ "eval_steps_per_second": 5.062,
264
+ "eval_wer": 1.0,
265
+ "step": 26192
266
+ },
267
+ {
268
+ "epoch": 17.0,
269
+ "learning_rate": 1.6666666666666667e-05,
270
+ "loss": 3.2139,
271
+ "step": 27829
272
+ },
273
+ {
274
+ "epoch": 17.0,
275
+ "eval_loss": 3.232697010040283,
276
+ "eval_per": 1.0,
277
+ "eval_runtime": 80.7791,
278
+ "eval_samples_per_second": 10.064,
279
+ "eval_steps_per_second": 5.038,
280
+ "eval_wer": 1.0,
281
+ "step": 27829
282
+ },
283
+ {
284
+ "epoch": 18.0,
285
+ "learning_rate": 1.1111111111111112e-05,
286
+ "loss": 3.2133,
287
+ "step": 29466
288
+ },
289
+ {
290
+ "epoch": 18.0,
291
+ "eval_loss": 3.2251381874084473,
292
+ "eval_per": 1.0,
293
+ "eval_runtime": 81.6112,
294
+ "eval_samples_per_second": 9.962,
295
+ "eval_steps_per_second": 4.987,
296
+ "eval_wer": 1.0,
297
+ "step": 29466
298
+ },
299
+ {
300
+ "epoch": 19.0,
301
+ "learning_rate": 5.555555555555556e-06,
302
+ "loss": 3.2128,
303
+ "step": 31103
304
+ },
305
+ {
306
+ "epoch": 19.0,
307
+ "eval_loss": 3.2287981510162354,
308
+ "eval_per": 1.0,
309
+ "eval_runtime": 83.1142,
310
+ "eval_samples_per_second": 9.782,
311
+ "eval_steps_per_second": 4.897,
312
+ "eval_wer": 1.0,
313
+ "step": 31103
314
+ },
315
+ {
316
+ "epoch": 20.0,
317
+ "learning_rate": 0.0,
318
+ "loss": 3.2113,
319
+ "step": 32740
320
+ },
321
+ {
322
+ "epoch": 20.0,
323
+ "eval_loss": 3.2265169620513916,
324
+ "eval_per": 1.0,
325
+ "eval_runtime": 81.1257,
326
+ "eval_samples_per_second": 10.021,
327
+ "eval_steps_per_second": 5.017,
328
+ "eval_wer": 1.0,
329
+ "step": 32740
330
+ },
331
+ {
332
+ "epoch": 20.0,
333
+ "step": 32740,
334
+ "total_flos": 3.133881121657545e+18,
335
+ "train_loss": 3.3939919333670585,
336
+ "train_runtime": 10237.0354,
337
+ "train_samples_per_second": 6.394,
338
+ "train_steps_per_second": 3.198
339
+ }
340
+ ],
341
+ "logging_steps": 500,
342
+ "max_steps": 32740,
343
+ "num_train_epochs": 20,
344
+ "save_steps": 500,
345
+ "total_flos": 3.133881121657545e+18,
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:aa4514283d7689fb966aa38c06eabb6fe1f41ea6e9a73cf2690c59f0fd9c5155
3
+ size 4600