Hugging Face's logo

speech31
/
wavlm-large-english-ipa

Automatic Speech Recognition
Transformers
PyTorch
wav2vec2
Inference Endpoints
Model card Files Community
speech31 commited on
Commit
b7deede
•
1 Parent(s): 66c9115

upload model

Browse files
Files changed (12) hide show
  1. config.json +99 -0
  2. optimizer.pt +3 -0
  3. preprocessor_config.json +9 -0
  4. pytorch_model.bin +3 -0
  5. rng_state.pth +3 -0
  6. scaler.pt +3 -0
  7. scheduler.pt +3 -0
  8. special_tokens_map.json +1 -0
  9. tokenizer_config.json +1 -0
  10. trainer_state.json +601 -0
  11. training_args.bin +3 -0
  12. vocab.json +1 -0
config.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/wavlm-large",
3
+ "activation_dropout": 0.0,
4
+ "adapter_kernel_size": 3,
5
+ "adapter_stride": 2,
6
+ "add_adapter": false,
7
+ "apply_spec_augment": true,
8
+ "architectures": [
9
+ "Wav2Vec2ForCTC"
10
+ ],
11
+ "attention_dropout": 0.0,
12
+ "bos_token_id": 1,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 768,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": false,
17
+ "conv_dim": [
18
+ 512,
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512
25
+ ],
26
+ "conv_kernel": [
27
+ 10,
28
+ 3,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 2,
33
+ 2
34
+ ],
35
+ "conv_stride": [
36
+ 5,
37
+ 2,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2
43
+ ],
44
+ "ctc_loss_reduction": "mean",
45
+ "ctc_zero_infinity": false,
46
+ "diversity_loss_weight": 0.1,
47
+ "do_stable_layer_norm": true,
48
+ "eos_token_id": 2,
49
+ "feat_extract_activation": "gelu",
50
+ "feat_extract_dropout": 0.0,
51
+ "feat_extract_norm": "layer",
52
+ "feat_proj_dropout": 0.0,
53
+ "feat_quantizer_dropout": 0.0,
54
+ "final_dropout": 0.0,
55
+ "gradient_checkpointing": false,
56
+ "hidden_act": "gelu",
57
+ "hidden_dropout": 0.0,
58
+ "hidden_size": 1024,
59
+ "initializer_range": 0.02,
60
+ "intermediate_size": 4096,
61
+ "layer_norm_eps": 1e-05,
62
+ "layerdrop": 0.0,
63
+ "mask_channel_length": 10,
64
+ "mask_channel_min_space": 1,
65
+ "mask_channel_other": 0.0,
66
+ "mask_channel_prob": 0.0,
67
+ "mask_channel_selection": "static",
68
+ "mask_feature_length": 10,
69
+ "mask_feature_min_masks": 0,
70
+ "mask_feature_prob": 0.0,
71
+ "mask_time_length": 10,
72
+ "mask_time_min_masks": 2,
73
+ "mask_time_min_space": 1,
74
+ "mask_time_other": 0.0,
75
+ "mask_time_prob": 0.05,
76
+ "mask_time_selection": "static",
77
+ "max_bucket_distance": 800,
78
+ "model_type": "wav2vec2",
79
+ "num_adapter_layers": 3,
80
+ "num_attention_heads": 16,
81
+ "num_buckets": 320,
82
+ "num_codevector_groups": 2,
83
+ "num_codevectors_per_group": 320,
84
+ "num_conv_pos_embedding_groups": 16,
85
+ "num_conv_pos_embeddings": 128,
86
+ "num_ctc_classes": 80,
87
+ "num_feat_extract_layers": 7,
88
+ "num_hidden_layers": 24,
89
+ "num_negatives": 100,
90
+ "output_hidden_size": 1024,
91
+ "pad_token_id": 42,
92
+ "proj_codevector_dim": 768,
93
+ "replace_prob": 0.5,
94
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
95
+ "torch_dtype": "float32",
96
+ "transformers_version": "4.11.3",
97
+ "use_weighted_layer_sum": false,
98
+ "vocab_size": 43
99
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46507e5fa1d44e79bab0c18b9eabf2a18f16d1e4081b09e8d204469f58a523d9
3
+ size 2490411473
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f50f0a5e4593e52a661119f0781b0a132d855be30259e0c4364681b333217a1
3
+ size 1262083263
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1233c8e11c71db13e502510231ab6138b867a153025d3498e035c655424c9a27
3
+ size 14567
scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04f3d820036fd6de9b35ec4dda4c2da42adb8e7e642417535fba4b6671a81cc1
3
+ size 559
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc9b08a31d09e16a1debad1cdf01fd7e172e3d8fa1dea22330415ddc352ec9e5
3
+ size 623
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]"}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "tokenizer_class": "Wav2Vec2CTCTokenizer"}
trainer_state.json ADDED
@@ -0,0 +1,601 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.20890280604362488,
3
+ "best_model_checkpoint": "wavlm-english-phoneme-v2/checkpoint-7500",
4
+ "epoch": 57.69230769230769,
5
+ "global_step": 7500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.77,
12
+ "learning_rate": 2.9999999999999997e-05,
13
+ "loss": 5.081,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 1.54,
18
+ "learning_rate": 5.9999999999999995e-05,
19
+ "loss": 2.595,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 2.31,
24
+ "learning_rate": 8.999999999999999e-05,
25
+ "loss": 2.5161,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 3.08,
30
+ "learning_rate": 0.00011999999999999999,
31
+ "loss": 2.226,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 3.85,
36
+ "learning_rate": 0.00015,
37
+ "loss": 1.8893,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 3.85,
42
+ "eval_cer": 0.7476650184443921,
43
+ "eval_loss": 1.3937348127365112,
44
+ "eval_runtime": 30.2371,
45
+ "eval_samples_per_second": 15.279,
46
+ "eval_steps_per_second": 1.918,
47
+ "step": 500
48
+ },
49
+ {
50
+ "epoch": 4.62,
51
+ "learning_rate": 0.00017999999999999998,
52
+ "loss": 1.1918,
53
+ "step": 600
54
+ },
55
+ {
56
+ "epoch": 5.38,
57
+ "learning_rate": 0.00020999999999999998,
58
+ "loss": 0.8122,
59
+ "step": 700
60
+ },
61
+ {
62
+ "epoch": 6.15,
63
+ "learning_rate": 0.00023999999999999998,
64
+ "loss": 0.6064,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 6.92,
69
+ "learning_rate": 0.00027,
70
+ "loss": 0.4715,
71
+ "step": 900
72
+ },
73
+ {
74
+ "epoch": 7.69,
75
+ "learning_rate": 0.0003,
76
+ "loss": 0.3875,
77
+ "step": 1000
78
+ },
79
+ {
80
+ "epoch": 7.69,
81
+ "eval_cer": 0.1790283337257672,
82
+ "eval_loss": 0.36919161677360535,
83
+ "eval_runtime": 29.9248,
84
+ "eval_samples_per_second": 15.439,
85
+ "eval_steps_per_second": 1.938,
86
+ "step": 1000
87
+ },
88
+ {
89
+ "epoch": 8.46,
90
+ "learning_rate": 0.00029558823529411763,
91
+ "loss": 0.3297,
92
+ "step": 1100
93
+ },
94
+ {
95
+ "epoch": 9.23,
96
+ "learning_rate": 0.00029117647058823524,
97
+ "loss": 0.2736,
98
+ "step": 1200
99
+ },
100
+ {
101
+ "epoch": 10.0,
102
+ "learning_rate": 0.00028676470588235296,
103
+ "loss": 0.2389,
104
+ "step": 1300
105
+ },
106
+ {
107
+ "epoch": 10.77,
108
+ "learning_rate": 0.00028235294117647056,
109
+ "loss": 0.1919,
110
+ "step": 1400
111
+ },
112
+ {
113
+ "epoch": 11.54,
114
+ "learning_rate": 0.0002779411764705882,
115
+ "loss": 0.1718,
116
+ "step": 1500
117
+ },
118
+ {
119
+ "epoch": 11.54,
120
+ "eval_cer": 0.11992779216701986,
121
+ "eval_loss": 0.2612331211566925,
122
+ "eval_runtime": 30.5393,
123
+ "eval_samples_per_second": 15.128,
124
+ "eval_steps_per_second": 1.899,
125
+ "step": 1500
126
+ },
127
+ {
128
+ "epoch": 12.31,
129
+ "learning_rate": 0.00027352941176470583,
130
+ "loss": 0.1501,
131
+ "step": 1600
132
+ },
133
+ {
134
+ "epoch": 13.08,
135
+ "learning_rate": 0.0002691176470588235,
136
+ "loss": 0.1404,
137
+ "step": 1700
138
+ },
139
+ {
140
+ "epoch": 13.85,
141
+ "learning_rate": 0.00026470588235294115,
142
+ "loss": 0.1198,
143
+ "step": 1800
144
+ },
145
+ {
146
+ "epoch": 14.62,
147
+ "learning_rate": 0.0002602941176470588,
148
+ "loss": 0.1128,
149
+ "step": 1900
150
+ },
151
+ {
152
+ "epoch": 15.38,
153
+ "learning_rate": 0.0002558823529411764,
154
+ "loss": 0.1022,
155
+ "step": 2000
156
+ },
157
+ {
158
+ "epoch": 15.38,
159
+ "eval_cer": 0.1011694529471784,
160
+ "eval_loss": 0.23615849018096924,
161
+ "eval_runtime": 37.8684,
162
+ "eval_samples_per_second": 12.2,
163
+ "eval_steps_per_second": 1.532,
164
+ "step": 2000
165
+ },
166
+ {
167
+ "epoch": 16.15,
168
+ "learning_rate": 0.0002514705882352941,
169
+ "loss": 0.0933,
170
+ "step": 2100
171
+ },
172
+ {
173
+ "epoch": 16.92,
174
+ "learning_rate": 0.00024705882352941174,
175
+ "loss": 0.0894,
176
+ "step": 2200
177
+ },
178
+ {
179
+ "epoch": 17.69,
180
+ "learning_rate": 0.0002426470588235294,
181
+ "loss": 0.0818,
182
+ "step": 2300
183
+ },
184
+ {
185
+ "epoch": 18.46,
186
+ "learning_rate": 0.000238235294117647,
187
+ "loss": 0.0745,
188
+ "step": 2400
189
+ },
190
+ {
191
+ "epoch": 19.23,
192
+ "learning_rate": 0.0002338676470588235,
193
+ "loss": 0.0735,
194
+ "step": 2500
195
+ },
196
+ {
197
+ "epoch": 19.23,
198
+ "eval_cer": 0.0909661721999843,
199
+ "eval_loss": 0.21879912912845612,
200
+ "eval_runtime": 19.7011,
201
+ "eval_samples_per_second": 23.451,
202
+ "eval_steps_per_second": 2.944,
203
+ "step": 2500
204
+ },
205
+ {
206
+ "epoch": 20.0,
207
+ "learning_rate": 0.00022945588235294116,
208
+ "loss": 0.0678,
209
+ "step": 2600
210
+ },
211
+ {
212
+ "epoch": 20.77,
213
+ "learning_rate": 0.0002250441176470588,
214
+ "loss": 0.0643,
215
+ "step": 2700
216
+ },
217
+ {
218
+ "epoch": 21.54,
219
+ "learning_rate": 0.00022063235294117646,
220
+ "loss": 0.0592,
221
+ "step": 2800
222
+ },
223
+ {
224
+ "epoch": 22.31,
225
+ "learning_rate": 0.0002162205882352941,
226
+ "loss": 0.0595,
227
+ "step": 2900
228
+ },
229
+ {
230
+ "epoch": 23.08,
231
+ "learning_rate": 0.00021180882352941175,
232
+ "loss": 0.0555,
233
+ "step": 3000
234
+ },
235
+ {
236
+ "epoch": 23.08,
237
+ "eval_cer": 0.08170473275253119,
238
+ "eval_loss": 0.22275137901306152,
239
+ "eval_runtime": 28.5681,
240
+ "eval_samples_per_second": 16.172,
241
+ "eval_steps_per_second": 2.03,
242
+ "step": 3000
243
+ },
244
+ {
245
+ "epoch": 23.85,
246
+ "learning_rate": 0.0002073970588235294,
247
+ "loss": 0.05,
248
+ "step": 3100
249
+ },
250
+ {
251
+ "epoch": 24.62,
252
+ "learning_rate": 0.00020298529411764705,
253
+ "loss": 0.0492,
254
+ "step": 3200
255
+ },
256
+ {
257
+ "epoch": 25.38,
258
+ "learning_rate": 0.00019857352941176468,
259
+ "loss": 0.0469,
260
+ "step": 3300
261
+ },
262
+ {
263
+ "epoch": 26.15,
264
+ "learning_rate": 0.00019416176470588234,
265
+ "loss": 0.045,
266
+ "step": 3400
267
+ },
268
+ {
269
+ "epoch": 26.92,
270
+ "learning_rate": 0.00018974999999999998,
271
+ "loss": 0.0432,
272
+ "step": 3500
273
+ },
274
+ {
275
+ "epoch": 26.92,
276
+ "eval_cer": 0.08044894435287654,
277
+ "eval_loss": 0.21576076745986938,
278
+ "eval_runtime": 28.5989,
279
+ "eval_samples_per_second": 16.154,
280
+ "eval_steps_per_second": 2.028,
281
+ "step": 3500
282
+ },
283
+ {
284
+ "epoch": 27.69,
285
+ "learning_rate": 0.00018533823529411764,
286
+ "loss": 0.0427,
287
+ "step": 3600
288
+ },
289
+ {
290
+ "epoch": 28.46,
291
+ "learning_rate": 0.00018092647058823527,
292
+ "loss": 0.0387,
293
+ "step": 3700
294
+ },
295
+ {
296
+ "epoch": 29.23,
297
+ "learning_rate": 0.00017651470588235293,
298
+ "loss": 0.0391,
299
+ "step": 3800
300
+ },
301
+ {
302
+ "epoch": 30.0,
303
+ "learning_rate": 0.0001721029411764706,
304
+ "loss": 0.0371,
305
+ "step": 3900
306
+ },
307
+ {
308
+ "epoch": 30.77,
309
+ "learning_rate": 0.00016769117647058823,
310
+ "loss": 0.0339,
311
+ "step": 4000
312
+ },
313
+ {
314
+ "epoch": 30.77,
315
+ "eval_cer": 0.07707401302880465,
316
+ "eval_loss": 0.21580535173416138,
317
+ "eval_runtime": 28.7435,
318
+ "eval_samples_per_second": 16.073,
319
+ "eval_steps_per_second": 2.018,
320
+ "step": 4000
321
+ },
322
+ {
323
+ "epoch": 31.54,
324
+ "learning_rate": 0.0001632794117647059,
325
+ "loss": 0.0335,
326
+ "step": 4100
327
+ },
328
+ {
329
+ "epoch": 32.31,
330
+ "learning_rate": 0.00015886764705882352,
331
+ "loss": 0.0312,
332
+ "step": 4200
333
+ },
334
+ {
335
+ "epoch": 33.08,
336
+ "learning_rate": 0.00015445588235294118,
337
+ "loss": 0.03,
338
+ "step": 4300
339
+ },
340
+ {
341
+ "epoch": 33.85,
342
+ "learning_rate": 0.0001500441176470588,
343
+ "loss": 0.0297,
344
+ "step": 4400
345
+ },
346
+ {
347
+ "epoch": 34.62,
348
+ "learning_rate": 0.00014563235294117645,
349
+ "loss": 0.0257,
350
+ "step": 4500
351
+ },
352
+ {
353
+ "epoch": 34.62,
354
+ "eval_cer": 0.07487638332940899,
355
+ "eval_loss": 0.2141367644071579,
356
+ "eval_runtime": 28.3525,
357
+ "eval_samples_per_second": 16.295,
358
+ "eval_steps_per_second": 2.046,
359
+ "step": 4500
360
+ },
361
+ {
362
+ "epoch": 35.38,
363
+ "learning_rate": 0.0001412205882352941,
364
+ "loss": 0.0271,
365
+ "step": 4600
366
+ },
367
+ {
368
+ "epoch": 36.15,
369
+ "learning_rate": 0.00013680882352941175,
370
+ "loss": 0.0259,
371
+ "step": 4700
372
+ },
373
+ {
374
+ "epoch": 36.92,
375
+ "learning_rate": 0.0001323970588235294,
376
+ "loss": 0.025,
377
+ "step": 4800
378
+ },
379
+ {
380
+ "epoch": 37.69,
381
+ "learning_rate": 0.00012798529411764704,
382
+ "loss": 0.0258,
383
+ "step": 4900
384
+ },
385
+ {
386
+ "epoch": 38.46,
387
+ "learning_rate": 0.0001235735294117647,
388
+ "loss": 0.0228,
389
+ "step": 5000
390
+ },
391
+ {
392
+ "epoch": 38.46,
393
+ "eval_cer": 0.0705596107055961,
394
+ "eval_loss": 0.21314646303653717,
395
+ "eval_runtime": 19.2794,
396
+ "eval_samples_per_second": 23.963,
397
+ "eval_steps_per_second": 3.008,
398
+ "step": 5000
399
+ },
400
+ {
401
+ "epoch": 39.23,
402
+ "learning_rate": 0.00011916176470588235,
403
+ "loss": 0.0233,
404
+ "step": 5100
405
+ },
406
+ {
407
+ "epoch": 40.0,
408
+ "learning_rate": 0.00011475,
409
+ "loss": 0.0204,
410
+ "step": 5200
411
+ },
412
+ {
413
+ "epoch": 40.77,
414
+ "learning_rate": 0.00011033823529411764,
415
+ "loss": 0.0194,
416
+ "step": 5300
417
+ },
418
+ {
419
+ "epoch": 41.54,
420
+ "learning_rate": 0.00010592647058823528,
421
+ "loss": 0.0167,
422
+ "step": 5400
423
+ },
424
+ {
425
+ "epoch": 42.31,
426
+ "learning_rate": 0.00010151470588235293,
427
+ "loss": 0.0189,
428
+ "step": 5500
429
+ },
430
+ {
431
+ "epoch": 42.31,
432
+ "eval_cer": 0.06828349423122204,
433
+ "eval_loss": 0.21500647068023682,
434
+ "eval_runtime": 19.6812,
435
+ "eval_samples_per_second": 23.474,
436
+ "eval_steps_per_second": 2.947,
437
+ "step": 5500
438
+ },
439
+ {
440
+ "epoch": 43.08,
441
+ "learning_rate": 9.710294117647057e-05,
442
+ "loss": 0.0189,
443
+ "step": 5600
444
+ },
445
+ {
446
+ "epoch": 43.85,
447
+ "learning_rate": 9.269117647058822e-05,
448
+ "loss": 0.0182,
449
+ "step": 5700
450
+ },
451
+ {
452
+ "epoch": 44.62,
453
+ "learning_rate": 8.827941176470587e-05,
454
+ "loss": 0.0189,
455
+ "step": 5800
456
+ },
457
+ {
458
+ "epoch": 45.38,
459
+ "learning_rate": 8.386764705882352e-05,
460
+ "loss": 0.0168,
461
+ "step": 5900
462
+ },
463
+ {
464
+ "epoch": 46.15,
465
+ "learning_rate": 7.945588235294116e-05,
466
+ "loss": 0.0141,
467
+ "step": 6000
468
+ },
469
+ {
470
+ "epoch": 46.15,
471
+ "eval_cer": 0.0661643513068048,
472
+ "eval_loss": 0.21131280064582825,
473
+ "eval_runtime": 20.235,
474
+ "eval_samples_per_second": 22.832,
475
+ "eval_steps_per_second": 2.866,
476
+ "step": 6000
477
+ },
478
+ {
479
+ "epoch": 46.92,
480
+ "learning_rate": 7.504411764705881e-05,
481
+ "loss": 0.0155,
482
+ "step": 6100
483
+ },
484
+ {
485
+ "epoch": 47.69,
486
+ "learning_rate": 7.063235294117646e-05,
487
+ "loss": 0.0142,
488
+ "step": 6200
489
+ },
490
+ {
491
+ "epoch": 48.46,
492
+ "learning_rate": 6.62205882352941e-05,
493
+ "loss": 0.0139,
494
+ "step": 6300
495
+ },
496
+ {
497
+ "epoch": 49.23,
498
+ "learning_rate": 6.180882352941175e-05,
499
+ "loss": 0.0146,
500
+ "step": 6400
501
+ },
502
+ {
503
+ "epoch": 50.0,
504
+ "learning_rate": 5.739705882352941e-05,
505
+ "loss": 0.0118,
506
+ "step": 6500
507
+ },
508
+ {
509
+ "epoch": 50.0,
510
+ "eval_cer": 0.06388823483243074,
511
+ "eval_loss": 0.2127513736486435,
512
+ "eval_runtime": 19.4356,
513
+ "eval_samples_per_second": 23.771,
514
+ "eval_steps_per_second": 2.984,
515
+ "step": 6500
516
+ },
517
+ {
518
+ "epoch": 50.77,
519
+ "learning_rate": 5.298529411764705e-05,
520
+ "loss": 0.0117,
521
+ "step": 6600
522
+ },
523
+ {
524
+ "epoch": 51.54,
525
+ "learning_rate": 4.85735294117647e-05,
526
+ "loss": 0.0123,
527
+ "step": 6700
528
+ },
529
+ {
530
+ "epoch": 52.31,
531
+ "learning_rate": 4.416176470588235e-05,
532
+ "loss": 0.0113,
533
+ "step": 6800
534
+ },
535
+ {
536
+ "epoch": 53.08,
537
+ "learning_rate": 3.975e-05,
538
+ "loss": 0.0111,
539
+ "step": 6900
540
+ },
541
+ {
542
+ "epoch": 53.85,
543
+ "learning_rate": 3.5338235294117645e-05,
544
+ "loss": 0.011,
545
+ "step": 7000
546
+ },
547
+ {
548
+ "epoch": 53.85,
549
+ "eval_cer": 0.06365277450749549,
550
+ "eval_loss": 0.20946960151195526,
551
+ "eval_runtime": 19.7474,
552
+ "eval_samples_per_second": 23.395,
553
+ "eval_steps_per_second": 2.937,
554
+ "step": 7000
555
+ },
556
+ {
557
+ "epoch": 54.62,
558
+ "learning_rate": 3.092647058823529e-05,
559
+ "loss": 0.0098,
560
+ "step": 7100
561
+ },
562
+ {
563
+ "epoch": 55.38,
564
+ "learning_rate": 2.651470588235294e-05,
565
+ "loss": 0.0105,
566
+ "step": 7200
567
+ },
568
+ {
569
+ "epoch": 56.15,
570
+ "learning_rate": 2.2102941176470584e-05,
571
+ "loss": 0.0097,
572
+ "step": 7300
573
+ },
574
+ {
575
+ "epoch": 56.92,
576
+ "learning_rate": 1.7691176470588234e-05,
577
+ "loss": 0.0095,
578
+ "step": 7400
579
+ },
580
+ {
581
+ "epoch": 57.69,
582
+ "learning_rate": 1.3279411764705882e-05,
583
+ "loss": 0.0097,
584
+ "step": 7500
585
+ },
586
+ {
587
+ "epoch": 57.69,
588
+ "eval_cer": 0.06121968448316459,
589
+ "eval_loss": 0.20890280604362488,
590
+ "eval_runtime": 20.5541,
591
+ "eval_samples_per_second": 22.477,
592
+ "eval_steps_per_second": 2.822,
593
+ "step": 7500
594
+ }
595
+ ],
596
+ "max_steps": 7800,
597
+ "num_train_epochs": 60,
598
+ "total_flos": 2.2797274353269838e+19,
599
+ "trial_name": null,
600
+ "trial_params": null
601
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d509dbe288188cdbc607ac6131a12e403dffe07b8c3e65dadbcd4531ba6c6980
3
+ size 2863
vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"s": 1, "ə": 2, "aj": 3, "θ": 4, "ej": 5, "t͡ʃ": 6, "i": 7, "f": 8, "w": 9, "ʊ": 10, "ŋ": 11, "z": 12, "t": 13, "ɪ": 14, "ɹ̩": 15, "l": 16, "ʒ": 17, "ʃ": 18, "aw": 19, "j": 20, "ɹ": 21, "ð": 22, "v": 23, "ɛ": 24, "æ": 25, "u": 26, "h": 27, "d": 28, "d͡ʒ": 29, "ɑ": 30, "oj": 31, "k": 32, "ɡ": 33, "p": 34, "ow": 35, "ʌ": 36, "b": 37, "ɔ": 38, "n": 39, "m": 40, "|": 0, "[UNK]": 41, "[PAD]": 42}