Hugging Face's logo

sanchit-gandhi
/
parler-tts-mini-v0.1-expresso-combined

Text2Text Generation
Transformers
Safetensors
parler_tts
Inference Endpoints
Model card Files Community
sanchit-gandhi HF staff commited on
Commit
4d6df9c
•
1 Parent(s): df60576

remove whispering

Browse files
Files changed (28) hide show
  1. .gitignore +1 -0
  2. checkpoint-1120-epoch-6/config.json +276 -0
  3. checkpoint-1120-epoch-6/generation_config.json +12 -0
  4. checkpoint-1120-epoch-6/optimizer.bin +1 -1
  5. checkpoint-1120-epoch-6/pytorch_model.bin +1 -1
  6. checkpoint-1120-epoch-6/random_states_0.pkl +2 -2
  7. checkpoint-1280-epoch-7/config.json +276 -0
  8. checkpoint-1280-epoch-7/generation_config.json +12 -0
  9. checkpoint-1280-epoch-7/optimizer.bin +1 -1
  10. checkpoint-1280-epoch-7/pytorch_model.bin +1 -1
  11. checkpoint-1280-epoch-7/random_states_0.pkl +2 -2
  12. checkpoint-640-epoch-3/config.json +276 -0
  13. checkpoint-640-epoch-3/generation_config.json +12 -0
  14. checkpoint-640-epoch-3/optimizer.bin +1 -1
  15. checkpoint-640-epoch-3/pytorch_model.bin +1 -1
  16. checkpoint-640-epoch-3/random_states_0.pkl +2 -2
  17. checkpoint-800-epoch-4/config.json +1 -1
  18. checkpoint-800-epoch-4/optimizer.bin +1 -1
  19. checkpoint-800-epoch-4/pytorch_model.bin +1 -1
  20. checkpoint-800-epoch-4/random_states_0.pkl +2 -2
  21. checkpoint-960-epoch-5/config.json +276 -0
  22. checkpoint-960-epoch-5/generation_config.json +12 -0
  23. checkpoint-960-epoch-5/optimizer.bin +1 -1
  24. checkpoint-960-epoch-5/pytorch_model.bin +1 -1
  25. checkpoint-960-epoch-5/random_states_0.pkl +2 -2
  26. finetuning_config.json +2 -1
  27. model.safetensors +1 -1
  28. run_parler_tts_training.py +10 -0
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ wandb
checkpoint-1120-epoch-6/config.json ADDED
@@ -0,0 +1,276 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
3
+ "architectures": [
4
+ "ParlerTTSForConditionalGeneration"
5
+ ],
6
+ "audio_encoder": {
7
+ "_name_or_path": "ylacombe/dac_44khZ_8kbps",
8
+ "add_cross_attention": false,
9
+ "architectures": [
10
+ "DACModel"
11
+ ],
12
+ "bad_words_ids": null,
13
+ "begin_suppress_tokens": null,
14
+ "bos_token_id": null,
15
+ "chunk_size_feed_forward": 0,
16
+ "codebook_size": 1024,
17
+ "cross_attention_hidden_size": null,
18
+ "decoder_start_token_id": null,
19
+ "diversity_penalty": 0.0,
20
+ "do_sample": false,
21
+ "early_stopping": false,
22
+ "encoder_no_repeat_ngram_size": 0,
23
+ "eos_token_id": null,
24
+ "exponential_decay_length_penalty": null,
25
+ "finetuning_task": null,
26
+ "forced_bos_token_id": null,
27
+ "forced_eos_token_id": null,
28
+ "frame_rate": 86,
29
+ "id2label": {
30
+ "0": "LABEL_0",
31
+ "1": "LABEL_1"
32
+ },
33
+ "is_decoder": false,
34
+ "is_encoder_decoder": false,
35
+ "label2id": {
36
+ "LABEL_0": 0,
37
+ "LABEL_1": 1
38
+ },
39
+ "latent_dim": 1024,
40
+ "length_penalty": 1.0,
41
+ "max_length": 20,
42
+ "min_length": 0,
43
+ "model_bitrate": 8,
44
+ "model_type": "dac",
45
+ "no_repeat_ngram_size": 0,
46
+ "num_beam_groups": 1,
47
+ "num_beams": 1,
48
+ "num_codebooks": 9,
49
+ "num_return_sequences": 1,
50
+ "output_attentions": false,
51
+ "output_hidden_states": false,
52
+ "output_scores": false,
53
+ "pad_token_id": null,
54
+ "prefix": null,
55
+ "problem_type": null,
56
+ "pruned_heads": {},
57
+ "remove_invalid_values": false,
58
+ "repetition_penalty": 1.0,
59
+ "return_dict": true,
60
+ "return_dict_in_generate": false,
61
+ "sampling_rate": 44100,
62
+ "sep_token_id": null,
63
+ "suppress_tokens": null,
64
+ "task_specific_params": null,
65
+ "temperature": 1.0,
66
+ "tf_legacy_loss": false,
67
+ "tie_encoder_decoder": false,
68
+ "tie_word_embeddings": true,
69
+ "tokenizer_class": null,
70
+ "top_k": 50,
71
+ "top_p": 1.0,
72
+ "torch_dtype": "float32",
73
+ "torchscript": false,
74
+ "typical_p": 1.0,
75
+ "use_bfloat16": false
76
+ },
77
+ "decoder": {
78
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
79
+ "activation_dropout": 0.0,
80
+ "activation_function": "gelu",
81
+ "add_cross_attention": true,
82
+ "architectures": [
83
+ "ParlerTTSForCausalLM"
84
+ ],
85
+ "attention_dropout": 0.0,
86
+ "bad_words_ids": null,
87
+ "begin_suppress_tokens": null,
88
+ "bos_token_id": 1025,
89
+ "chunk_size_feed_forward": 0,
90
+ "cross_attention_hidden_size": null,
91
+ "decoder_start_token_id": null,
92
+ "diversity_penalty": 0.0,
93
+ "do_sample": false,
94
+ "dropout": 0.1,
95
+ "early_stopping": false,
96
+ "encoder_no_repeat_ngram_size": 0,
97
+ "eos_token_id": 1024,
98
+ "exponential_decay_length_penalty": null,
99
+ "ffn_dim": 4096,
100
+ "finetuning_task": null,
101
+ "forced_bos_token_id": null,
102
+ "forced_eos_token_id": null,
103
+ "hidden_size": 1024,
104
+ "id2label": {
105
+ "0": "LABEL_0",
106
+ "1": "LABEL_1"
107
+ },
108
+ "initializer_factor": 0.02,
109
+ "is_decoder": true,
110
+ "is_encoder_decoder": false,
111
+ "label2id": {
112
+ "LABEL_0": 0,
113
+ "LABEL_1": 1
114
+ },
115
+ "layerdrop": 0.0,
116
+ "length_penalty": 1.0,
117
+ "max_length": 20,
118
+ "max_position_embeddings": 4096,
119
+ "min_length": 0,
120
+ "model_type": "parler_tts_decoder",
121
+ "no_repeat_ngram_size": 0,
122
+ "num_attention_heads": 16,
123
+ "num_beam_groups": 1,
124
+ "num_beams": 1,
125
+ "num_codebooks": 9,
126
+ "num_hidden_layers": 24,
127
+ "num_return_sequences": 1,
128
+ "output_attentions": false,
129
+ "output_hidden_states": false,
130
+ "output_scores": false,
131
+ "pad_token_id": 1024,
132
+ "prefix": null,
133
+ "problem_type": null,
134
+ "pruned_heads": {},
135
+ "remove_invalid_values": false,
136
+ "repetition_penalty": 1.0,
137
+ "return_dict": true,
138
+ "return_dict_in_generate": false,
139
+ "scale_embedding": false,
140
+ "sep_token_id": null,
141
+ "suppress_tokens": null,
142
+ "task_specific_params": null,
143
+ "temperature": 1.0,
144
+ "tf_legacy_loss": false,
145
+ "tie_encoder_decoder": false,
146
+ "tie_word_embeddings": false,
147
+ "tokenizer_class": null,
148
+ "top_k": 50,
149
+ "top_p": 1.0,
150
+ "torch_dtype": "float32",
151
+ "torchscript": false,
152
+ "typical_p": 1.0,
153
+ "use_bfloat16": false,
154
+ "use_cache": true,
155
+ "vocab_size": 1088
156
+ },
157
+ "decoder_start_token_id": 1025,
158
+ "is_encoder_decoder": true,
159
+ "model_type": "parler_tts",
160
+ "pad_token_id": 1024,
161
+ "text_encoder": {
162
+ "_name_or_path": "google/flan-t5-base",
163
+ "add_cross_attention": false,
164
+ "architectures": [
165
+ "T5ForConditionalGeneration"
166
+ ],
167
+ "bad_words_ids": null,
168
+ "begin_suppress_tokens": null,
169
+ "bos_token_id": null,
170
+ "chunk_size_feed_forward": 0,
171
+ "classifier_dropout": 0.0,
172
+ "cross_attention_hidden_size": null,
173
+ "d_ff": 2048,
174
+ "d_kv": 64,
175
+ "d_model": 768,
176
+ "decoder_start_token_id": 0,
177
+ "dense_act_fn": "gelu_new",
178
+ "diversity_penalty": 0.0,
179
+ "do_sample": false,
180
+ "dropout_rate": 0.1,
181
+ "early_stopping": false,
182
+ "encoder_no_repeat_ngram_size": 0,
183
+ "eos_token_id": 1,
184
+ "exponential_decay_length_penalty": null,
185
+ "feed_forward_proj": "gated-gelu",
186
+ "finetuning_task": null,
187
+ "forced_bos_token_id": null,
188
+ "forced_eos_token_id": null,
189
+ "id2label": {
190
+ "0": "LABEL_0",
191
+ "1": "LABEL_1"
192
+ },
193
+ "initializer_factor": 1.0,
194
+ "is_decoder": false,
195
+ "is_encoder_decoder": true,
196
+ "is_gated_act": true,
197
+ "label2id": {
198
+ "LABEL_0": 0,
199
+ "LABEL_1": 1
200
+ },
201
+ "layer_norm_epsilon": 1e-06,
202
+ "length_penalty": 1.0,
203
+ "max_length": 20,
204
+ "min_length": 0,
205
+ "model_type": "t5",
206
+ "n_positions": 512,
207
+ "no_repeat_ngram_size": 0,
208
+ "num_beam_groups": 1,
209
+ "num_beams": 1,
210
+ "num_decoder_layers": 12,
211
+ "num_heads": 12,
212
+ "num_layers": 12,
213
+ "num_return_sequences": 1,
214
+ "output_attentions": false,
215
+ "output_hidden_states": false,
216
+ "output_past": true,
217
+ "output_scores": false,
218
+ "pad_token_id": 0,
219
+ "prefix": null,
220
+ "problem_type": null,
221
+ "pruned_heads": {},
222
+ "relative_attention_max_distance": 128,
223
+ "relative_attention_num_buckets": 32,
224
+ "remove_invalid_values": false,
225
+ "repetition_penalty": 1.0,
226
+ "return_dict": true,
227
+ "return_dict_in_generate": false,
228
+ "sep_token_id": null,
229
+ "suppress_tokens": null,
230
+ "task_specific_params": {
231
+ "summarization": {
232
+ "early_stopping": true,
233
+ "length_penalty": 2.0,
234
+ "max_length": 200,
235
+ "min_length": 30,
236
+ "no_repeat_ngram_size": 3,
237
+ "num_beams": 4,
238
+ "prefix": "summarize: "
239
+ },
240
+ "translation_en_to_de": {
241
+ "early_stopping": true,
242
+ "max_length": 300,
243
+ "num_beams": 4,
244
+ "prefix": "translate English to German: "
245
+ },
246
+ "translation_en_to_fr": {
247
+ "early_stopping": true,
248
+ "max_length": 300,
249
+ "num_beams": 4,
250
+ "prefix": "translate English to French: "
251
+ },
252
+ "translation_en_to_ro": {
253
+ "early_stopping": true,
254
+ "max_length": 300,
255
+ "num_beams": 4,
256
+ "prefix": "translate English to Romanian: "
257
+ }
258
+ },
259
+ "temperature": 1.0,
260
+ "tf_legacy_loss": false,
261
+ "tie_encoder_decoder": false,
262
+ "tie_word_embeddings": false,
263
+ "tokenizer_class": null,
264
+ "top_k": 50,
265
+ "top_p": 1.0,
266
+ "torch_dtype": null,
267
+ "torchscript": false,
268
+ "typical_p": 1.0,
269
+ "use_bfloat16": false,
270
+ "use_cache": true,
271
+ "vocab_size": 32128
272
+ },
273
+ "torch_dtype": "float32",
274
+ "transformers_version": "4.41.0.dev0",
275
+ "vocab_size": 32128
276
+ }
checkpoint-1120-epoch-6/generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1025,
4
+ "decoder_start_token_id": 1025,
5
+ "do_sample": true,
6
+ "eos_token_id": 1024,
7
+ "guidance_scale": 1.0,
8
+ "max_length": 2580,
9
+ "min_new_tokens": 50,
10
+ "pad_token_id": 1024,
11
+ "transformers_version": "4.41.0.dev0"
12
+ }
checkpoint-1120-epoch-6/optimizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:72090bb2fa144ea827c28f55280c9b5d4e50d3f5a89359fcc850079f35309b1e
3
  size 3652763351
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ceac1375e3f4f9a0294fa96223571ba0291d376395749594203cabeb92074bdb
3
  size 3652763351
checkpoint-1120-epoch-6/pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bce47e0f3b778b4ad7ddf6164868262be9682558bb083a0d53841a87279b3537
3
  size 2588462170
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a572a271a6d88d582d7461f7551734c92aa80b7e0c59bacd96fb654f5e4df57d
3
  size 2588462170
checkpoint-1120-epoch-6/random_states_0.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0ca49fbae7eb9705c8b805b1ab5f1049040cacf26d75207ae76b7e964251141d
3
- size 14344
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3024823d6d4fad0542aece913542e358ad9ccd95c7c4a6f90d57ab09b7f19c4
3
+ size 15384
checkpoint-1280-epoch-7/config.json ADDED
@@ -0,0 +1,276 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
3
+ "architectures": [
4
+ "ParlerTTSForConditionalGeneration"
5
+ ],
6
+ "audio_encoder": {
7
+ "_name_or_path": "ylacombe/dac_44khZ_8kbps",
8
+ "add_cross_attention": false,
9
+ "architectures": [
10
+ "DACModel"
11
+ ],
12
+ "bad_words_ids": null,
13
+ "begin_suppress_tokens": null,
14
+ "bos_token_id": null,
15
+ "chunk_size_feed_forward": 0,
16
+ "codebook_size": 1024,
17
+ "cross_attention_hidden_size": null,
18
+ "decoder_start_token_id": null,
19
+ "diversity_penalty": 0.0,
20
+ "do_sample": false,
21
+ "early_stopping": false,
22
+ "encoder_no_repeat_ngram_size": 0,
23
+ "eos_token_id": null,
24
+ "exponential_decay_length_penalty": null,
25
+ "finetuning_task": null,
26
+ "forced_bos_token_id": null,
27
+ "forced_eos_token_id": null,
28
+ "frame_rate": 86,
29
+ "id2label": {
30
+ "0": "LABEL_0",
31
+ "1": "LABEL_1"
32
+ },
33
+ "is_decoder": false,
34
+ "is_encoder_decoder": false,
35
+ "label2id": {
36
+ "LABEL_0": 0,
37
+ "LABEL_1": 1
38
+ },
39
+ "latent_dim": 1024,
40
+ "length_penalty": 1.0,
41
+ "max_length": 20,
42
+ "min_length": 0,
43
+ "model_bitrate": 8,
44
+ "model_type": "dac",
45
+ "no_repeat_ngram_size": 0,
46
+ "num_beam_groups": 1,
47
+ "num_beams": 1,
48
+ "num_codebooks": 9,
49
+ "num_return_sequences": 1,
50
+ "output_attentions": false,
51
+ "output_hidden_states": false,
52
+ "output_scores": false,
53
+ "pad_token_id": null,
54
+ "prefix": null,
55
+ "problem_type": null,
56
+ "pruned_heads": {},
57
+ "remove_invalid_values": false,
58
+ "repetition_penalty": 1.0,
59
+ "return_dict": true,
60
+ "return_dict_in_generate": false,
61
+ "sampling_rate": 44100,
62
+ "sep_token_id": null,
63
+ "suppress_tokens": null,
64
+ "task_specific_params": null,
65
+ "temperature": 1.0,
66
+ "tf_legacy_loss": false,
67
+ "tie_encoder_decoder": false,
68
+ "tie_word_embeddings": true,
69
+ "tokenizer_class": null,
70
+ "top_k": 50,
71
+ "top_p": 1.0,
72
+ "torch_dtype": "float32",
73
+ "torchscript": false,
74
+ "typical_p": 1.0,
75
+ "use_bfloat16": false
76
+ },
77
+ "decoder": {
78
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
79
+ "activation_dropout": 0.0,
80
+ "activation_function": "gelu",
81
+ "add_cross_attention": true,
82
+ "architectures": [
83
+ "ParlerTTSForCausalLM"
84
+ ],
85
+ "attention_dropout": 0.0,
86
+ "bad_words_ids": null,
87
+ "begin_suppress_tokens": null,
88
+ "bos_token_id": 1025,
89
+ "chunk_size_feed_forward": 0,
90
+ "cross_attention_hidden_size": null,
91
+ "decoder_start_token_id": null,
92
+ "diversity_penalty": 0.0,
93
+ "do_sample": false,
94
+ "dropout": 0.1,
95
+ "early_stopping": false,
96
+ "encoder_no_repeat_ngram_size": 0,
97
+ "eos_token_id": 1024,
98
+ "exponential_decay_length_penalty": null,
99
+ "ffn_dim": 4096,
100
+ "finetuning_task": null,
101
+ "forced_bos_token_id": null,
102
+ "forced_eos_token_id": null,
103
+ "hidden_size": 1024,
104
+ "id2label": {
105
+ "0": "LABEL_0",
106
+ "1": "LABEL_1"
107
+ },
108
+ "initializer_factor": 0.02,
109
+ "is_decoder": true,
110
+ "is_encoder_decoder": false,
111
+ "label2id": {
112
+ "LABEL_0": 0,
113
+ "LABEL_1": 1
114
+ },
115
+ "layerdrop": 0.0,
116
+ "length_penalty": 1.0,
117
+ "max_length": 20,
118
+ "max_position_embeddings": 4096,
119
+ "min_length": 0,
120
+ "model_type": "parler_tts_decoder",
121
+ "no_repeat_ngram_size": 0,
122
+ "num_attention_heads": 16,
123
+ "num_beam_groups": 1,
124
+ "num_beams": 1,
125
+ "num_codebooks": 9,
126
+ "num_hidden_layers": 24,
127
+ "num_return_sequences": 1,
128
+ "output_attentions": false,
129
+ "output_hidden_states": false,
130
+ "output_scores": false,
131
+ "pad_token_id": 1024,
132
+ "prefix": null,
133
+ "problem_type": null,
134
+ "pruned_heads": {},
135
+ "remove_invalid_values": false,
136
+ "repetition_penalty": 1.0,
137
+ "return_dict": true,
138
+ "return_dict_in_generate": false,
139
+ "scale_embedding": false,
140
+ "sep_token_id": null,
141
+ "suppress_tokens": null,
142
+ "task_specific_params": null,
143
+ "temperature": 1.0,
144
+ "tf_legacy_loss": false,
145
+ "tie_encoder_decoder": false,
146
+ "tie_word_embeddings": false,
147
+ "tokenizer_class": null,
148
+ "top_k": 50,
149
+ "top_p": 1.0,
150
+ "torch_dtype": "float32",
151
+ "torchscript": false,
152
+ "typical_p": 1.0,
153
+ "use_bfloat16": false,
154
+ "use_cache": true,
155
+ "vocab_size": 1088
156
+ },
157
+ "decoder_start_token_id": 1025,
158
+ "is_encoder_decoder": true,
159
+ "model_type": "parler_tts",
160
+ "pad_token_id": 1024,
161
+ "text_encoder": {
162
+ "_name_or_path": "google/flan-t5-base",
163
+ "add_cross_attention": false,
164
+ "architectures": [
165
+ "T5ForConditionalGeneration"
166
+ ],
167
+ "bad_words_ids": null,
168
+ "begin_suppress_tokens": null,
169
+ "bos_token_id": null,
170
+ "chunk_size_feed_forward": 0,
171
+ "classifier_dropout": 0.0,
172
+ "cross_attention_hidden_size": null,
173
+ "d_ff": 2048,
174
+ "d_kv": 64,
175
+ "d_model": 768,
176
+ "decoder_start_token_id": 0,
177
+ "dense_act_fn": "gelu_new",
178
+ "diversity_penalty": 0.0,
179
+ "do_sample": false,
180
+ "dropout_rate": 0.1,
181
+ "early_stopping": false,
182
+ "encoder_no_repeat_ngram_size": 0,
183
+ "eos_token_id": 1,
184
+ "exponential_decay_length_penalty": null,
185
+ "feed_forward_proj": "gated-gelu",
186
+ "finetuning_task": null,
187
+ "forced_bos_token_id": null,
188
+ "forced_eos_token_id": null,
189
+ "id2label": {
190
+ "0": "LABEL_0",
191
+ "1": "LABEL_1"
192
+ },
193
+ "initializer_factor": 1.0,
194
+ "is_decoder": false,
195
+ "is_encoder_decoder": true,
196
+ "is_gated_act": true,
197
+ "label2id": {
198
+ "LABEL_0": 0,
199
+ "LABEL_1": 1
200
+ },
201
+ "layer_norm_epsilon": 1e-06,
202
+ "length_penalty": 1.0,
203
+ "max_length": 20,
204
+ "min_length": 0,
205
+ "model_type": "t5",
206
+ "n_positions": 512,
207
+ "no_repeat_ngram_size": 0,
208
+ "num_beam_groups": 1,
209
+ "num_beams": 1,
210
+ "num_decoder_layers": 12,
211
+ "num_heads": 12,
212
+ "num_layers": 12,
213
+ "num_return_sequences": 1,
214
+ "output_attentions": false,
215
+ "output_hidden_states": false,
216
+ "output_past": true,
217
+ "output_scores": false,
218
+ "pad_token_id": 0,
219
+ "prefix": null,
220
+ "problem_type": null,
221
+ "pruned_heads": {},
222
+ "relative_attention_max_distance": 128,
223
+ "relative_attention_num_buckets": 32,
224
+ "remove_invalid_values": false,
225
+ "repetition_penalty": 1.0,
226
+ "return_dict": true,
227
+ "return_dict_in_generate": false,
228
+ "sep_token_id": null,
229
+ "suppress_tokens": null,
230
+ "task_specific_params": {
231
+ "summarization": {
232
+ "early_stopping": true,
233
+ "length_penalty": 2.0,
234
+ "max_length": 200,
235
+ "min_length": 30,
236
+ "no_repeat_ngram_size": 3,
237
+ "num_beams": 4,
238
+ "prefix": "summarize: "
239
+ },
240
+ "translation_en_to_de": {
241
+ "early_stopping": true,
242
+ "max_length": 300,
243
+ "num_beams": 4,
244
+ "prefix": "translate English to German: "
245
+ },
246
+ "translation_en_to_fr": {
247
+ "early_stopping": true,
248
+ "max_length": 300,
249
+ "num_beams": 4,
250
+ "prefix": "translate English to French: "
251
+ },
252
+ "translation_en_to_ro": {
253
+ "early_stopping": true,
254
+ "max_length": 300,
255
+ "num_beams": 4,
256
+ "prefix": "translate English to Romanian: "
257
+ }
258
+ },
259
+ "temperature": 1.0,
260
+ "tf_legacy_loss": false,
261
+ "tie_encoder_decoder": false,
262
+ "tie_word_embeddings": false,
263
+ "tokenizer_class": null,
264
+ "top_k": 50,
265
+ "top_p": 1.0,
266
+ "torch_dtype": null,
267
+ "torchscript": false,
268
+ "typical_p": 1.0,
269
+ "use_bfloat16": false,
270
+ "use_cache": true,
271
+ "vocab_size": 32128
272
+ },
273
+ "torch_dtype": "float32",
274
+ "transformers_version": "4.41.0.dev0",
275
+ "vocab_size": 32128
276
+ }
checkpoint-1280-epoch-7/generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1025,
4
+ "decoder_start_token_id": 1025,
5
+ "do_sample": true,
6
+ "eos_token_id": 1024,
7
+ "guidance_scale": 1.0,
8
+ "max_length": 2580,
9
+ "min_new_tokens": 50,
10
+ "pad_token_id": 1024,
11
+ "transformers_version": "4.41.0.dev0"
12
+ }
checkpoint-1280-epoch-7/optimizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0a22ddbb565705b39d5bb86e2494d49f115f58b308e1207e78b98dc9c8456295
3
  size 3652763351
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:517aa375d2eebc8426f582764f5bda6d1563fc7a7bdd1294abd4b34a389151a5
3
  size 3652763351
checkpoint-1280-epoch-7/pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ee2430b336c101ba8a0bf42fb405360191213a0085e9e16d048026292650a0a0
3
  size 2588462170
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6858c2439f04655903a513f549f28338cbe76fd997db08fb9221553265014a23
3
  size 2588462170
checkpoint-1280-epoch-7/random_states_0.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2858b9303b7791ffdd884017cf45488bdde5e7e57da3915a2f8642af64676749
3
- size 14408
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:983b1223b6f4d653f99a5b3d5892e6f89c685ecf06caf30bd1f297a89e33b0e0
3
+ size 15320
checkpoint-640-epoch-3/config.json ADDED
@@ -0,0 +1,276 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
3
+ "architectures": [
4
+ "ParlerTTSForConditionalGeneration"
5
+ ],
6
+ "audio_encoder": {
7
+ "_name_or_path": "ylacombe/dac_44khZ_8kbps",
8
+ "add_cross_attention": false,
9
+ "architectures": [
10
+ "DACModel"
11
+ ],
12
+ "bad_words_ids": null,
13
+ "begin_suppress_tokens": null,
14
+ "bos_token_id": null,
15
+ "chunk_size_feed_forward": 0,
16
+ "codebook_size": 1024,
17
+ "cross_attention_hidden_size": null,
18
+ "decoder_start_token_id": null,
19
+ "diversity_penalty": 0.0,
20
+ "do_sample": false,
21
+ "early_stopping": false,
22
+ "encoder_no_repeat_ngram_size": 0,
23
+ "eos_token_id": null,
24
+ "exponential_decay_length_penalty": null,
25
+ "finetuning_task": null,
26
+ "forced_bos_token_id": null,
27
+ "forced_eos_token_id": null,
28
+ "frame_rate": 86,
29
+ "id2label": {
30
+ "0": "LABEL_0",
31
+ "1": "LABEL_1"
32
+ },
33
+ "is_decoder": false,
34
+ "is_encoder_decoder": false,
35
+ "label2id": {
36
+ "LABEL_0": 0,
37
+ "LABEL_1": 1
38
+ },
39
+ "latent_dim": 1024,
40
+ "length_penalty": 1.0,
41
+ "max_length": 20,
42
+ "min_length": 0,
43
+ "model_bitrate": 8,
44
+ "model_type": "dac",
45
+ "no_repeat_ngram_size": 0,
46
+ "num_beam_groups": 1,
47
+ "num_beams": 1,
48
+ "num_codebooks": 9,
49
+ "num_return_sequences": 1,
50
+ "output_attentions": false,
51
+ "output_hidden_states": false,
52
+ "output_scores": false,
53
+ "pad_token_id": null,
54
+ "prefix": null,
55
+ "problem_type": null,
56
+ "pruned_heads": {},
57
+ "remove_invalid_values": false,
58
+ "repetition_penalty": 1.0,
59
+ "return_dict": true,
60
+ "return_dict_in_generate": false,
61
+ "sampling_rate": 44100,
62
+ "sep_token_id": null,
63
+ "suppress_tokens": null,
64
+ "task_specific_params": null,
65
+ "temperature": 1.0,
66
+ "tf_legacy_loss": false,
67
+ "tie_encoder_decoder": false,
68
+ "tie_word_embeddings": true,
69
+ "tokenizer_class": null,
70
+ "top_k": 50,
71
+ "top_p": 1.0,
72
+ "torch_dtype": "float32",
73
+ "torchscript": false,
74
+ "typical_p": 1.0,
75
+ "use_bfloat16": false
76
+ },
77
+ "decoder": {
78
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
79
+ "activation_dropout": 0.0,
80
+ "activation_function": "gelu",
81
+ "add_cross_attention": true,
82
+ "architectures": [
83
+ "ParlerTTSForCausalLM"
84
+ ],
85
+ "attention_dropout": 0.0,
86
+ "bad_words_ids": null,
87
+ "begin_suppress_tokens": null,
88
+ "bos_token_id": 1025,
89
+ "chunk_size_feed_forward": 0,
90
+ "cross_attention_hidden_size": null,
91
+ "decoder_start_token_id": null,
92
+ "diversity_penalty": 0.0,
93
+ "do_sample": false,
94
+ "dropout": 0.1,
95
+ "early_stopping": false,
96
+ "encoder_no_repeat_ngram_size": 0,
97
+ "eos_token_id": 1024,
98
+ "exponential_decay_length_penalty": null,
99
+ "ffn_dim": 4096,
100
+ "finetuning_task": null,
101
+ "forced_bos_token_id": null,
102
+ "forced_eos_token_id": null,
103
+ "hidden_size": 1024,
104
+ "id2label": {
105
+ "0": "LABEL_0",
106
+ "1": "LABEL_1"
107
+ },
108
+ "initializer_factor": 0.02,
109
+ "is_decoder": true,
110
+ "is_encoder_decoder": false,
111
+ "label2id": {
112
+ "LABEL_0": 0,
113
+ "LABEL_1": 1
114
+ },
115
+ "layerdrop": 0.0,
116
+ "length_penalty": 1.0,
117
+ "max_length": 20,
118
+ "max_position_embeddings": 4096,
119
+ "min_length": 0,
120
+ "model_type": "parler_tts_decoder",
121
+ "no_repeat_ngram_size": 0,
122
+ "num_attention_heads": 16,
123
+ "num_beam_groups": 1,
124
+ "num_beams": 1,
125
+ "num_codebooks": 9,
126
+ "num_hidden_layers": 24,
127
+ "num_return_sequences": 1,
128
+ "output_attentions": false,
129
+ "output_hidden_states": false,
130
+ "output_scores": false,
131
+ "pad_token_id": 1024,
132
+ "prefix": null,
133
+ "problem_type": null,
134
+ "pruned_heads": {},
135
+ "remove_invalid_values": false,
136
+ "repetition_penalty": 1.0,
137
+ "return_dict": true,
138
+ "return_dict_in_generate": false,
139
+ "scale_embedding": false,
140
+ "sep_token_id": null,
141
+ "suppress_tokens": null,
142
+ "task_specific_params": null,
143
+ "temperature": 1.0,
144
+ "tf_legacy_loss": false,
145
+ "tie_encoder_decoder": false,
146
+ "tie_word_embeddings": false,
147
+ "tokenizer_class": null,
148
+ "top_k": 50,
149
+ "top_p": 1.0,
150
+ "torch_dtype": "float32",
151
+ "torchscript": false,
152
+ "typical_p": 1.0,
153
+ "use_bfloat16": false,
154
+ "use_cache": true,
155
+ "vocab_size": 1088
156
+ },
157
+ "decoder_start_token_id": 1025,
158
+ "is_encoder_decoder": true,
159
+ "model_type": "parler_tts",
160
+ "pad_token_id": 1024,
161
+ "text_encoder": {
162
+ "_name_or_path": "google/flan-t5-base",
163
+ "add_cross_attention": false,
164
+ "architectures": [
165
+ "T5ForConditionalGeneration"
166
+ ],
167
+ "bad_words_ids": null,
168
+ "begin_suppress_tokens": null,
169
+ "bos_token_id": null,
170
+ "chunk_size_feed_forward": 0,
171
+ "classifier_dropout": 0.0,
172
+ "cross_attention_hidden_size": null,
173
+ "d_ff": 2048,
174
+ "d_kv": 64,
175
+ "d_model": 768,
176
+ "decoder_start_token_id": 0,
177
+ "dense_act_fn": "gelu_new",
178
+ "diversity_penalty": 0.0,
179
+ "do_sample": false,
180
+ "dropout_rate": 0.1,
181
+ "early_stopping": false,
182
+ "encoder_no_repeat_ngram_size": 0,
183
+ "eos_token_id": 1,
184
+ "exponential_decay_length_penalty": null,
185
+ "feed_forward_proj": "gated-gelu",
186
+ "finetuning_task": null,
187
+ "forced_bos_token_id": null,
188
+ "forced_eos_token_id": null,
189
+ "id2label": {
190
+ "0": "LABEL_0",
191
+ "1": "LABEL_1"
192
+ },
193
+ "initializer_factor": 1.0,
194
+ "is_decoder": false,
195
+ "is_encoder_decoder": true,
196
+ "is_gated_act": true,
197
+ "label2id": {
198
+ "LABEL_0": 0,
199
+ "LABEL_1": 1
200
+ },
201
+ "layer_norm_epsilon": 1e-06,
202
+ "length_penalty": 1.0,
203
+ "max_length": 20,
204
+ "min_length": 0,
205
+ "model_type": "t5",
206
+ "n_positions": 512,
207
+ "no_repeat_ngram_size": 0,
208
+ "num_beam_groups": 1,
209
+ "num_beams": 1,
210
+ "num_decoder_layers": 12,
211
+ "num_heads": 12,
212
+ "num_layers": 12,
213
+ "num_return_sequences": 1,
214
+ "output_attentions": false,
215
+ "output_hidden_states": false,
216
+ "output_past": true,
217
+ "output_scores": false,
218
+ "pad_token_id": 0,
219
+ "prefix": null,
220
+ "problem_type": null,
221
+ "pruned_heads": {},
222
+ "relative_attention_max_distance": 128,
223
+ "relative_attention_num_buckets": 32,
224
+ "remove_invalid_values": false,
225
+ "repetition_penalty": 1.0,
226
+ "return_dict": true,
227
+ "return_dict_in_generate": false,
228
+ "sep_token_id": null,
229
+ "suppress_tokens": null,
230
+ "task_specific_params": {
231
+ "summarization": {
232
+ "early_stopping": true,
233
+ "length_penalty": 2.0,
234
+ "max_length": 200,
235
+ "min_length": 30,
236
+ "no_repeat_ngram_size": 3,
237
+ "num_beams": 4,
238
+ "prefix": "summarize: "
239
+ },
240
+ "translation_en_to_de": {
241
+ "early_stopping": true,
242
+ "max_length": 300,
243
+ "num_beams": 4,
244
+ "prefix": "translate English to German: "
245
+ },
246
+ "translation_en_to_fr": {
247
+ "early_stopping": true,
248
+ "max_length": 300,
249
+ "num_beams": 4,
250
+ "prefix": "translate English to French: "
251
+ },
252
+ "translation_en_to_ro": {
253
+ "early_stopping": true,
254
+ "max_length": 300,
255
+ "num_beams": 4,
256
+ "prefix": "translate English to Romanian: "
257
+ }
258
+ },
259
+ "temperature": 1.0,
260
+ "tf_legacy_loss": false,
261
+ "tie_encoder_decoder": false,
262
+ "tie_word_embeddings": false,
263
+ "tokenizer_class": null,
264
+ "top_k": 50,
265
+ "top_p": 1.0,
266
+ "torch_dtype": null,
267
+ "torchscript": false,
268
+ "typical_p": 1.0,
269
+ "use_bfloat16": false,
270
+ "use_cache": true,
271
+ "vocab_size": 32128
272
+ },
273
+ "torch_dtype": "float32",
274
+ "transformers_version": "4.41.0.dev0",
275
+ "vocab_size": 32128
276
+ }
checkpoint-640-epoch-3/generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1025,
4
+ "decoder_start_token_id": 1025,
5
+ "do_sample": true,
6
+ "eos_token_id": 1024,
7
+ "guidance_scale": 1.0,
8
+ "max_length": 2580,
9
+ "min_new_tokens": 50,
10
+ "pad_token_id": 1024,
11
+ "transformers_version": "4.41.0.dev0"
12
+ }
checkpoint-640-epoch-3/optimizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7676577b9a00401b141b802905c27c0e36fbbfb39be9fe4ab5c416440d084b97
3
  size 3652763351
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d44b46c3d88fc3095d5a7059bf0fe3b6277c435dacd898d1b67a0a63081172a7
3
  size 3652763351
checkpoint-640-epoch-3/pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c7dc0f7aee183a683eaf160c4cb24b324321cd016f5d56da009862f2436b65f3
3
  size 2588462170
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a8bb393767b101298c2f0341346e25753532c254862b1fe8fee4cab4fcb0795
3
  size 2588462170
checkpoint-640-epoch-3/random_states_0.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:256df18c89f256f371cd05fd9c75d1352980db04c485621dcf7c689ad1f421d6
3
- size 14408
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e7dcd9631675ac60c3a95317663b1450b8a9ec0ee5e88f502e1c965f4b3240d
3
+ size 15384
checkpoint-800-epoch-4/config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "parler-tts/parler_tts_mini_v0.1",
3
  "architectures": [
4
  "ParlerTTSForConditionalGeneration"
5
  ],
 
1
  {
2
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
3
  "architectures": [
4
  "ParlerTTSForConditionalGeneration"
5
  ],
checkpoint-800-epoch-4/optimizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f1be446c0dbd4b6c8512cfca027c114f7ad54b35c3d4076c76436c03f8c7e8b1
3
  size 3652763351
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6d90b139c9ed7c7d7bc04e54d80d61d87261ef7dfa24df06bea51df0b39b8ab
3
  size 3652763351
checkpoint-800-epoch-4/pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:241896d2452407ad54eeff936427883c44a2e230e23b765744eacfd779d02213
3
  size 2588462170
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5679d7147314711051361c581d412fc9fd275d2087edad280b8b1896cde2cc2
3
  size 2588462170
checkpoint-800-epoch-4/random_states_0.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:62803c1cff666cd37d714e1f111d0780398deb4f04ea085c8ed15a8efecea7f5
3
- size 14344
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:914a3adfb74f00c1341c340b3a25428f6f2b1d9356c903324b6fd116d6de5b41
3
+ size 15384
checkpoint-960-epoch-5/config.json ADDED
@@ -0,0 +1,276 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/training-400M-punctuated-v2/",
3
+ "architectures": [
4
+ "ParlerTTSForConditionalGeneration"
5
+ ],
6
+ "audio_encoder": {
7
+ "_name_or_path": "ylacombe/dac_44khZ_8kbps",
8
+ "add_cross_attention": false,
9
+ "architectures": [
10
+ "DACModel"
11
+ ],
12
+ "bad_words_ids": null,
13
+ "begin_suppress_tokens": null,
14
+ "bos_token_id": null,
15
+ "chunk_size_feed_forward": 0,
16
+ "codebook_size": 1024,
17
+ "cross_attention_hidden_size": null,
18
+ "decoder_start_token_id": null,
19
+ "diversity_penalty": 0.0,
20
+ "do_sample": false,
21
+ "early_stopping": false,
22
+ "encoder_no_repeat_ngram_size": 0,
23
+ "eos_token_id": null,
24
+ "exponential_decay_length_penalty": null,
25
+ "finetuning_task": null,
26
+ "forced_bos_token_id": null,
27
+ "forced_eos_token_id": null,
28
+ "frame_rate": 86,
29
+ "id2label": {
30
+ "0": "LABEL_0",
31
+ "1": "LABEL_1"
32
+ },
33
+ "is_decoder": false,
34
+ "is_encoder_decoder": false,
35
+ "label2id": {
36
+ "LABEL_0": 0,
37
+ "LABEL_1": 1
38
+ },
39
+ "latent_dim": 1024,
40
+ "length_penalty": 1.0,
41
+ "max_length": 20,
42
+ "min_length": 0,
43
+ "model_bitrate": 8,
44
+ "model_type": "dac",
45
+ "no_repeat_ngram_size": 0,
46
+ "num_beam_groups": 1,
47
+ "num_beams": 1,
48
+ "num_codebooks": 9,
49
+ "num_return_sequences": 1,
50
+ "output_attentions": false,
51
+ "output_hidden_states": false,
52
+ "output_scores": false,
53
+ "pad_token_id": null,
54
+ "prefix": null,
55
+ "problem_type": null,
56
+ "pruned_heads": {},
57
+ "remove_invalid_values": false,
58
+ "repetition_penalty": 1.0,
59
+ "return_dict": true,
60
+ "return_dict_in_generate": false,
61
+ "sampling_rate": 44100,
62
+ "sep_token_id": null,
63
+ "suppress_tokens": null,
64
+ "task_specific_params": null,
65
+ "temperature": 1.0,
66
+ "tf_legacy_loss": false,
67
+ "tie_encoder_decoder": false,
68
+ "tie_word_embeddings": true,
69
+ "tokenizer_class": null,
70
+ "top_k": 50,
71
+ "top_p": 1.0,
72
+ "torch_dtype": "float32",
73
+ "torchscript": false,
74
+ "typical_p": 1.0,
75
+ "use_bfloat16": false
76
+ },
77
+ "decoder": {
78
+ "_name_or_path": "/fsx/yoach/tmp/artefacts/decoder_400M/",
79
+ "activation_dropout": 0.0,
80
+ "activation_function": "gelu",
81
+ "add_cross_attention": true,
82
+ "architectures": [
83
+ "ParlerTTSForCausalLM"
84
+ ],
85
+ "attention_dropout": 0.0,
86
+ "bad_words_ids": null,
87
+ "begin_suppress_tokens": null,
88
+ "bos_token_id": 1025,
89
+ "chunk_size_feed_forward": 0,
90
+ "cross_attention_hidden_size": null,
91
+ "decoder_start_token_id": null,
92
+ "diversity_penalty": 0.0,
93
+ "do_sample": false,
94
+ "dropout": 0.1,
95
+ "early_stopping": false,
96
+ "encoder_no_repeat_ngram_size": 0,
97
+ "eos_token_id": 1024,
98
+ "exponential_decay_length_penalty": null,
99
+ "ffn_dim": 4096,
100
+ "finetuning_task": null,
101
+ "forced_bos_token_id": null,
102
+ "forced_eos_token_id": null,
103
+ "hidden_size": 1024,
104
+ "id2label": {
105
+ "0": "LABEL_0",
106
+ "1": "LABEL_1"
107
+ },
108
+ "initializer_factor": 0.02,
109
+ "is_decoder": true,
110
+ "is_encoder_decoder": false,
111
+ "label2id": {
112
+ "LABEL_0": 0,
113
+ "LABEL_1": 1
114
+ },
115
+ "layerdrop": 0.0,
116
+ "length_penalty": 1.0,
117
+ "max_length": 20,
118
+ "max_position_embeddings": 4096,
119
+ "min_length": 0,
120
+ "model_type": "parler_tts_decoder",
121
+ "no_repeat_ngram_size": 0,
122
+ "num_attention_heads": 16,
123
+ "num_beam_groups": 1,
124
+ "num_beams": 1,
125
+ "num_codebooks": 9,
126
+ "num_hidden_layers": 24,
127
+ "num_return_sequences": 1,
128
+ "output_attentions": false,
129
+ "output_hidden_states": false,
130
+ "output_scores": false,
131
+ "pad_token_id": 1024,
132
+ "prefix": null,
133
+ "problem_type": null,
134
+ "pruned_heads": {},
135
+ "remove_invalid_values": false,
136
+ "repetition_penalty": 1.0,
137
+ "return_dict": true,
138
+ "return_dict_in_generate": false,
139
+ "scale_embedding": false,
140
+ "sep_token_id": null,
141
+ "suppress_tokens": null,
142
+ "task_specific_params": null,
143
+ "temperature": 1.0,
144
+ "tf_legacy_loss": false,
145
+ "tie_encoder_decoder": false,
146
+ "tie_word_embeddings": false,
147
+ "tokenizer_class": null,
148
+ "top_k": 50,
149
+ "top_p": 1.0,
150
+ "torch_dtype": "float32",
151
+ "torchscript": false,
152
+ "typical_p": 1.0,
153
+ "use_bfloat16": false,
154
+ "use_cache": true,
155
+ "vocab_size": 1088
156
+ },
157
+ "decoder_start_token_id": 1025,
158
+ "is_encoder_decoder": true,
159
+ "model_type": "parler_tts",
160
+ "pad_token_id": 1024,
161
+ "text_encoder": {
162
+ "_name_or_path": "google/flan-t5-base",
163
+ "add_cross_attention": false,
164
+ "architectures": [
165
+ "T5ForConditionalGeneration"
166
+ ],
167
+ "bad_words_ids": null,
168
+ "begin_suppress_tokens": null,
169
+ "bos_token_id": null,
170
+ "chunk_size_feed_forward": 0,
171
+ "classifier_dropout": 0.0,
172
+ "cross_attention_hidden_size": null,
173
+ "d_ff": 2048,
174
+ "d_kv": 64,
175
+ "d_model": 768,
176
+ "decoder_start_token_id": 0,
177
+ "dense_act_fn": "gelu_new",
178
+ "diversity_penalty": 0.0,
179
+ "do_sample": false,
180
+ "dropout_rate": 0.1,
181
+ "early_stopping": false,
182
+ "encoder_no_repeat_ngram_size": 0,
183
+ "eos_token_id": 1,
184
+ "exponential_decay_length_penalty": null,
185
+ "feed_forward_proj": "gated-gelu",
186
+ "finetuning_task": null,
187
+ "forced_bos_token_id": null,
188
+ "forced_eos_token_id": null,
189
+ "id2label": {
190
+ "0": "LABEL_0",
191
+ "1": "LABEL_1"
192
+ },
193
+ "initializer_factor": 1.0,
194
+ "is_decoder": false,
195
+ "is_encoder_decoder": true,
196
+ "is_gated_act": true,
197
+ "label2id": {
198
+ "LABEL_0": 0,
199
+ "LABEL_1": 1
200
+ },
201
+ "layer_norm_epsilon": 1e-06,
202
+ "length_penalty": 1.0,
203
+ "max_length": 20,
204
+ "min_length": 0,
205
+ "model_type": "t5",
206
+ "n_positions": 512,
207
+ "no_repeat_ngram_size": 0,
208
+ "num_beam_groups": 1,
209
+ "num_beams": 1,
210
+ "num_decoder_layers": 12,
211
+ "num_heads": 12,
212
+ "num_layers": 12,
213
+ "num_return_sequences": 1,
214
+ "output_attentions": false,
215
+ "output_hidden_states": false,
216
+ "output_past": true,
217
+ "output_scores": false,
218
+ "pad_token_id": 0,
219
+ "prefix": null,
220
+ "problem_type": null,
221
+ "pruned_heads": {},
222
+ "relative_attention_max_distance": 128,
223
+ "relative_attention_num_buckets": 32,
224
+ "remove_invalid_values": false,
225
+ "repetition_penalty": 1.0,
226
+ "return_dict": true,
227
+ "return_dict_in_generate": false,
228
+ "sep_token_id": null,
229
+ "suppress_tokens": null,
230
+ "task_specific_params": {
231
+ "summarization": {
232
+ "early_stopping": true,
233
+ "length_penalty": 2.0,
234
+ "max_length": 200,
235
+ "min_length": 30,
236
+ "no_repeat_ngram_size": 3,
237
+ "num_beams": 4,
238
+ "prefix": "summarize: "
239
+ },
240
+ "translation_en_to_de": {
241
+ "early_stopping": true,
242
+ "max_length": 300,
243
+ "num_beams": 4,
244
+ "prefix": "translate English to German: "
245
+ },
246
+ "translation_en_to_fr": {
247
+ "early_stopping": true,
248
+ "max_length": 300,
249
+ "num_beams": 4,
250
+ "prefix": "translate English to French: "
251
+ },
252
+ "translation_en_to_ro": {
253
+ "early_stopping": true,
254
+ "max_length": 300,
255
+ "num_beams": 4,
256
+ "prefix": "translate English to Romanian: "
257
+ }
258
+ },
259
+ "temperature": 1.0,
260
+ "tf_legacy_loss": false,
261
+ "tie_encoder_decoder": false,
262
+ "tie_word_embeddings": false,
263
+ "tokenizer_class": null,
264
+ "top_k": 50,
265
+ "top_p": 1.0,
266
+ "torch_dtype": null,
267
+ "torchscript": false,
268
+ "typical_p": 1.0,
269
+ "use_bfloat16": false,
270
+ "use_cache": true,
271
+ "vocab_size": 32128
272
+ },
273
+ "torch_dtype": "float32",
274
+ "transformers_version": "4.41.0.dev0",
275
+ "vocab_size": 32128
276
+ }
checkpoint-960-epoch-5/generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1025,
4
+ "decoder_start_token_id": 1025,
5
+ "do_sample": true,
6
+ "eos_token_id": 1024,
7
+ "guidance_scale": 1.0,
8
+ "max_length": 2580,
9
+ "min_new_tokens": 50,
10
+ "pad_token_id": 1024,
11
+ "transformers_version": "4.41.0.dev0"
12
+ }
checkpoint-960-epoch-5/optimizer.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fd7754aeea66398a842117378c23e420c4b9bfd2ee4e1c15ef940d894a0b6e35
3
  size 3652763351
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfd6e345787ff4520cc8d8b7b7efb53d67a530395b17adbf2f3af7568eef715d
3
  size 3652763351
checkpoint-960-epoch-5/pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:52b3e76b7fde9e53c910cfcb44db88274cc60c24569f3f51786d34fd6b06e9f1
3
  size 2588462170
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d238cad039497527495bc020c99c2d6a1280f5767f25d1f7f720db1feeedf8cc
3
  size 2588462170
checkpoint-960-epoch-5/random_states_0.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:43e8de9bf1af8ea2926a2a17dba99400420a9bbdbc76642417f1086a83a5227e
3
- size 14344
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:986938c8f8a39798c83a0eb882227050b61879583c253c88f2f2b6ea3138464c
3
+ size 15384
finetuning_config.json CHANGED
@@ -49,6 +49,7 @@
49
  "save_strategy": "epoch",
50
  "evaluation_strategy": "epoch",
51
  "save_total_limit": 5,
52
- "group_by_length": true
 
53
  }
54
 
 
49
  "save_strategy": "epoch",
50
  "evaluation_strategy": "epoch",
51
  "save_total_limit": 5,
52
+ "group_by_length": true,
53
+ "push_to_hub": true
54
  }
55
 
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:45937d8dfa9f232e79f76f536602301dde6f2b5be7200414d69ff07bff1228c3
3
  size 2588215392
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab690bfb1b231f648dc3c141200f7a753fe4c84b4e3e30307f4491bccd5e7ee7
3
  size 2588215392
run_parler_tts_training.py CHANGED
@@ -942,6 +942,7 @@ def main():
942
  columns_to_keep = {
943
  "target_audio_column_name": data_args.target_audio_column_name,
944
  "prompt_column_name": data_args.prompt_column_name,
 
945
  }
946
  if data_args.description_column_name is not None:
947
  columns_to_keep["description_column_name"] = data_args.description_column_name
@@ -1020,6 +1021,11 @@ def main():
1020
  }
1021
  )
1022
 
 
 
 
 
 
1023
  # create model
1024
  model = ParlerTTSForConditionalGeneration.from_pretrained(
1025
  model_args.model_name_or_path,
@@ -1028,6 +1034,7 @@ def main():
1028
  token=data_args.token,
1029
  trust_remote_code=data_args.trust_remote_code,
1030
  )
 
1031
 
1032
  # enable gradient checkpointing if necessary
1033
  if training_args.gradient_checkpointing:
@@ -1619,6 +1626,8 @@ def main():
1619
  # https://github.com/huggingface/transformers/issues/27293#issuecomment-1872560074
1620
  accelerator.save_state(output_dir=intermediate_dir, safe_serialization=False)
1621
  accelerator.wait_for_everyone()
 
 
1622
  if accelerator.is_main_process:
1623
  rotate_checkpoints(training_args.save_total_limit, output_dir=training_args.output_dir)
1624
 
@@ -1761,3 +1770,4 @@ def main():
1761
  if __name__ == "__main__":
1762
  set_start_method("spawn")
1763
  main()
 
 
942
  columns_to_keep = {
943
  "target_audio_column_name": data_args.target_audio_column_name,
944
  "prompt_column_name": data_args.prompt_column_name,
945
+ "style": "style",
946
  }
947
  if data_args.description_column_name is not None:
948
  columns_to_keep["description_column_name"] = data_args.description_column_name
 
1021
  }
1022
  )
1023
 
1024
+ def remove_whispering(style):
1025
+ return style != "whisper"
1026
+
1027
+ raw_datasets = raw_datasets.filter(remove_whispering)
1028
+
1029
  # create model
1030
  model = ParlerTTSForConditionalGeneration.from_pretrained(
1031
  model_args.model_name_or_path,
 
1034
  token=data_args.token,
1035
  trust_remote_code=data_args.trust_remote_code,
1036
  )
1037
+ generation_config = model.generation_config
1038
 
1039
  # enable gradient checkpointing if necessary
1040
  if training_args.gradient_checkpointing:
 
1626
  # https://github.com/huggingface/transformers/issues/27293#issuecomment-1872560074
1627
  accelerator.save_state(output_dir=intermediate_dir, safe_serialization=False)
1628
  accelerator.wait_for_everyone()
1629
+ config.save_pretrained(intermediate_dir)
1630
+ generation_config.save_pretrained(intermediate_dir)
1631
  if accelerator.is_main_process:
1632
  rotate_checkpoints(training_args.save_total_limit, output_dir=training_args.output_dir)
1633
 
 
1770
  if __name__ == "__main__":
1771
  set_start_method("spawn")
1772
  main()
1773
+