comodoro commited on
Commit
52d319d
1 Parent(s): f45d287

Initial CV 8.0 model

Browse files
config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-xls-r-300m",
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": 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": 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_feature_length": 10,
64
+ "mask_feature_min_masks": 0,
65
+ "mask_feature_prob": 0.0,
66
+ "mask_time_length": 10,
67
+ "mask_time_min_masks": 2,
68
+ "mask_time_prob": 0.05,
69
+ "model_type": "wav2vec2",
70
+ "num_adapter_layers": 3,
71
+ "num_attention_heads": 16,
72
+ "num_codevector_groups": 2,
73
+ "num_codevectors_per_group": 320,
74
+ "num_conv_pos_embedding_groups": 16,
75
+ "num_conv_pos_embeddings": 128,
76
+ "num_feat_extract_layers": 7,
77
+ "num_hidden_layers": 24,
78
+ "num_negatives": 100,
79
+ "output_hidden_size": 1024,
80
+ "pad_token_id": 43,
81
+ "proj_codevector_dim": 768,
82
+ "tdnn_dilation": [
83
+ 1,
84
+ 2,
85
+ 3,
86
+ 1,
87
+ 1
88
+ ],
89
+ "tdnn_dim": [
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 512,
94
+ 1500
95
+ ],
96
+ "tdnn_kernel": [
97
+ 5,
98
+ 3,
99
+ 3,
100
+ 1,
101
+ 1
102
+ ],
103
+ "torch_dtype": "float32",
104
+ "transformers_version": "4.16.0.dev0",
105
+ "use_weighted_layer_sum": false,
106
+ "vocab_size": 46,
107
+ "xvector_output_dim": 512
108
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9a9d761e6f780e99089191e7ffa24e2a89ad23a8410feb77c290632bc296de4
3
+ size 2490436113
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": 0.0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bac8b9bb8d3e9f421a115fc0eb137f2bbd90c1eeee93d9a2de231922740eb66d
3
+ size 1262112241
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:490ca80b725ae7cc84f79c45a41ad8d08cf8b31084d829c0315d48f6217697f4
3
+ size 14567
scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27b32535e1492f5590f3993b428ea4ab0f5e62a1483a35f70f699d94966b2d49
3
+ size 559
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a12aff91ce999c303898b72fd3524f256ca19e4a494318ba7aec968280f26bd
3
+ size 623
trainer_state.json ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 137.09493670886076,
5
+ "global_step": 4250,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 8.06,
12
+ "learning_rate": 3.5e-05,
13
+ "loss": 7.2926,
14
+ "step": 250
15
+ },
16
+ {
17
+ "epoch": 8.06,
18
+ "eval_cer": 1.0,
19
+ "eval_loss": 3.8496713638305664,
20
+ "eval_runtime": 271.7106,
21
+ "eval_samples_per_second": 26.745,
22
+ "eval_steps_per_second": 3.345,
23
+ "eval_wer": 1.0,
24
+ "step": 250
25
+ },
26
+ {
27
+ "epoch": 16.13,
28
+ "learning_rate": 7e-05,
29
+ "loss": 3.417,
30
+ "step": 500
31
+ },
32
+ {
33
+ "epoch": 16.13,
34
+ "eval_cer": 0.9856759584948347,
35
+ "eval_loss": 3.285226583480835,
36
+ "eval_runtime": 275.8724,
37
+ "eval_samples_per_second": 26.342,
38
+ "eval_steps_per_second": 3.295,
39
+ "eval_wer": 1.0,
40
+ "step": 500
41
+ },
42
+ {
43
+ "epoch": 24.19,
44
+ "learning_rate": 6.578313253012048e-05,
45
+ "loss": 2.0264,
46
+ "step": 750
47
+ },
48
+ {
49
+ "epoch": 24.19,
50
+ "eval_cer": 0.176804441300443,
51
+ "eval_loss": 0.7098603248596191,
52
+ "eval_runtime": 269.2535,
53
+ "eval_samples_per_second": 26.989,
54
+ "eval_steps_per_second": 3.376,
55
+ "eval_wer": 0.7342138090806487,
56
+ "step": 750
57
+ },
58
+ {
59
+ "epoch": 32.25,
60
+ "learning_rate": 6.156626506024095e-05,
61
+ "loss": 0.4018,
62
+ "step": 1000
63
+ },
64
+ {
65
+ "epoch": 32.25,
66
+ "eval_cer": 0.15511300475620768,
67
+ "eval_loss": 0.6187673211097717,
68
+ "eval_runtime": 271.6934,
69
+ "eval_samples_per_second": 26.747,
70
+ "eval_steps_per_second": 3.346,
71
+ "eval_wer": 0.6415460467694989,
72
+ "step": 1000
73
+ },
74
+ {
75
+ "epoch": 40.32,
76
+ "learning_rate": 5.734939759036144e-05,
77
+ "loss": 0.2444,
78
+ "step": 1250
79
+ },
80
+ {
81
+ "epoch": 40.32,
82
+ "eval_cer": 0.15995397008055237,
83
+ "eval_loss": 0.6631603837013245,
84
+ "eval_runtime": 276.2198,
85
+ "eval_samples_per_second": 26.309,
86
+ "eval_steps_per_second": 3.291,
87
+ "eval_wer": 0.6361800289091737,
88
+ "step": 1250
89
+ },
90
+ {
91
+ "epoch": 48.38,
92
+ "learning_rate": 5.313253012048192e-05,
93
+ "loss": 0.1882,
94
+ "step": 1500
95
+ },
96
+ {
97
+ "epoch": 48.38,
98
+ "eval_cer": 0.13876129966064343,
99
+ "eval_loss": 0.6070172190666199,
100
+ "eval_runtime": 272.4136,
101
+ "eval_samples_per_second": 26.676,
102
+ "eval_steps_per_second": 3.337,
103
+ "eval_wer": 0.578262677464705,
104
+ "step": 1500
105
+ },
106
+ {
107
+ "epoch": 56.44,
108
+ "learning_rate": 4.891566265060241e-05,
109
+ "loss": 0.153,
110
+ "step": 1750
111
+ },
112
+ {
113
+ "epoch": 56.44,
114
+ "eval_cer": 0.13767574986063888,
115
+ "eval_loss": 0.6425250172615051,
116
+ "eval_runtime": 272.6101,
117
+ "eval_samples_per_second": 26.657,
118
+ "eval_steps_per_second": 3.334,
119
+ "eval_wer": 0.5720056234283112,
120
+ "step": 1750
121
+ },
122
+ {
123
+ "epoch": 64.51,
124
+ "learning_rate": 4.469879518072288e-05,
125
+ "loss": 0.1214,
126
+ "step": 2000
127
+ },
128
+ {
129
+ "epoch": 64.51,
130
+ "eval_cer": 0.1337117001405021,
131
+ "eval_loss": 0.6362873315811157,
132
+ "eval_runtime": 271.8138,
133
+ "eval_samples_per_second": 26.735,
134
+ "eval_steps_per_second": 3.344,
135
+ "eval_wer": 0.5546007167891016,
136
+ "step": 2000
137
+ },
138
+ {
139
+ "epoch": 72.57,
140
+ "learning_rate": 4.048192771084337e-05,
141
+ "loss": 0.1011,
142
+ "step": 2250
143
+ },
144
+ {
145
+ "epoch": 72.57,
146
+ "eval_cer": 0.12238677519991394,
147
+ "eval_loss": 0.6309632658958435,
148
+ "eval_runtime": 276.7543,
149
+ "eval_samples_per_second": 26.258,
150
+ "eval_steps_per_second": 3.285,
151
+ "eval_wer": 0.5221669999801992,
152
+ "step": 2250
153
+ },
154
+ {
155
+ "epoch": 80.63,
156
+ "learning_rate": 3.6265060240963855e-05,
157
+ "loss": 0.0879,
158
+ "step": 2500
159
+ },
160
+ {
161
+ "epoch": 80.63,
162
+ "eval_cer": 0.12531743366899534,
163
+ "eval_loss": 0.6352854371070862,
164
+ "eval_runtime": 270.8217,
165
+ "eval_samples_per_second": 26.833,
166
+ "eval_steps_per_second": 3.356,
167
+ "eval_wer": 0.5258301487040374,
168
+ "step": 2500
169
+ },
170
+ {
171
+ "epoch": 88.7,
172
+ "learning_rate": 3.2048192771084335e-05,
173
+ "loss": 0.0782,
174
+ "step": 2750
175
+ },
176
+ {
177
+ "epoch": 88.7,
178
+ "eval_cer": 0.11265268600227542,
179
+ "eval_loss": 0.607792854309082,
180
+ "eval_runtime": 270.7843,
181
+ "eval_samples_per_second": 26.837,
182
+ "eval_steps_per_second": 3.357,
183
+ "eval_wer": 0.4904263113082391,
184
+ "step": 2750
185
+ },
186
+ {
187
+ "epoch": 96.76,
188
+ "learning_rate": 2.783132530120482e-05,
189
+ "loss": 0.0709,
190
+ "step": 3000
191
+ },
192
+ {
193
+ "epoch": 96.76,
194
+ "eval_cer": 0.11539100982210675,
195
+ "eval_loss": 0.6464908123016357,
196
+ "eval_runtime": 272.9311,
197
+ "eval_samples_per_second": 26.626,
198
+ "eval_steps_per_second": 3.331,
199
+ "eval_wer": 0.49601013801160326,
200
+ "step": 3000
201
+ },
202
+ {
203
+ "epoch": 104.82,
204
+ "learning_rate": 2.36144578313253e-05,
205
+ "loss": 0.0661,
206
+ "step": 3250
207
+ },
208
+ {
209
+ "epoch": 104.82,
210
+ "eval_cer": 0.11656783708277235,
211
+ "eval_loss": 0.6621575951576233,
212
+ "eval_runtime": 270.3992,
213
+ "eval_samples_per_second": 26.875,
214
+ "eval_steps_per_second": 3.362,
215
+ "eval_wer": 0.494544878522068,
216
+ "step": 3250
217
+ },
218
+ {
219
+ "epoch": 112.89,
220
+ "learning_rate": 1.9397590361445782e-05,
221
+ "loss": 0.0616,
222
+ "step": 3500
223
+ },
224
+ {
225
+ "epoch": 112.89,
226
+ "eval_cer": 0.11035119002989337,
227
+ "eval_loss": 0.6440250277519226,
228
+ "eval_runtime": 271.0109,
229
+ "eval_samples_per_second": 26.814,
230
+ "eval_steps_per_second": 3.354,
231
+ "eval_wer": 0.47860523137239375,
232
+ "step": 3500
233
+ },
234
+ {
235
+ "epoch": 120.95,
236
+ "learning_rate": 1.5180722891566264e-05,
237
+ "loss": 0.0579,
238
+ "step": 3750
239
+ },
240
+ {
241
+ "epoch": 120.95,
242
+ "eval_cer": 0.11444237621309375,
243
+ "eval_loss": 0.6815317273139954,
244
+ "eval_runtime": 269.1998,
245
+ "eval_samples_per_second": 26.995,
246
+ "eval_steps_per_second": 3.377,
247
+ "eval_wer": 0.4887432429756648,
248
+ "step": 3750
249
+ },
250
+ {
251
+ "epoch": 129.03,
252
+ "learning_rate": 1.0963855421686746e-05,
253
+ "loss": 0.0549,
254
+ "step": 4000
255
+ },
256
+ {
257
+ "epoch": 129.03,
258
+ "eval_cer": 0.11051744540466885,
259
+ "eval_loss": 0.6602992415428162,
260
+ "eval_runtime": 274.1737,
261
+ "eval_samples_per_second": 26.505,
262
+ "eval_steps_per_second": 3.315,
263
+ "eval_wer": 0.47799140645110194,
264
+ "step": 4000
265
+ },
266
+ {
267
+ "epoch": 137.09,
268
+ "learning_rate": 6.746987951807228e-06,
269
+ "loss": 0.0527,
270
+ "step": 4250
271
+ },
272
+ {
273
+ "epoch": 137.09,
274
+ "eval_cer": 0.10900158757583364,
275
+ "eval_loss": 0.6652226448059082,
276
+ "eval_runtime": 271.842,
277
+ "eval_samples_per_second": 26.732,
278
+ "eval_steps_per_second": 3.344,
279
+ "eval_wer": 0.47486287943290495,
280
+ "step": 4250
281
+ }
282
+ ],
283
+ "max_steps": 4650,
284
+ "num_train_epochs": 150,
285
+ "total_flos": 3.524250563411945e+20,
286
+ "trial_name": null,
287
+ "trial_params": null
288
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74b41d01827c8f175a52458dbdf0a94c6386cb87c0c1454cee96f5e60cef0701
3
+ size 3055