mikhail-panzo commited on
Commit
28ac5bc
1 Parent(s): 0093cd6

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 sha256:623862acf0ccc94122a296e6fbed8c9c4412e67661cd9ff9629e112a95188db3
3
+ size 577789320
checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:837edb13ef2872550b838a6ea5f0d2cb40fb544d3496b0c5a258791e7dfd19f4
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 sha256:4831366ec1dbcb558f23d85b2403f0009413fae4ecc65be8d93f368ca0b82cd3
3
+ size 14244
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8200c7911b70674e95f6ea3c4fd4f7c5a623c58e7efa3752996c3106e6a97146
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": 0.4244868755340576,
3
+ "best_model_checkpoint": "mikhail-panzo/fil_b32_le5_s4000/checkpoint-2000",
4
+ "epoch": 44.44444444444444,
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": 1.1111111111111112,
13
+ "grad_norm": 4.398683071136475,
14
+ "learning_rate": 2.4500000000000004e-07,
15
+ "loss": 0.8108,
16
+ "step": 50
17
+ },
18
+ {
19
+ "epoch": 2.2222222222222223,
20
+ "grad_norm": 3.726436138153076,
21
+ "learning_rate": 4.95e-07,
22
+ "loss": 0.792,
23
+ "step": 100
24
+ },
25
+ {
26
+ "epoch": 3.3333333333333335,
27
+ "grad_norm": 2.0379676818847656,
28
+ "learning_rate": 7.450000000000001e-07,
29
+ "loss": 0.7413,
30
+ "step": 150
31
+ },
32
+ {
33
+ "epoch": 4.444444444444445,
34
+ "grad_norm": 2.0460364818573,
35
+ "learning_rate": 9.950000000000002e-07,
36
+ "loss": 0.7342,
37
+ "step": 200
38
+ },
39
+ {
40
+ "epoch": 5.555555555555555,
41
+ "grad_norm": 2.803539276123047,
42
+ "learning_rate": 1.2450000000000002e-06,
43
+ "loss": 0.7147,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 6.666666666666667,
48
+ "grad_norm": 1.9818241596221924,
49
+ "learning_rate": 1.495e-06,
50
+ "loss": 0.7072,
51
+ "step": 300
52
+ },
53
+ {
54
+ "epoch": 7.777777777777778,
55
+ "grad_norm": 3.081399917602539,
56
+ "learning_rate": 1.745e-06,
57
+ "loss": 0.6843,
58
+ "step": 350
59
+ },
60
+ {
61
+ "epoch": 8.88888888888889,
62
+ "grad_norm": 2.4465785026550293,
63
+ "learning_rate": 1.9950000000000004e-06,
64
+ "loss": 0.657,
65
+ "step": 400
66
+ },
67
+ {
68
+ "epoch": 10.0,
69
+ "grad_norm": 2.929323434829712,
70
+ "learning_rate": 2.245e-06,
71
+ "loss": 0.6597,
72
+ "step": 450
73
+ },
74
+ {
75
+ "epoch": 11.11111111111111,
76
+ "grad_norm": 1.419311285018921,
77
+ "learning_rate": 2.4950000000000003e-06,
78
+ "loss": 0.6246,
79
+ "step": 500
80
+ },
81
+ {
82
+ "epoch": 11.11111111111111,
83
+ "eval_loss": 0.5375720262527466,
84
+ "eval_runtime": 9.2568,
85
+ "eval_samples_per_second": 17.177,
86
+ "eval_steps_per_second": 2.161,
87
+ "step": 500
88
+ },
89
+ {
90
+ "epoch": 12.222222222222221,
91
+ "grad_norm": 2.3438374996185303,
92
+ "learning_rate": 2.7450000000000004e-06,
93
+ "loss": 0.6018,
94
+ "step": 550
95
+ },
96
+ {
97
+ "epoch": 13.333333333333334,
98
+ "grad_norm": 1.7441222667694092,
99
+ "learning_rate": 2.995e-06,
100
+ "loss": 0.5547,
101
+ "step": 600
102
+ },
103
+ {
104
+ "epoch": 14.444444444444445,
105
+ "grad_norm": 1.5592021942138672,
106
+ "learning_rate": 3.2450000000000003e-06,
107
+ "loss": 0.5466,
108
+ "step": 650
109
+ },
110
+ {
111
+ "epoch": 15.555555555555555,
112
+ "grad_norm": 1.305444359779358,
113
+ "learning_rate": 3.495e-06,
114
+ "loss": 0.535,
115
+ "step": 700
116
+ },
117
+ {
118
+ "epoch": 16.666666666666668,
119
+ "grad_norm": 2.2067477703094482,
120
+ "learning_rate": 3.745e-06,
121
+ "loss": 0.5343,
122
+ "step": 750
123
+ },
124
+ {
125
+ "epoch": 17.77777777777778,
126
+ "grad_norm": 1.667480707168579,
127
+ "learning_rate": 3.995000000000001e-06,
128
+ "loss": 0.5296,
129
+ "step": 800
130
+ },
131
+ {
132
+ "epoch": 18.88888888888889,
133
+ "grad_norm": 1.5465797185897827,
134
+ "learning_rate": 4.245e-06,
135
+ "loss": 0.5198,
136
+ "step": 850
137
+ },
138
+ {
139
+ "epoch": 20.0,
140
+ "grad_norm": 2.760589361190796,
141
+ "learning_rate": 4.495e-06,
142
+ "loss": 0.5193,
143
+ "step": 900
144
+ },
145
+ {
146
+ "epoch": 21.11111111111111,
147
+ "grad_norm": 1.4570050239562988,
148
+ "learning_rate": 4.745e-06,
149
+ "loss": 0.5124,
150
+ "step": 950
151
+ },
152
+ {
153
+ "epoch": 22.22222222222222,
154
+ "grad_norm": 2.1914265155792236,
155
+ "learning_rate": 4.9950000000000005e-06,
156
+ "loss": 0.5034,
157
+ "step": 1000
158
+ },
159
+ {
160
+ "epoch": 22.22222222222222,
161
+ "eval_loss": 0.45661377906799316,
162
+ "eval_runtime": 9.2158,
163
+ "eval_samples_per_second": 17.253,
164
+ "eval_steps_per_second": 2.17,
165
+ "step": 1000
166
+ },
167
+ {
168
+ "epoch": 23.333333333333332,
169
+ "grad_norm": 1.941751480102539,
170
+ "learning_rate": 5.245e-06,
171
+ "loss": 0.5046,
172
+ "step": 1050
173
+ },
174
+ {
175
+ "epoch": 24.444444444444443,
176
+ "grad_norm": 1.9293475151062012,
177
+ "learning_rate": 5.495000000000001e-06,
178
+ "loss": 0.4948,
179
+ "step": 1100
180
+ },
181
+ {
182
+ "epoch": 25.555555555555557,
183
+ "grad_norm": 1.3278849124908447,
184
+ "learning_rate": 5.745000000000001e-06,
185
+ "loss": 0.4931,
186
+ "step": 1150
187
+ },
188
+ {
189
+ "epoch": 26.666666666666668,
190
+ "grad_norm": 1.1657841205596924,
191
+ "learning_rate": 5.995000000000001e-06,
192
+ "loss": 0.496,
193
+ "step": 1200
194
+ },
195
+ {
196
+ "epoch": 27.77777777777778,
197
+ "grad_norm": 1.8730145692825317,
198
+ "learning_rate": 6.245000000000001e-06,
199
+ "loss": 0.4939,
200
+ "step": 1250
201
+ },
202
+ {
203
+ "epoch": 28.88888888888889,
204
+ "grad_norm": 2.1156740188598633,
205
+ "learning_rate": 6.4950000000000005e-06,
206
+ "loss": 0.4908,
207
+ "step": 1300
208
+ },
209
+ {
210
+ "epoch": 30.0,
211
+ "grad_norm": 1.799806833267212,
212
+ "learning_rate": 6.745000000000001e-06,
213
+ "loss": 0.4875,
214
+ "step": 1350
215
+ },
216
+ {
217
+ "epoch": 31.11111111111111,
218
+ "grad_norm": 1.762253761291504,
219
+ "learning_rate": 6.995000000000001e-06,
220
+ "loss": 0.4836,
221
+ "step": 1400
222
+ },
223
+ {
224
+ "epoch": 32.22222222222222,
225
+ "grad_norm": 1.305191993713379,
226
+ "learning_rate": 7.245000000000001e-06,
227
+ "loss": 0.4811,
228
+ "step": 1450
229
+ },
230
+ {
231
+ "epoch": 33.333333333333336,
232
+ "grad_norm": 1.5870022773742676,
233
+ "learning_rate": 7.495000000000001e-06,
234
+ "loss": 0.4807,
235
+ "step": 1500
236
+ },
237
+ {
238
+ "epoch": 33.333333333333336,
239
+ "eval_loss": 0.43601250648498535,
240
+ "eval_runtime": 9.5435,
241
+ "eval_samples_per_second": 16.66,
242
+ "eval_steps_per_second": 2.096,
243
+ "step": 1500
244
+ },
245
+ {
246
+ "epoch": 34.44444444444444,
247
+ "grad_norm": 2.632877826690674,
248
+ "learning_rate": 7.745e-06,
249
+ "loss": 0.4782,
250
+ "step": 1550
251
+ },
252
+ {
253
+ "epoch": 35.55555555555556,
254
+ "grad_norm": 1.3705276250839233,
255
+ "learning_rate": 7.995e-06,
256
+ "loss": 0.48,
257
+ "step": 1600
258
+ },
259
+ {
260
+ "epoch": 36.666666666666664,
261
+ "grad_norm": 1.9232587814331055,
262
+ "learning_rate": 8.245000000000002e-06,
263
+ "loss": 0.47,
264
+ "step": 1650
265
+ },
266
+ {
267
+ "epoch": 37.77777777777778,
268
+ "grad_norm": 1.3885018825531006,
269
+ "learning_rate": 8.495e-06,
270
+ "loss": 0.4711,
271
+ "step": 1700
272
+ },
273
+ {
274
+ "epoch": 38.888888888888886,
275
+ "grad_norm": 2.4857094287872314,
276
+ "learning_rate": 8.745000000000002e-06,
277
+ "loss": 0.4721,
278
+ "step": 1750
279
+ },
280
+ {
281
+ "epoch": 40.0,
282
+ "grad_norm": 1.7662663459777832,
283
+ "learning_rate": 8.995000000000001e-06,
284
+ "loss": 0.4656,
285
+ "step": 1800
286
+ },
287
+ {
288
+ "epoch": 41.111111111111114,
289
+ "grad_norm": 1.906548261642456,
290
+ "learning_rate": 9.245e-06,
291
+ "loss": 0.463,
292
+ "step": 1850
293
+ },
294
+ {
295
+ "epoch": 42.22222222222222,
296
+ "grad_norm": 1.7204883098602295,
297
+ "learning_rate": 9.495000000000001e-06,
298
+ "loss": 0.4592,
299
+ "step": 1900
300
+ },
301
+ {
302
+ "epoch": 43.333333333333336,
303
+ "grad_norm": 1.3313167095184326,
304
+ "learning_rate": 9.745e-06,
305
+ "loss": 0.4581,
306
+ "step": 1950
307
+ },
308
+ {
309
+ "epoch": 44.44444444444444,
310
+ "grad_norm": 1.94256591796875,
311
+ "learning_rate": 9.995000000000002e-06,
312
+ "loss": 0.4582,
313
+ "step": 2000
314
+ },
315
+ {
316
+ "epoch": 44.44444444444444,
317
+ "eval_loss": 0.4244868755340576,
318
+ "eval_runtime": 9.1316,
319
+ "eval_samples_per_second": 17.412,
320
+ "eval_steps_per_second": 2.19,
321
+ "step": 2000
322
+ }
323
+ ],
324
+ "logging_steps": 50,
325
+ "max_steps": 4000,
326
+ "num_input_tokens_seen": 0,
327
+ "num_train_epochs": 89,
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": 1.3753846024576416e+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 sha256:05b7ecdded947540b9670cedf7cfe87afc7199b237efb642b915ffbb505d51dc
3
+ size 5304