mikhail-panzo commited on
Commit
fcec71d
1 Parent(s): 03c12fb

Training in progress, step 2000, checkpoint

Browse files
checkpoint-2000/added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "<ctc_blank>": 80,
3
+ "<mask>": 79
4
+ }
checkpoint-2000/config.json ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/speecht5_tts",
3
+ "activation_dropout": 0.1,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "SpeechT5ForTextToSpeech"
7
+ ],
8
+ "attention_dropout": 0.1,
9
+ "bos_token_id": 0,
10
+ "conv_bias": false,
11
+ "conv_dim": [
12
+ 512,
13
+ 512,
14
+ 512,
15
+ 512,
16
+ 512,
17
+ 512,
18
+ 512
19
+ ],
20
+ "conv_kernel": [
21
+ 10,
22
+ 3,
23
+ 3,
24
+ 3,
25
+ 3,
26
+ 2,
27
+ 2
28
+ ],
29
+ "conv_stride": [
30
+ 5,
31
+ 2,
32
+ 2,
33
+ 2,
34
+ 2,
35
+ 2,
36
+ 2
37
+ ],
38
+ "decoder_attention_heads": 12,
39
+ "decoder_ffn_dim": 3072,
40
+ "decoder_layerdrop": 0.1,
41
+ "decoder_layers": 6,
42
+ "decoder_start_token_id": 2,
43
+ "encoder_attention_heads": 12,
44
+ "encoder_ffn_dim": 3072,
45
+ "encoder_layerdrop": 0.1,
46
+ "encoder_layers": 12,
47
+ "encoder_max_relative_position": 160,
48
+ "eos_token_id": 2,
49
+ "feat_extract_activation": "gelu",
50
+ "feat_extract_norm": "group",
51
+ "feat_proj_dropout": 0.0,
52
+ "guided_attention_loss_num_heads": 2,
53
+ "guided_attention_loss_scale": 10.0,
54
+ "guided_attention_loss_sigma": 0.4,
55
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.1,
57
+ "hidden_size": 768,
58
+ "initializer_range": 0.02,
59
+ "is_encoder_decoder": true,
60
+ "layer_norm_eps": 1e-05,
61
+ "mask_feature_length": 10,
62
+ "mask_feature_min_masks": 0,
63
+ "mask_feature_prob": 0.0,
64
+ "mask_time_length": 10,
65
+ "mask_time_min_masks": 2,
66
+ "mask_time_prob": 0.05,
67
+ "max_length": 1876,
68
+ "max_speech_positions": 1876,
69
+ "max_text_positions": 600,
70
+ "model_type": "speecht5",
71
+ "num_conv_pos_embedding_groups": 16,
72
+ "num_conv_pos_embeddings": 128,
73
+ "num_feat_extract_layers": 7,
74
+ "num_mel_bins": 80,
75
+ "pad_token_id": 1,
76
+ "positional_dropout": 0.1,
77
+ "reduction_factor": 2,
78
+ "scale_embedding": false,
79
+ "speaker_embedding_dim": 512,
80
+ "speech_decoder_postnet_dropout": 0.5,
81
+ "speech_decoder_postnet_kernel": 5,
82
+ "speech_decoder_postnet_layers": 5,
83
+ "speech_decoder_postnet_units": 256,
84
+ "speech_decoder_prenet_dropout": 0.5,
85
+ "speech_decoder_prenet_layers": 2,
86
+ "speech_decoder_prenet_units": 256,
87
+ "torch_dtype": "float32",
88
+ "transformers_version": "4.41.0.dev0",
89
+ "use_cache": false,
90
+ "use_guided_attention_loss": true,
91
+ "vocab_size": 81
92
+ }
checkpoint-2000/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "decoder_start_token_id": 2,
5
+ "eos_token_id": 2,
6
+ "max_length": 1876,
7
+ "pad_token_id": 1,
8
+ "transformers_version": "4.41.0.dev0"
9
+ }
checkpoint-2000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52912110366f353648456d94db4eb3eededa1a7bd1e1ee948a08978fa689fc86
3
+ size 577789320
checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc09339c0743ae93f2c6d529ea7fd41ef8b4d7558d5651b00ff3f3def1959bb5
3
+ size 1155772233
checkpoint-2000/preprocessor_config.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": false,
3
+ "feature_extractor_type": "SpeechT5FeatureExtractor",
4
+ "feature_size": 1,
5
+ "fmax": 7600,
6
+ "fmin": 80,
7
+ "frame_signal_scale": 1.0,
8
+ "hop_length": 16,
9
+ "mel_floor": 1e-10,
10
+ "num_mel_bins": 80,
11
+ "padding_side": "right",
12
+ "padding_value": 0.0,
13
+ "processor_class": "SpeechT5Processor",
14
+ "reduction_factor": 2,
15
+ "return_attention_mask": true,
16
+ "sampling_rate": 16000,
17
+ "win_function": "hann_window",
18
+ "win_length": 64
19
+ }
checkpoint-2000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:024d0cf30a7dc2cd699fa88d8139043fbfd6dbc8be6ed6931372b1fb1049dcda
3
+ size 14244
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b0b2a4fb4f9beb82c2cdc2dd6c09abca2bfc108bd9784a5fcb4addd2fba2f91
3
+ size 1064
checkpoint-2000/special_tokens_map.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "mask_token": {
5
+ "content": "<mask>",
6
+ "lstrip": true,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false
10
+ },
11
+ "pad_token": "<pad>",
12
+ "unk_token": "<unk>"
13
+ }
checkpoint-2000/spm_char.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7fcc48f3e225f627b1641db410ceb0c8649bd2b0c982e150b03f8be3728ab560
3
+ size 238473
checkpoint-2000/tokenizer_config.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<pad>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "</s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<unk>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "79": {
36
+ "content": "<mask>",
37
+ "lstrip": true,
38
+ "normalized": true,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "80": {
44
+ "content": "<ctc_blank>",
45
+ "lstrip": false,
46
+ "normalized": true,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": false
50
+ }
51
+ },
52
+ "bos_token": "<s>",
53
+ "clean_up_tokenization_spaces": true,
54
+ "eos_token": "</s>",
55
+ "mask_token": "<mask>",
56
+ "model_max_length": 600,
57
+ "normalize": false,
58
+ "pad_token": "<pad>",
59
+ "processor_class": "SpeechT5Processor",
60
+ "sp_model_kwargs": {},
61
+ "tokenizer_class": "SpeechT5Tokenizer",
62
+ "unk_token": "<unk>"
63
+ }
checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,345 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.4031977355480194,
3
+ "best_model_checkpoint": "mikhail-panzo/ceb_b64_le5_s4000/checkpoint-2000",
4
+ "epoch": 79.20792079207921,
5
+ "eval_steps": 500,
6
+ "global_step": 2000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 1.9801980198019802,
13
+ "grad_norm": 2.883972644805908,
14
+ "learning_rate": 2.5000000000000004e-07,
15
+ "loss": 0.8004,
16
+ "step": 50
17
+ },
18
+ {
19
+ "epoch": 3.9603960396039604,
20
+ "grad_norm": 2.5420947074890137,
21
+ "learning_rate": 5.000000000000001e-07,
22
+ "loss": 0.77,
23
+ "step": 100
24
+ },
25
+ {
26
+ "epoch": 5.9405940594059405,
27
+ "grad_norm": 1.3930779695510864,
28
+ "learning_rate": 7.5e-07,
29
+ "loss": 0.7525,
30
+ "step": 150
31
+ },
32
+ {
33
+ "epoch": 7.920792079207921,
34
+ "grad_norm": 1.8414520025253296,
35
+ "learning_rate": 1.0000000000000002e-06,
36
+ "loss": 0.722,
37
+ "step": 200
38
+ },
39
+ {
40
+ "epoch": 9.900990099009901,
41
+ "grad_norm": 2.0665407180786133,
42
+ "learning_rate": 1.25e-06,
43
+ "loss": 0.7076,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 11.881188118811881,
48
+ "grad_norm": 1.2595587968826294,
49
+ "learning_rate": 1.5e-06,
50
+ "loss": 0.6757,
51
+ "step": 300
52
+ },
53
+ {
54
+ "epoch": 13.861386138613861,
55
+ "grad_norm": 1.4587587118148804,
56
+ "learning_rate": 1.75e-06,
57
+ "loss": 0.6562,
58
+ "step": 350
59
+ },
60
+ {
61
+ "epoch": 15.841584158415841,
62
+ "grad_norm": 2.241546154022217,
63
+ "learning_rate": 2.0000000000000003e-06,
64
+ "loss": 0.6016,
65
+ "step": 400
66
+ },
67
+ {
68
+ "epoch": 17.821782178217823,
69
+ "grad_norm": 1.4243369102478027,
70
+ "learning_rate": 2.25e-06,
71
+ "loss": 0.5593,
72
+ "step": 450
73
+ },
74
+ {
75
+ "epoch": 19.801980198019802,
76
+ "grad_norm": 1.6889928579330444,
77
+ "learning_rate": 2.5e-06,
78
+ "loss": 0.536,
79
+ "step": 500
80
+ },
81
+ {
82
+ "epoch": 19.801980198019802,
83
+ "eval_loss": 0.4640245735645294,
84
+ "eval_runtime": 6.7454,
85
+ "eval_samples_per_second": 26.685,
86
+ "eval_steps_per_second": 3.41,
87
+ "step": 500
88
+ },
89
+ {
90
+ "epoch": 21.782178217821784,
91
+ "grad_norm": 1.1601781845092773,
92
+ "learning_rate": 2.7500000000000004e-06,
93
+ "loss": 0.5198,
94
+ "step": 550
95
+ },
96
+ {
97
+ "epoch": 23.762376237623762,
98
+ "grad_norm": 1.1969332695007324,
99
+ "learning_rate": 3e-06,
100
+ "loss": 0.5114,
101
+ "step": 600
102
+ },
103
+ {
104
+ "epoch": 25.742574257425744,
105
+ "grad_norm": 1.1022064685821533,
106
+ "learning_rate": 3.2500000000000002e-06,
107
+ "loss": 0.5092,
108
+ "step": 650
109
+ },
110
+ {
111
+ "epoch": 27.722772277227723,
112
+ "grad_norm": 1.2418344020843506,
113
+ "learning_rate": 3.5e-06,
114
+ "loss": 0.4951,
115
+ "step": 700
116
+ },
117
+ {
118
+ "epoch": 29.702970297029704,
119
+ "grad_norm": 1.1083300113677979,
120
+ "learning_rate": 3.7500000000000005e-06,
121
+ "loss": 0.4911,
122
+ "step": 750
123
+ },
124
+ {
125
+ "epoch": 31.683168316831683,
126
+ "grad_norm": 1.2230323553085327,
127
+ "learning_rate": 4.000000000000001e-06,
128
+ "loss": 0.4866,
129
+ "step": 800
130
+ },
131
+ {
132
+ "epoch": 33.663366336633665,
133
+ "grad_norm": 1.1232423782348633,
134
+ "learning_rate": 4.25e-06,
135
+ "loss": 0.4876,
136
+ "step": 850
137
+ },
138
+ {
139
+ "epoch": 35.64356435643565,
140
+ "grad_norm": 0.9751402139663696,
141
+ "learning_rate": 4.5e-06,
142
+ "loss": 0.4772,
143
+ "step": 900
144
+ },
145
+ {
146
+ "epoch": 37.62376237623762,
147
+ "grad_norm": 1.0373197793960571,
148
+ "learning_rate": 4.75e-06,
149
+ "loss": 0.4791,
150
+ "step": 950
151
+ },
152
+ {
153
+ "epoch": 39.603960396039604,
154
+ "grad_norm": 1.0164889097213745,
155
+ "learning_rate": 5e-06,
156
+ "loss": 0.4674,
157
+ "step": 1000
158
+ },
159
+ {
160
+ "epoch": 39.603960396039604,
161
+ "eval_loss": 0.4247289001941681,
162
+ "eval_runtime": 6.5527,
163
+ "eval_samples_per_second": 27.47,
164
+ "eval_steps_per_second": 3.51,
165
+ "step": 1000
166
+ },
167
+ {
168
+ "epoch": 41.584158415841586,
169
+ "grad_norm": 1.087668538093567,
170
+ "learning_rate": 5.2500000000000006e-06,
171
+ "loss": 0.4754,
172
+ "step": 1050
173
+ },
174
+ {
175
+ "epoch": 43.56435643564357,
176
+ "grad_norm": 0.891859769821167,
177
+ "learning_rate": 5.500000000000001e-06,
178
+ "loss": 0.4677,
179
+ "step": 1100
180
+ },
181
+ {
182
+ "epoch": 45.54455445544554,
183
+ "grad_norm": 1.018160343170166,
184
+ "learning_rate": 5.75e-06,
185
+ "loss": 0.4642,
186
+ "step": 1150
187
+ },
188
+ {
189
+ "epoch": 47.524752475247524,
190
+ "grad_norm": 1.4765816926956177,
191
+ "learning_rate": 6e-06,
192
+ "loss": 0.4602,
193
+ "step": 1200
194
+ },
195
+ {
196
+ "epoch": 49.504950495049506,
197
+ "grad_norm": 1.3534387350082397,
198
+ "learning_rate": 6.25e-06,
199
+ "loss": 0.4634,
200
+ "step": 1250
201
+ },
202
+ {
203
+ "epoch": 51.48514851485149,
204
+ "grad_norm": 1.031537413597107,
205
+ "learning_rate": 6.5000000000000004e-06,
206
+ "loss": 0.4567,
207
+ "step": 1300
208
+ },
209
+ {
210
+ "epoch": 53.46534653465346,
211
+ "grad_norm": 1.0172531604766846,
212
+ "learning_rate": 6.750000000000001e-06,
213
+ "loss": 0.4592,
214
+ "step": 1350
215
+ },
216
+ {
217
+ "epoch": 55.445544554455445,
218
+ "grad_norm": 0.8680336475372314,
219
+ "learning_rate": 7e-06,
220
+ "loss": 0.4583,
221
+ "step": 1400
222
+ },
223
+ {
224
+ "epoch": 57.42574257425743,
225
+ "grad_norm": 1.2397912740707397,
226
+ "learning_rate": 7.25e-06,
227
+ "loss": 0.4459,
228
+ "step": 1450
229
+ },
230
+ {
231
+ "epoch": 59.40594059405941,
232
+ "grad_norm": 1.021185040473938,
233
+ "learning_rate": 7.500000000000001e-06,
234
+ "loss": 0.4479,
235
+ "step": 1500
236
+ },
237
+ {
238
+ "epoch": 59.40594059405941,
239
+ "eval_loss": 0.4102235734462738,
240
+ "eval_runtime": 7.6974,
241
+ "eval_samples_per_second": 23.384,
242
+ "eval_steps_per_second": 2.988,
243
+ "step": 1500
244
+ },
245
+ {
246
+ "epoch": 61.386138613861384,
247
+ "grad_norm": 1.005591630935669,
248
+ "learning_rate": 7.75e-06,
249
+ "loss": 0.4469,
250
+ "step": 1550
251
+ },
252
+ {
253
+ "epoch": 63.366336633663366,
254
+ "grad_norm": 0.8193312287330627,
255
+ "learning_rate": 8.000000000000001e-06,
256
+ "loss": 0.4425,
257
+ "step": 1600
258
+ },
259
+ {
260
+ "epoch": 65.34653465346534,
261
+ "grad_norm": 0.9143276810646057,
262
+ "learning_rate": 8.25e-06,
263
+ "loss": 0.4424,
264
+ "step": 1650
265
+ },
266
+ {
267
+ "epoch": 67.32673267326733,
268
+ "grad_norm": 1.0127445459365845,
269
+ "learning_rate": 8.5e-06,
270
+ "loss": 0.4446,
271
+ "step": 1700
272
+ },
273
+ {
274
+ "epoch": 69.3069306930693,
275
+ "grad_norm": 1.045551061630249,
276
+ "learning_rate": 8.750000000000001e-06,
277
+ "loss": 0.4417,
278
+ "step": 1750
279
+ },
280
+ {
281
+ "epoch": 71.2871287128713,
282
+ "grad_norm": 0.9629473090171814,
283
+ "learning_rate": 9e-06,
284
+ "loss": 0.4362,
285
+ "step": 1800
286
+ },
287
+ {
288
+ "epoch": 73.26732673267327,
289
+ "grad_norm": 0.9655708074569702,
290
+ "learning_rate": 9.250000000000001e-06,
291
+ "loss": 0.4393,
292
+ "step": 1850
293
+ },
294
+ {
295
+ "epoch": 75.24752475247524,
296
+ "grad_norm": 1.167992115020752,
297
+ "learning_rate": 9.5e-06,
298
+ "loss": 0.4373,
299
+ "step": 1900
300
+ },
301
+ {
302
+ "epoch": 77.22772277227723,
303
+ "grad_norm": 0.9866817593574524,
304
+ "learning_rate": 9.75e-06,
305
+ "loss": 0.4369,
306
+ "step": 1950
307
+ },
308
+ {
309
+ "epoch": 79.20792079207921,
310
+ "grad_norm": 0.8923723101615906,
311
+ "learning_rate": 1e-05,
312
+ "loss": 0.4356,
313
+ "step": 2000
314
+ },
315
+ {
316
+ "epoch": 79.20792079207921,
317
+ "eval_loss": 0.4031977355480194,
318
+ "eval_runtime": 7.0669,
319
+ "eval_samples_per_second": 25.471,
320
+ "eval_steps_per_second": 3.255,
321
+ "step": 2000
322
+ }
323
+ ],
324
+ "logging_steps": 50,
325
+ "max_steps": 4000,
326
+ "num_input_tokens_seen": 0,
327
+ "num_train_epochs": 160,
328
+ "save_steps": 500,
329
+ "stateful_callbacks": {
330
+ "TrainerControl": {
331
+ "args": {
332
+ "should_epoch_stop": false,
333
+ "should_evaluate": false,
334
+ "should_log": false,
335
+ "should_save": true,
336
+ "should_training_stop": false
337
+ },
338
+ "attributes": {}
339
+ }
340
+ },
341
+ "total_flos": 2.161155031107257e+16,
342
+ "train_batch_size": 16,
343
+ "trial_name": null,
344
+ "trial_params": null
345
+ }
checkpoint-2000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:893b77dc1258076eac6823109e549347547b5ddcab0a290335f22ad5e88711bb
3
+ size 5304