Hugging Face's logo

patrickvonplaten
/
xls-r-300m-sv-phoneme

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
mozilla-foundation/common_voice_3_0
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
2
patrickvonplaten commited on
Commit
ffb6ea4
•
1 Parent(s): b5891db

up

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. all_results.json +14 -0
  2. checkpoint-4000/config.json +107 -0
  3. checkpoint-4000/optimizer.pt +3 -0
  4. checkpoint-4000/preprocessor_config.json +9 -0
  5. checkpoint-4000/pytorch_model.bin +3 -0
  6. checkpoint-4000/rng_state_0.pth +3 -0
  7. checkpoint-4000/rng_state_1.pth +3 -0
  8. checkpoint-4000/rng_state_2.pth +3 -0
  9. checkpoint-4000/rng_state_3.pth +3 -0
  10. checkpoint-4000/rng_state_4.pth +3 -0
  11. checkpoint-4000/rng_state_5.pth +3 -0
  12. checkpoint-4000/rng_state_6.pth +3 -0
  13. checkpoint-4000/rng_state_7.pth +3 -0
  14. checkpoint-4000/scaler.pt +3 -0
  15. checkpoint-4000/scheduler.pt +3 -0
  16. checkpoint-4000/trainer_state.json +328 -0
  17. checkpoint-4000/training_args.bin +3 -0
  18. checkpoint-4500/config.json +107 -0
  19. checkpoint-4500/optimizer.pt +3 -0
  20. checkpoint-4500/preprocessor_config.json +9 -0
  21. checkpoint-4500/pytorch_model.bin +3 -0
  22. checkpoint-4500/rng_state_0.pth +3 -0
  23. checkpoint-4500/rng_state_1.pth +3 -0
  24. checkpoint-4500/rng_state_2.pth +3 -0
  25. checkpoint-4500/rng_state_3.pth +3 -0
  26. checkpoint-4500/rng_state_4.pth +3 -0
  27. checkpoint-4500/rng_state_5.pth +3 -0
  28. checkpoint-4500/rng_state_6.pth +3 -0
  29. checkpoint-4500/rng_state_7.pth +3 -0
  30. checkpoint-4500/scaler.pt +3 -0
  31. checkpoint-4500/scheduler.pt +3 -0
  32. checkpoint-4500/trainer_state.json +367 -0
  33. checkpoint-4500/training_args.bin +3 -0
  34. checkpoint-5000/config.json +107 -0
  35. checkpoint-5000/optimizer.pt +3 -0
  36. checkpoint-5000/preprocessor_config.json +9 -0
  37. checkpoint-5000/pytorch_model.bin +3 -0
  38. checkpoint-5000/rng_state_0.pth +3 -0
  39. checkpoint-5000/rng_state_1.pth +3 -0
  40. checkpoint-5000/rng_state_2.pth +3 -0
  41. checkpoint-5000/rng_state_3.pth +3 -0
  42. checkpoint-5000/rng_state_4.pth +3 -0
  43. checkpoint-5000/rng_state_5.pth +3 -0
  44. checkpoint-5000/rng_state_6.pth +3 -0
  45. checkpoint-5000/rng_state_7.pth +3 -0
  46. checkpoint-5000/scaler.pt +3 -0
  47. checkpoint-5000/scheduler.pt +3 -0
  48. checkpoint-5000/trainer_state.json +406 -0
  49. checkpoint-5000/training_args.bin +3 -0
  50. config.json +107 -0
all_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 150.0,
3
+ "eval_loss": 0.4879043400287628,
4
+ "eval_runtime": 2.1613,
5
+ "eval_samples": 348,
6
+ "eval_samples_per_second": 161.012,
7
+ "eval_steps_per_second": 5.089,
8
+ "eval_wer": 0.16115374437681926,
9
+ "train_loss": 1.9894317994294344,
10
+ "train_runtime": 2069.4343,
11
+ "train_samples": 1150,
12
+ "train_samples_per_second": 83.356,
13
+ "train_steps_per_second": 2.609
14
+ }
checkpoint-4000/config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "./",
3
+ "activation_dropout": 0.1,
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": 42,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 768,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
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": 43,
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
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.0,
57
+ "hidden_size": 1024,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 4096,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.0,
62
+ "mask_feature_length": 64,
63
+ "mask_feature_min_masks": 0,
64
+ "mask_feature_prob": 0.25,
65
+ "mask_time_length": 10,
66
+ "mask_time_min_masks": 2,
67
+ "mask_time_prob": 0.75,
68
+ "model_type": "wav2vec2",
69
+ "num_adapter_layers": 3,
70
+ "num_attention_heads": 16,
71
+ "num_codevector_groups": 2,
72
+ "num_codevectors_per_group": 320,
73
+ "num_conv_pos_embedding_groups": 16,
74
+ "num_conv_pos_embeddings": 128,
75
+ "num_feat_extract_layers": 7,
76
+ "num_hidden_layers": 24,
77
+ "num_negatives": 100,
78
+ "output_hidden_size": 1024,
79
+ "pad_token_id": 41,
80
+ "proj_codevector_dim": 768,
81
+ "tdnn_dilation": [
82
+ 1,
83
+ 2,
84
+ 3,
85
+ 1,
86
+ 1
87
+ ],
88
+ "tdnn_dim": [
89
+ 512,
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 1500
94
+ ],
95
+ "tdnn_kernel": [
96
+ 5,
97
+ 3,
98
+ 3,
99
+ 1,
100
+ 1
101
+ ],
102
+ "torch_dtype": "float32",
103
+ "transformers_version": "4.15.0.dev0",
104
+ "use_weighted_layer_sum": false,
105
+ "vocab_size": 45,
106
+ "xvector_output_dim": 512
107
+ }
checkpoint-4000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:814463ec6828f20ef29c843136f0e3a850465e3f75025efda87b5183c6548488
3
+ size 2490427921
checkpoint-4000/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,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
checkpoint-4000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f314f001edb26513c74f89e5d6bb4bd4ca9c73826e79a3345f0bc4aa9b9b07ac
3
+ size 1262108145
checkpoint-4000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1367aebe9ecb8adafc4a4659468ef3ba0bcf50722cf266ca2a92ce5f03c34538
3
+ size 14567
checkpoint-4000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e8ea05e8985945e3e53e54abed1e790048d859b91830957c9fa5b04a557d6e8
3
+ size 14503
checkpoint-4000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:661854f01a4ee9d3af50fd74039159509a98c6a52cec04c032f10adda6240f2d
3
+ size 14567
checkpoint-4000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40da3e01b2738b76b68ba706a816f620c65936b70207738a4021a10c438a3abc
3
+ size 14567
checkpoint-4000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37ec9fcd16ec11ec80a7bcb9c44eee117d5cb55a66ea35593c091cd5967c1637
3
+ size 14503
checkpoint-4000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72cb23d5bfdb5f7ae2f148749f76dd88114e30de937b7612ff507faa0ab1b841
3
+ size 14503
checkpoint-4000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:569f197926a3b01457d517205dfdd03e6efb4dc9594fe5e5b7582eb4be0e6a5c
3
+ size 14503
checkpoint-4000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c535e383c0858212d99b3229de4ab437738af5bf836d4be1e02c7aa3ea618946
3
+ size 14503
checkpoint-4000/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d35b9eb5aeb68521669bf414d2eba186ec6873bfce68abf8ca262f523b0fecfc
3
+ size 559
checkpoint-4000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50b96a1fe2edfebdf673e78eef1c30984dab1f9e9b2694fb08e5a3835eec7a0f
3
+ size 623
checkpoint-4000/trainer_state.json ADDED
@@ -0,0 +1,328 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 111.11111111111111,
5
+ "global_step": 4000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 2.78,
12
+ "learning_rate": 7.2e-06,
13
+ "loss": 22.7336,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 5.56,
18
+ "learning_rate": 1.47e-05,
19
+ "loss": 8.3017,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 8.33,
24
+ "learning_rate": 2.2199999999999998e-05,
25
+ "loss": 5.4371,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 11.11,
30
+ "learning_rate": 2.97e-05,
31
+ "loss": 4.1583,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 13.89,
36
+ "learning_rate": 3.7199999999999996e-05,
37
+ "loss": 3.6082,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 13.89,
42
+ "eval_loss": 3.501119613647461,
43
+ "eval_runtime": 2.3778,
44
+ "eval_samples_per_second": 146.354,
45
+ "eval_steps_per_second": 4.626,
46
+ "eval_wer": 0.9825350621857635,
47
+ "step": 500
48
+ },
49
+ {
50
+ "epoch": 16.67,
51
+ "learning_rate": 4.4699999999999996e-05,
52
+ "loss": 3.4894,
53
+ "step": 600
54
+ },
55
+ {
56
+ "epoch": 19.44,
57
+ "learning_rate": 5.2199999999999995e-05,
58
+ "loss": 3.448,
59
+ "step": 700
60
+ },
61
+ {
62
+ "epoch": 22.22,
63
+ "learning_rate": 5.97e-05,
64
+ "loss": 3.4155,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 25.0,
69
+ "learning_rate": 6.72e-05,
70
+ "loss": 3.3702,
71
+ "step": 900
72
+ },
73
+ {
74
+ "epoch": 27.78,
75
+ "learning_rate": 7.47e-05,
76
+ "loss": 3.2566,
77
+ "step": 1000
78
+ },
79
+ {
80
+ "epoch": 27.78,
81
+ "eval_loss": 2.9146299362182617,
82
+ "eval_runtime": 2.224,
83
+ "eval_samples_per_second": 156.477,
84
+ "eval_steps_per_second": 4.946,
85
+ "eval_wer": 0.9367557554908706,
86
+ "step": 1000
87
+ },
88
+ {
89
+ "epoch": 30.56,
90
+ "learning_rate": 7.336363636363636e-05,
91
+ "loss": 2.832,
92
+ "step": 1100
93
+ },
94
+ {
95
+ "epoch": 33.33,
96
+ "learning_rate": 7.16590909090909e-05,
97
+ "loss": 2.2848,
98
+ "step": 1200
99
+ },
100
+ {
101
+ "epoch": 36.11,
102
+ "learning_rate": 6.995454545454545e-05,
103
+ "loss": 1.9336,
104
+ "step": 1300
105
+ },
106
+ {
107
+ "epoch": 38.89,
108
+ "learning_rate": 6.824999999999999e-05,
109
+ "loss": 1.7366,
110
+ "step": 1400
111
+ },
112
+ {
113
+ "epoch": 41.67,
114
+ "learning_rate": 6.654545454545455e-05,
115
+ "loss": 1.5744,
116
+ "step": 1500
117
+ },
118
+ {
119
+ "epoch": 41.67,
120
+ "eval_loss": 0.6705893278121948,
121
+ "eval_runtime": 2.2526,
122
+ "eval_samples_per_second": 154.49,
123
+ "eval_steps_per_second": 4.883,
124
+ "eval_wer": 0.20706536120666844,
125
+ "step": 1500
126
+ },
127
+ {
128
+ "epoch": 44.44,
129
+ "learning_rate": 6.484090909090909e-05,
130
+ "loss": 1.4626,
131
+ "step": 1600
132
+ },
133
+ {
134
+ "epoch": 47.22,
135
+ "learning_rate": 6.313636363636363e-05,
136
+ "loss": 1.4203,
137
+ "step": 1700
138
+ },
139
+ {
140
+ "epoch": 50.0,
141
+ "learning_rate": 6.143181818181818e-05,
142
+ "loss": 1.3699,
143
+ "step": 1800
144
+ },
145
+ {
146
+ "epoch": 52.78,
147
+ "learning_rate": 5.9727272727272724e-05,
148
+ "loss": 1.3123,
149
+ "step": 1900
150
+ },
151
+ {
152
+ "epoch": 55.56,
153
+ "learning_rate": 5.802272727272727e-05,
154
+ "loss": 1.2744,
155
+ "step": 2000
156
+ },
157
+ {
158
+ "epoch": 55.56,
159
+ "eval_loss": 0.540650486946106,
160
+ "eval_runtime": 2.2596,
161
+ "eval_samples_per_second": 154.01,
162
+ "eval_steps_per_second": 4.868,
163
+ "eval_wer": 0.1749139984122784,
164
+ "step": 2000
165
+ },
166
+ {
167
+ "epoch": 58.33,
168
+ "learning_rate": 5.631818181818181e-05,
169
+ "loss": 1.2163,
170
+ "step": 2100
171
+ },
172
+ {
173
+ "epoch": 61.11,
174
+ "learning_rate": 5.463068181818181e-05,
175
+ "loss": 1.187,
176
+ "step": 2200
177
+ },
178
+ {
179
+ "epoch": 63.89,
180
+ "learning_rate": 5.292613636363636e-05,
181
+ "loss": 1.1699,
182
+ "step": 2300
183
+ },
184
+ {
185
+ "epoch": 66.67,
186
+ "learning_rate": 5.122159090909091e-05,
187
+ "loss": 1.1208,
188
+ "step": 2400
189
+ },
190
+ {
191
+ "epoch": 69.44,
192
+ "learning_rate": 4.9517045454545456e-05,
193
+ "loss": 1.0734,
194
+ "step": 2500
195
+ },
196
+ {
197
+ "epoch": 69.44,
198
+ "eval_loss": 0.527560830116272,
199
+ "eval_runtime": 2.3157,
200
+ "eval_samples_per_second": 150.278,
201
+ "eval_steps_per_second": 4.75,
202
+ "eval_wer": 0.17610478962688542,
203
+ "step": 2500
204
+ },
205
+ {
206
+ "epoch": 72.22,
207
+ "learning_rate": 4.781249999999999e-05,
208
+ "loss": 1.0427,
209
+ "step": 2600
210
+ },
211
+ {
212
+ "epoch": 75.0,
213
+ "learning_rate": 4.610795454545454e-05,
214
+ "loss": 1.0227,
215
+ "step": 2700
216
+ },
217
+ {
218
+ "epoch": 77.78,
219
+ "learning_rate": 4.4420454545454544e-05,
220
+ "loss": 0.9966,
221
+ "step": 2800
222
+ },
223
+ {
224
+ "epoch": 80.56,
225
+ "learning_rate": 4.271590909090909e-05,
226
+ "loss": 0.9804,
227
+ "step": 2900
228
+ },
229
+ {
230
+ "epoch": 83.33,
231
+ "learning_rate": 4.101136363636364e-05,
232
+ "loss": 0.9451,
233
+ "step": 3000
234
+ },
235
+ {
236
+ "epoch": 83.33,
237
+ "eval_loss": 0.4845634400844574,
238
+ "eval_runtime": 2.2522,
239
+ "eval_samples_per_second": 154.518,
240
+ "eval_steps_per_second": 4.884,
241
+ "eval_wer": 0.17068007409367558,
242
+ "step": 3000
243
+ },
244
+ {
245
+ "epoch": 86.11,
246
+ "learning_rate": 3.9306818181818174e-05,
247
+ "loss": 0.9217,
248
+ "step": 3100
249
+ },
250
+ {
251
+ "epoch": 88.89,
252
+ "learning_rate": 3.760227272727272e-05,
253
+ "loss": 0.904,
254
+ "step": 3200
255
+ },
256
+ {
257
+ "epoch": 91.67,
258
+ "learning_rate": 3.591477272727273e-05,
259
+ "loss": 0.8857,
260
+ "step": 3300
261
+ },
262
+ {
263
+ "epoch": 94.44,
264
+ "learning_rate": 3.421022727272727e-05,
265
+ "loss": 0.88,
266
+ "step": 3400
267
+ },
268
+ {
269
+ "epoch": 97.22,
270
+ "learning_rate": 3.250568181818182e-05,
271
+ "loss": 0.8559,
272
+ "step": 3500
273
+ },
274
+ {
275
+ "epoch": 97.22,
276
+ "eval_loss": 0.494262158870697,
277
+ "eval_runtime": 2.0455,
278
+ "eval_samples_per_second": 170.13,
279
+ "eval_steps_per_second": 5.378,
280
+ "eval_wer": 0.1672400105848108,
281
+ "step": 3500
282
+ },
283
+ {
284
+ "epoch": 100.0,
285
+ "learning_rate": 3.0801136363636364e-05,
286
+ "loss": 0.819,
287
+ "step": 3600
288
+ },
289
+ {
290
+ "epoch": 102.78,
291
+ "learning_rate": 2.9096590909090905e-05,
292
+ "loss": 0.8242,
293
+ "step": 3700
294
+ },
295
+ {
296
+ "epoch": 105.56,
297
+ "learning_rate": 2.7392045454545453e-05,
298
+ "loss": 0.7983,
299
+ "step": 3800
300
+ },
301
+ {
302
+ "epoch": 108.33,
303
+ "learning_rate": 2.56875e-05,
304
+ "loss": 0.7911,
305
+ "step": 3900
306
+ },
307
+ {
308
+ "epoch": 111.11,
309
+ "learning_rate": 2.3982954545454542e-05,
310
+ "loss": 0.7725,
311
+ "step": 4000
312
+ },
313
+ {
314
+ "epoch": 111.11,
315
+ "eval_loss": 0.4919077455997467,
316
+ "eval_runtime": 2.1172,
317
+ "eval_samples_per_second": 164.365,
318
+ "eval_steps_per_second": 5.195,
319
+ "eval_wer": 0.1672400105848108,
320
+ "step": 4000
321
+ }
322
+ ],
323
+ "max_steps": 5400,
324
+ "num_train_epochs": 150,
325
+ "total_flos": 1.359349696556748e+19,
326
+ "trial_name": null,
327
+ "trial_params": null
328
+ }
checkpoint-4000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4481500ba24008d68d7f470ba6c1d6973e0fb92d1b02b0e9fa3a4fd2489c41d0
3
+ size 2927
checkpoint-4500/config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "./",
3
+ "activation_dropout": 0.1,
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": 42,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 768,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
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": 43,
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
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.0,
57
+ "hidden_size": 1024,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 4096,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.0,
62
+ "mask_feature_length": 64,
63
+ "mask_feature_min_masks": 0,
64
+ "mask_feature_prob": 0.25,
65
+ "mask_time_length": 10,
66
+ "mask_time_min_masks": 2,
67
+ "mask_time_prob": 0.75,
68
+ "model_type": "wav2vec2",
69
+ "num_adapter_layers": 3,
70
+ "num_attention_heads": 16,
71
+ "num_codevector_groups": 2,
72
+ "num_codevectors_per_group": 320,
73
+ "num_conv_pos_embedding_groups": 16,
74
+ "num_conv_pos_embeddings": 128,
75
+ "num_feat_extract_layers": 7,
76
+ "num_hidden_layers": 24,
77
+ "num_negatives": 100,
78
+ "output_hidden_size": 1024,
79
+ "pad_token_id": 41,
80
+ "proj_codevector_dim": 768,
81
+ "tdnn_dilation": [
82
+ 1,
83
+ 2,
84
+ 3,
85
+ 1,
86
+ 1
87
+ ],
88
+ "tdnn_dim": [
89
+ 512,
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 1500
94
+ ],
95
+ "tdnn_kernel": [
96
+ 5,
97
+ 3,
98
+ 3,
99
+ 1,
100
+ 1
101
+ ],
102
+ "torch_dtype": "float32",
103
+ "transformers_version": "4.15.0.dev0",
104
+ "use_weighted_layer_sum": false,
105
+ "vocab_size": 45,
106
+ "xvector_output_dim": 512
107
+ }
checkpoint-4500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:502260a39a07a3d3017dd875ba45ce94cf8e4c66877cb4789918095d07305dcf
3
+ size 2490427921
checkpoint-4500/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,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
checkpoint-4500/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f0946c648cb5f889ed93507f5d23ed9c769023300d95e69eb502fca6b410f8e
3
+ size 1262108145
checkpoint-4500/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d757537fc7c8864a582a9f6a377f343e401842190b0998b74554e77f7d5fc4ac
3
+ size 14503
checkpoint-4500/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c2a685299438493e5bd8f1651a6116a412089064ef6021a3ead10c08449639e
3
+ size 14567
checkpoint-4500/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ce7e241a96d84601502f8193306ba61e2fa861dcf3fc2d6c5d538017e28f9c2
3
+ size 14503
checkpoint-4500/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fdddd051e9bcbf1485f0081a44eabc2ad52be9c843ce027e925fbec9e797e89b
3
+ size 14567
checkpoint-4500/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3065335beb8d97915dc7576f14787f3b5783eef888dfbf16a4c72d42134094cf
3
+ size 14567
checkpoint-4500/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfc2ab2e4d543b1cde35ff4d2c624349ee8b118dfb0deb9f2f35bd31c757f898
3
+ size 14503
checkpoint-4500/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7ef101d29e12b2203287d943c064e762489ebd7233e5721994329f62dadf4ed
3
+ size 14567
checkpoint-4500/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb2a447b437db81b760f7a242bf1c3e50873506ff4296e3659f5480024daf1d5
3
+ size 14567
checkpoint-4500/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:202f5be028c59e51a03602ebf7c46edb21052f445964dc7c6d89ce4f4cf61320
3
+ size 559
checkpoint-4500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:117490191d77c46a0812610fe700ebe8ecb1676a3d376da1d8c1a0f010f192a4
3
+ size 623
checkpoint-4500/trainer_state.json ADDED
@@ -0,0 +1,367 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 125.0,
5
+ "global_step": 4500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 2.78,
12
+ "learning_rate": 7.2e-06,
13
+ "loss": 22.7336,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 5.56,
18
+ "learning_rate": 1.47e-05,
19
+ "loss": 8.3017,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 8.33,
24
+ "learning_rate": 2.2199999999999998e-05,
25
+ "loss": 5.4371,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 11.11,
30
+ "learning_rate": 2.97e-05,
31
+ "loss": 4.1583,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 13.89,
36
+ "learning_rate": 3.7199999999999996e-05,
37
+ "loss": 3.6082,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 13.89,
42
+ "eval_loss": 3.501119613647461,
43
+ "eval_runtime": 2.3778,
44
+ "eval_samples_per_second": 146.354,
45
+ "eval_steps_per_second": 4.626,
46
+ "eval_wer": 0.9825350621857635,
47
+ "step": 500
48
+ },
49
+ {
50
+ "epoch": 16.67,
51
+ "learning_rate": 4.4699999999999996e-05,
52
+ "loss": 3.4894,
53
+ "step": 600
54
+ },
55
+ {
56
+ "epoch": 19.44,
57
+ "learning_rate": 5.2199999999999995e-05,
58
+ "loss": 3.448,
59
+ "step": 700
60
+ },
61
+ {
62
+ "epoch": 22.22,
63
+ "learning_rate": 5.97e-05,
64
+ "loss": 3.4155,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 25.0,
69
+ "learning_rate": 6.72e-05,
70
+ "loss": 3.3702,
71
+ "step": 900
72
+ },
73
+ {
74
+ "epoch": 27.78,
75
+ "learning_rate": 7.47e-05,
76
+ "loss": 3.2566,
77
+ "step": 1000
78
+ },
79
+ {
80
+ "epoch": 27.78,
81
+ "eval_loss": 2.9146299362182617,
82
+ "eval_runtime": 2.224,
83
+ "eval_samples_per_second": 156.477,
84
+ "eval_steps_per_second": 4.946,
85
+ "eval_wer": 0.9367557554908706,
86
+ "step": 1000
87
+ },
88
+ {
89
+ "epoch": 30.56,
90
+ "learning_rate": 7.336363636363636e-05,
91
+ "loss": 2.832,
92
+ "step": 1100
93
+ },
94
+ {
95
+ "epoch": 33.33,
96
+ "learning_rate": 7.16590909090909e-05,
97
+ "loss": 2.2848,
98
+ "step": 1200
99
+ },
100
+ {
101
+ "epoch": 36.11,
102
+ "learning_rate": 6.995454545454545e-05,
103
+ "loss": 1.9336,
104
+ "step": 1300
105
+ },
106
+ {
107
+ "epoch": 38.89,
108
+ "learning_rate": 6.824999999999999e-05,
109
+ "loss": 1.7366,
110
+ "step": 1400
111
+ },
112
+ {
113
+ "epoch": 41.67,
114
+ "learning_rate": 6.654545454545455e-05,
115
+ "loss": 1.5744,
116
+ "step": 1500
117
+ },
118
+ {
119
+ "epoch": 41.67,
120
+ "eval_loss": 0.6705893278121948,
121
+ "eval_runtime": 2.2526,
122
+ "eval_samples_per_second": 154.49,
123
+ "eval_steps_per_second": 4.883,
124
+ "eval_wer": 0.20706536120666844,
125
+ "step": 1500
126
+ },
127
+ {
128
+ "epoch": 44.44,
129
+ "learning_rate": 6.484090909090909e-05,
130
+ "loss": 1.4626,
131
+ "step": 1600
132
+ },
133
+ {
134
+ "epoch": 47.22,
135
+ "learning_rate": 6.313636363636363e-05,
136
+ "loss": 1.4203,
137
+ "step": 1700
138
+ },
139
+ {
140
+ "epoch": 50.0,
141
+ "learning_rate": 6.143181818181818e-05,
142
+ "loss": 1.3699,
143
+ "step": 1800
144
+ },
145
+ {
146
+ "epoch": 52.78,
147
+ "learning_rate": 5.9727272727272724e-05,
148
+ "loss": 1.3123,
149
+ "step": 1900
150
+ },
151
+ {
152
+ "epoch": 55.56,
153
+ "learning_rate": 5.802272727272727e-05,
154
+ "loss": 1.2744,
155
+ "step": 2000
156
+ },
157
+ {
158
+ "epoch": 55.56,
159
+ "eval_loss": 0.540650486946106,
160
+ "eval_runtime": 2.2596,
161
+ "eval_samples_per_second": 154.01,
162
+ "eval_steps_per_second": 4.868,
163
+ "eval_wer": 0.1749139984122784,
164
+ "step": 2000
165
+ },
166
+ {
167
+ "epoch": 58.33,
168
+ "learning_rate": 5.631818181818181e-05,
169
+ "loss": 1.2163,
170
+ "step": 2100
171
+ },
172
+ {
173
+ "epoch": 61.11,
174
+ "learning_rate": 5.463068181818181e-05,
175
+ "loss": 1.187,
176
+ "step": 2200
177
+ },
178
+ {
179
+ "epoch": 63.89,
180
+ "learning_rate": 5.292613636363636e-05,
181
+ "loss": 1.1699,
182
+ "step": 2300
183
+ },
184
+ {
185
+ "epoch": 66.67,
186
+ "learning_rate": 5.122159090909091e-05,
187
+ "loss": 1.1208,
188
+ "step": 2400
189
+ },
190
+ {
191
+ "epoch": 69.44,
192
+ "learning_rate": 4.9517045454545456e-05,
193
+ "loss": 1.0734,
194
+ "step": 2500
195
+ },
196
+ {
197
+ "epoch": 69.44,
198
+ "eval_loss": 0.527560830116272,
199
+ "eval_runtime": 2.3157,
200
+ "eval_samples_per_second": 150.278,
201
+ "eval_steps_per_second": 4.75,
202
+ "eval_wer": 0.17610478962688542,
203
+ "step": 2500
204
+ },
205
+ {
206
+ "epoch": 72.22,
207
+ "learning_rate": 4.781249999999999e-05,
208
+ "loss": 1.0427,
209
+ "step": 2600
210
+ },
211
+ {
212
+ "epoch": 75.0,
213
+ "learning_rate": 4.610795454545454e-05,
214
+ "loss": 1.0227,
215
+ "step": 2700
216
+ },
217
+ {
218
+ "epoch": 77.78,
219
+ "learning_rate": 4.4420454545454544e-05,
220
+ "loss": 0.9966,
221
+ "step": 2800
222
+ },
223
+ {
224
+ "epoch": 80.56,
225
+ "learning_rate": 4.271590909090909e-05,
226
+ "loss": 0.9804,
227
+ "step": 2900
228
+ },
229
+ {
230
+ "epoch": 83.33,
231
+ "learning_rate": 4.101136363636364e-05,
232
+ "loss": 0.9451,
233
+ "step": 3000
234
+ },
235
+ {
236
+ "epoch": 83.33,
237
+ "eval_loss": 0.4845634400844574,
238
+ "eval_runtime": 2.2522,
239
+ "eval_samples_per_second": 154.518,
240
+ "eval_steps_per_second": 4.884,
241
+ "eval_wer": 0.17068007409367558,
242
+ "step": 3000
243
+ },
244
+ {
245
+ "epoch": 86.11,
246
+ "learning_rate": 3.9306818181818174e-05,
247
+ "loss": 0.9217,
248
+ "step": 3100
249
+ },
250
+ {
251
+ "epoch": 88.89,
252
+ "learning_rate": 3.760227272727272e-05,
253
+ "loss": 0.904,
254
+ "step": 3200
255
+ },
256
+ {
257
+ "epoch": 91.67,
258
+ "learning_rate": 3.591477272727273e-05,
259
+ "loss": 0.8857,
260
+ "step": 3300
261
+ },
262
+ {
263
+ "epoch": 94.44,
264
+ "learning_rate": 3.421022727272727e-05,
265
+ "loss": 0.88,
266
+ "step": 3400
267
+ },
268
+ {
269
+ "epoch": 97.22,
270
+ "learning_rate": 3.250568181818182e-05,
271
+ "loss": 0.8559,
272
+ "step": 3500
273
+ },
274
+ {
275
+ "epoch": 97.22,
276
+ "eval_loss": 0.494262158870697,
277
+ "eval_runtime": 2.0455,
278
+ "eval_samples_per_second": 170.13,
279
+ "eval_steps_per_second": 5.378,
280
+ "eval_wer": 0.1672400105848108,
281
+ "step": 3500
282
+ },
283
+ {
284
+ "epoch": 100.0,
285
+ "learning_rate": 3.0801136363636364e-05,
286
+ "loss": 0.819,
287
+ "step": 3600
288
+ },
289
+ {
290
+ "epoch": 102.78,
291
+ "learning_rate": 2.9096590909090905e-05,
292
+ "loss": 0.8242,
293
+ "step": 3700
294
+ },
295
+ {
296
+ "epoch": 105.56,
297
+ "learning_rate": 2.7392045454545453e-05,
298
+ "loss": 0.7983,
299
+ "step": 3800
300
+ },
301
+ {
302
+ "epoch": 108.33,
303
+ "learning_rate": 2.56875e-05,
304
+ "loss": 0.7911,
305
+ "step": 3900
306
+ },
307
+ {
308
+ "epoch": 111.11,
309
+ "learning_rate": 2.3982954545454542e-05,
310
+ "loss": 0.7725,
311
+ "step": 4000
312
+ },
313
+ {
314
+ "epoch": 111.11,
315
+ "eval_loss": 0.4919077455997467,
316
+ "eval_runtime": 2.1172,
317
+ "eval_samples_per_second": 164.365,
318
+ "eval_steps_per_second": 5.195,
319
+ "eval_wer": 0.1672400105848108,
320
+ "step": 4000
321
+ },
322
+ {
323
+ "epoch": 113.89,
324
+ "learning_rate": 2.227840909090909e-05,
325
+ "loss": 0.7691,
326
+ "step": 4100
327
+ },
328
+ {
329
+ "epoch": 116.67,
330
+ "learning_rate": 2.0573863636363634e-05,
331
+ "loss": 0.739,
332
+ "step": 4200
333
+ },
334
+ {
335
+ "epoch": 119.44,
336
+ "learning_rate": 1.886931818181818e-05,
337
+ "loss": 0.7452,
338
+ "step": 4300
339
+ },
340
+ {
341
+ "epoch": 122.22,
342
+ "learning_rate": 1.7164772727272726e-05,
343
+ "loss": 0.7218,
344
+ "step": 4400
345
+ },
346
+ {
347
+ "epoch": 125.0,
348
+ "learning_rate": 1.546022727272727e-05,
349
+ "loss": 0.7171,
350
+ "step": 4500
351
+ },
352
+ {
353
+ "epoch": 125.0,
354
+ "eval_loss": 0.5054941773414612,
355
+ "eval_runtime": 2.2494,
356
+ "eval_samples_per_second": 154.711,
357
+ "eval_steps_per_second": 4.89,
358
+ "eval_wer": 0.16115374437681926,
359
+ "step": 4500
360
+ }
361
+ ],
362
+ "max_steps": 5400,
363
+ "num_train_epochs": 150,
364
+ "total_flos": 1.5292173814083879e+19,
365
+ "trial_name": null,
366
+ "trial_params": null
367
+ }
checkpoint-4500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4481500ba24008d68d7f470ba6c1d6973e0fb92d1b02b0e9fa3a4fd2489c41d0
3
+ size 2927
checkpoint-5000/config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "./",
3
+ "activation_dropout": 0.1,
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": 42,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 768,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
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": 43,
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
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.0,
57
+ "hidden_size": 1024,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 4096,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.0,
62
+ "mask_feature_length": 64,
63
+ "mask_feature_min_masks": 0,
64
+ "mask_feature_prob": 0.25,
65
+ "mask_time_length": 10,
66
+ "mask_time_min_masks": 2,
67
+ "mask_time_prob": 0.75,
68
+ "model_type": "wav2vec2",
69
+ "num_adapter_layers": 3,
70
+ "num_attention_heads": 16,
71
+ "num_codevector_groups": 2,
72
+ "num_codevectors_per_group": 320,
73
+ "num_conv_pos_embedding_groups": 16,
74
+ "num_conv_pos_embeddings": 128,
75
+ "num_feat_extract_layers": 7,
76
+ "num_hidden_layers": 24,
77
+ "num_negatives": 100,
78
+ "output_hidden_size": 1024,
79
+ "pad_token_id": 41,
80
+ "proj_codevector_dim": 768,
81
+ "tdnn_dilation": [
82
+ 1,
83
+ 2,
84
+ 3,
85
+ 1,
86
+ 1
87
+ ],
88
+ "tdnn_dim": [
89
+ 512,
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 1500
94
+ ],
95
+ "tdnn_kernel": [
96
+ 5,
97
+ 3,
98
+ 3,
99
+ 1,
100
+ 1
101
+ ],
102
+ "torch_dtype": "float32",
103
+ "transformers_version": "4.15.0.dev0",
104
+ "use_weighted_layer_sum": false,
105
+ "vocab_size": 45,
106
+ "xvector_output_dim": 512
107
+ }
checkpoint-5000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60003d29a8d90a0c1902fe096002b5d4269a2651d2dd04202f0fdb1130a79c04
3
+ size 2490427921
checkpoint-5000/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,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
checkpoint-5000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:acdf48af6d2a322351befc7cdd675231549e4b202dd0408ddebf1e57224b37b3
3
+ size 1262108145
checkpoint-5000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:173d7f49613dfa5d261da368ae50e065fe341c2dd38800e101f2a0cc3918bc01
3
+ size 14567
checkpoint-5000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:83bb5b11149882ac18f318c13966d4f30539fe29bab2efabd1da1d00818e262a
3
+ size 14503
checkpoint-5000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8758ab8923a28eb9ca9de158bf9044841096ed853ef032e7d7c9808d1bbdb60d
3
+ size 14503
checkpoint-5000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e576a97e7f6721ffacc4adccc849e840a81007881f207c84db69e49b3cd5d02e
3
+ size 14503
checkpoint-5000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4aa1bd576fd7ce768dc78f4426efd9bb2b0916a1ccd5dbfa12bbc68b7a0ead06
3
+ size 14567
checkpoint-5000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d4abfbe5eff201064163933fbb50be02d2411d76f9958a4ac2503a726e7fd76
3
+ size 14567
checkpoint-5000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a47cd95fa3b9540c512aa6eddd3b6b996019c38ae454db073f0b28d8e8f9e42
3
+ size 14567
checkpoint-5000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:401f29c8fbd5e679dc661f0d90971cb0b68beb67a52e09eaf574e3c837ea3551
3
+ size 14503
checkpoint-5000/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fdc0a5e44b0ff5d9bf798bbf6fb99f8e67698c58083668fef838cbc2313a5bac
3
+ size 559
checkpoint-5000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4b90497758e9b369af350ce2158433c5e77e14ffce1a17a16b76230e1126fc0
3
+ size 623
checkpoint-5000/trainer_state.json ADDED
@@ -0,0 +1,406 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 138.88888888888889,
5
+ "global_step": 5000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 2.78,
12
+ "learning_rate": 7.2e-06,
13
+ "loss": 22.7336,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 5.56,
18
+ "learning_rate": 1.47e-05,
19
+ "loss": 8.3017,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 8.33,
24
+ "learning_rate": 2.2199999999999998e-05,
25
+ "loss": 5.4371,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 11.11,
30
+ "learning_rate": 2.97e-05,
31
+ "loss": 4.1583,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 13.89,
36
+ "learning_rate": 3.7199999999999996e-05,
37
+ "loss": 3.6082,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 13.89,
42
+ "eval_loss": 3.501119613647461,
43
+ "eval_runtime": 2.3778,
44
+ "eval_samples_per_second": 146.354,
45
+ "eval_steps_per_second": 4.626,
46
+ "eval_wer": 0.9825350621857635,
47
+ "step": 500
48
+ },
49
+ {
50
+ "epoch": 16.67,
51
+ "learning_rate": 4.4699999999999996e-05,
52
+ "loss": 3.4894,
53
+ "step": 600
54
+ },
55
+ {
56
+ "epoch": 19.44,
57
+ "learning_rate": 5.2199999999999995e-05,
58
+ "loss": 3.448,
59
+ "step": 700
60
+ },
61
+ {
62
+ "epoch": 22.22,
63
+ "learning_rate": 5.97e-05,
64
+ "loss": 3.4155,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 25.0,
69
+ "learning_rate": 6.72e-05,
70
+ "loss": 3.3702,
71
+ "step": 900
72
+ },
73
+ {
74
+ "epoch": 27.78,
75
+ "learning_rate": 7.47e-05,
76
+ "loss": 3.2566,
77
+ "step": 1000
78
+ },
79
+ {
80
+ "epoch": 27.78,
81
+ "eval_loss": 2.9146299362182617,
82
+ "eval_runtime": 2.224,
83
+ "eval_samples_per_second": 156.477,
84
+ "eval_steps_per_second": 4.946,
85
+ "eval_wer": 0.9367557554908706,
86
+ "step": 1000
87
+ },
88
+ {
89
+ "epoch": 30.56,
90
+ "learning_rate": 7.336363636363636e-05,
91
+ "loss": 2.832,
92
+ "step": 1100
93
+ },
94
+ {
95
+ "epoch": 33.33,
96
+ "learning_rate": 7.16590909090909e-05,
97
+ "loss": 2.2848,
98
+ "step": 1200
99
+ },
100
+ {
101
+ "epoch": 36.11,
102
+ "learning_rate": 6.995454545454545e-05,
103
+ "loss": 1.9336,
104
+ "step": 1300
105
+ },
106
+ {
107
+ "epoch": 38.89,
108
+ "learning_rate": 6.824999999999999e-05,
109
+ "loss": 1.7366,
110
+ "step": 1400
111
+ },
112
+ {
113
+ "epoch": 41.67,
114
+ "learning_rate": 6.654545454545455e-05,
115
+ "loss": 1.5744,
116
+ "step": 1500
117
+ },
118
+ {
119
+ "epoch": 41.67,
120
+ "eval_loss": 0.6705893278121948,
121
+ "eval_runtime": 2.2526,
122
+ "eval_samples_per_second": 154.49,
123
+ "eval_steps_per_second": 4.883,
124
+ "eval_wer": 0.20706536120666844,
125
+ "step": 1500
126
+ },
127
+ {
128
+ "epoch": 44.44,
129
+ "learning_rate": 6.484090909090909e-05,
130
+ "loss": 1.4626,
131
+ "step": 1600
132
+ },
133
+ {
134
+ "epoch": 47.22,
135
+ "learning_rate": 6.313636363636363e-05,
136
+ "loss": 1.4203,
137
+ "step": 1700
138
+ },
139
+ {
140
+ "epoch": 50.0,
141
+ "learning_rate": 6.143181818181818e-05,
142
+ "loss": 1.3699,
143
+ "step": 1800
144
+ },
145
+ {
146
+ "epoch": 52.78,
147
+ "learning_rate": 5.9727272727272724e-05,
148
+ "loss": 1.3123,
149
+ "step": 1900
150
+ },
151
+ {
152
+ "epoch": 55.56,
153
+ "learning_rate": 5.802272727272727e-05,
154
+ "loss": 1.2744,
155
+ "step": 2000
156
+ },
157
+ {
158
+ "epoch": 55.56,
159
+ "eval_loss": 0.540650486946106,
160
+ "eval_runtime": 2.2596,
161
+ "eval_samples_per_second": 154.01,
162
+ "eval_steps_per_second": 4.868,
163
+ "eval_wer": 0.1749139984122784,
164
+ "step": 2000
165
+ },
166
+ {
167
+ "epoch": 58.33,
168
+ "learning_rate": 5.631818181818181e-05,
169
+ "loss": 1.2163,
170
+ "step": 2100
171
+ },
172
+ {
173
+ "epoch": 61.11,
174
+ "learning_rate": 5.463068181818181e-05,
175
+ "loss": 1.187,
176
+ "step": 2200
177
+ },
178
+ {
179
+ "epoch": 63.89,
180
+ "learning_rate": 5.292613636363636e-05,
181
+ "loss": 1.1699,
182
+ "step": 2300
183
+ },
184
+ {
185
+ "epoch": 66.67,
186
+ "learning_rate": 5.122159090909091e-05,
187
+ "loss": 1.1208,
188
+ "step": 2400
189
+ },
190
+ {
191
+ "epoch": 69.44,
192
+ "learning_rate": 4.9517045454545456e-05,
193
+ "loss": 1.0734,
194
+ "step": 2500
195
+ },
196
+ {
197
+ "epoch": 69.44,
198
+ "eval_loss": 0.527560830116272,
199
+ "eval_runtime": 2.3157,
200
+ "eval_samples_per_second": 150.278,
201
+ "eval_steps_per_second": 4.75,
202
+ "eval_wer": 0.17610478962688542,
203
+ "step": 2500
204
+ },
205
+ {
206
+ "epoch": 72.22,
207
+ "learning_rate": 4.781249999999999e-05,
208
+ "loss": 1.0427,
209
+ "step": 2600
210
+ },
211
+ {
212
+ "epoch": 75.0,
213
+ "learning_rate": 4.610795454545454e-05,
214
+ "loss": 1.0227,
215
+ "step": 2700
216
+ },
217
+ {
218
+ "epoch": 77.78,
219
+ "learning_rate": 4.4420454545454544e-05,
220
+ "loss": 0.9966,
221
+ "step": 2800
222
+ },
223
+ {
224
+ "epoch": 80.56,
225
+ "learning_rate": 4.271590909090909e-05,
226
+ "loss": 0.9804,
227
+ "step": 2900
228
+ },
229
+ {
230
+ "epoch": 83.33,
231
+ "learning_rate": 4.101136363636364e-05,
232
+ "loss": 0.9451,
233
+ "step": 3000
234
+ },
235
+ {
236
+ "epoch": 83.33,
237
+ "eval_loss": 0.4845634400844574,
238
+ "eval_runtime": 2.2522,
239
+ "eval_samples_per_second": 154.518,
240
+ "eval_steps_per_second": 4.884,
241
+ "eval_wer": 0.17068007409367558,
242
+ "step": 3000
243
+ },
244
+ {
245
+ "epoch": 86.11,
246
+ "learning_rate": 3.9306818181818174e-05,
247
+ "loss": 0.9217,
248
+ "step": 3100
249
+ },
250
+ {
251
+ "epoch": 88.89,
252
+ "learning_rate": 3.760227272727272e-05,
253
+ "loss": 0.904,
254
+ "step": 3200
255
+ },
256
+ {
257
+ "epoch": 91.67,
258
+ "learning_rate": 3.591477272727273e-05,
259
+ "loss": 0.8857,
260
+ "step": 3300
261
+ },
262
+ {
263
+ "epoch": 94.44,
264
+ "learning_rate": 3.421022727272727e-05,
265
+ "loss": 0.88,
266
+ "step": 3400
267
+ },
268
+ {
269
+ "epoch": 97.22,
270
+ "learning_rate": 3.250568181818182e-05,
271
+ "loss": 0.8559,
272
+ "step": 3500
273
+ },
274
+ {
275
+ "epoch": 97.22,
276
+ "eval_loss": 0.494262158870697,
277
+ "eval_runtime": 2.0455,
278
+ "eval_samples_per_second": 170.13,
279
+ "eval_steps_per_second": 5.378,
280
+ "eval_wer": 0.1672400105848108,
281
+ "step": 3500
282
+ },
283
+ {
284
+ "epoch": 100.0,
285
+ "learning_rate": 3.0801136363636364e-05,
286
+ "loss": 0.819,
287
+ "step": 3600
288
+ },
289
+ {
290
+ "epoch": 102.78,
291
+ "learning_rate": 2.9096590909090905e-05,
292
+ "loss": 0.8242,
293
+ "step": 3700
294
+ },
295
+ {
296
+ "epoch": 105.56,
297
+ "learning_rate": 2.7392045454545453e-05,
298
+ "loss": 0.7983,
299
+ "step": 3800
300
+ },
301
+ {
302
+ "epoch": 108.33,
303
+ "learning_rate": 2.56875e-05,
304
+ "loss": 0.7911,
305
+ "step": 3900
306
+ },
307
+ {
308
+ "epoch": 111.11,
309
+ "learning_rate": 2.3982954545454542e-05,
310
+ "loss": 0.7725,
311
+ "step": 4000
312
+ },
313
+ {
314
+ "epoch": 111.11,
315
+ "eval_loss": 0.4919077455997467,
316
+ "eval_runtime": 2.1172,
317
+ "eval_samples_per_second": 164.365,
318
+ "eval_steps_per_second": 5.195,
319
+ "eval_wer": 0.1672400105848108,
320
+ "step": 4000
321
+ },
322
+ {
323
+ "epoch": 113.89,
324
+ "learning_rate": 2.227840909090909e-05,
325
+ "loss": 0.7691,
326
+ "step": 4100
327
+ },
328
+ {
329
+ "epoch": 116.67,
330
+ "learning_rate": 2.0573863636363634e-05,
331
+ "loss": 0.739,
332
+ "step": 4200
333
+ },
334
+ {
335
+ "epoch": 119.44,
336
+ "learning_rate": 1.886931818181818e-05,
337
+ "loss": 0.7452,
338
+ "step": 4300
339
+ },
340
+ {
341
+ "epoch": 122.22,
342
+ "learning_rate": 1.7164772727272726e-05,
343
+ "loss": 0.7218,
344
+ "step": 4400
345
+ },
346
+ {
347
+ "epoch": 125.0,
348
+ "learning_rate": 1.546022727272727e-05,
349
+ "loss": 0.7171,
350
+ "step": 4500
351
+ },
352
+ {
353
+ "epoch": 125.0,
354
+ "eval_loss": 0.5054941773414612,
355
+ "eval_runtime": 2.2494,
356
+ "eval_samples_per_second": 154.711,
357
+ "eval_steps_per_second": 4.89,
358
+ "eval_wer": 0.16115374437681926,
359
+ "step": 4500
360
+ },
361
+ {
362
+ "epoch": 127.78,
363
+ "learning_rate": 1.3755681818181817e-05,
364
+ "loss": 0.7135,
365
+ "step": 4600
366
+ },
367
+ {
368
+ "epoch": 130.56,
369
+ "learning_rate": 1.2051136363636363e-05,
370
+ "loss": 0.6951,
371
+ "step": 4700
372
+ },
373
+ {
374
+ "epoch": 133.33,
375
+ "learning_rate": 1.0346590909090907e-05,
376
+ "loss": 0.698,
377
+ "step": 4800
378
+ },
379
+ {
380
+ "epoch": 136.11,
381
+ "learning_rate": 8.642045454545455e-06,
382
+ "loss": 0.6711,
383
+ "step": 4900
384
+ },
385
+ {
386
+ "epoch": 138.89,
387
+ "learning_rate": 6.937499999999999e-06,
388
+ "loss": 0.6749,
389
+ "step": 5000
390
+ },
391
+ {
392
+ "epoch": 138.89,
393
+ "eval_loss": 0.48401492834091187,
394
+ "eval_runtime": 2.2496,
395
+ "eval_samples_per_second": 154.697,
396
+ "eval_steps_per_second": 4.89,
397
+ "eval_wer": 0.16260915586133898,
398
+ "step": 5000
399
+ }
400
+ ],
401
+ "max_steps": 5400,
402
+ "num_train_epochs": 150,
403
+ "total_flos": 1.6991775248800023e+19,
404
+ "trial_name": null,
405
+ "trial_params": null
406
+ }
checkpoint-5000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4481500ba24008d68d7f470ba6c1d6973e0fb92d1b02b0e9fa3a4fd2489c41d0
3
+ size 2927
config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "./",
3
+ "activation_dropout": 0.1,
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": 42,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 768,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
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": 43,
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
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.0,
57
+ "hidden_size": 1024,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 4096,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.0,
62
+ "mask_feature_length": 64,
63
+ "mask_feature_min_masks": 0,
64
+ "mask_feature_prob": 0.25,
65
+ "mask_time_length": 10,
66
+ "mask_time_min_masks": 2,
67
+ "mask_time_prob": 0.75,
68
+ "model_type": "wav2vec2",
69
+ "num_adapter_layers": 3,
70
+ "num_attention_heads": 16,
71
+ "num_codevector_groups": 2,
72
+ "num_codevectors_per_group": 320,
73
+ "num_conv_pos_embedding_groups": 16,
74
+ "num_conv_pos_embeddings": 128,
75
+ "num_feat_extract_layers": 7,
76
+ "num_hidden_layers": 24,
77
+ "num_negatives": 100,
78
+ "output_hidden_size": 1024,
79
+ "pad_token_id": 41,
80
+ "proj_codevector_dim": 768,
81
+ "tdnn_dilation": [
82
+ 1,
83
+ 2,
84
+ 3,
85
+ 1,
86
+ 1
87
+ ],
88
+ "tdnn_dim": [
89
+ 512,
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 1500
94
+ ],
95
+ "tdnn_kernel": [
96
+ 5,
97
+ 3,
98
+ 3,
99
+ 1,
100
+ 1
101
+ ],
102
+ "torch_dtype": "float32",
103
+ "transformers_version": "4.15.0.dev0",
104
+ "use_weighted_layer_sum": false,
105
+ "vocab_size": 45,
106
+ "xvector_output_dim": 512
107
+ }