sanchit-gandhi commited on
Commit
1ef1ef1
·
verified ·
1 Parent(s): bf791d4

remove whispering

Browse files
.gitattributes CHANGED
@@ -33,18 +33,3 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_4899e0da4615e883ad13.wav filter=lfs diff=lfs merge=lfs -text
37
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_492597073098578f0605.wav filter=lfs diff=lfs merge=lfs -text
38
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_67d9409a306e3614ec3f.wav filter=lfs diff=lfs merge=lfs -text
39
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_7c47ba927ac118ffaacc.wav filter=lfs diff=lfs merge=lfs -text
40
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_89ad32d31f3e70178cc1.wav filter=lfs diff=lfs merge=lfs -text
41
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_cd644667186ae0518a3c.wav filter=lfs diff=lfs merge=lfs -text
42
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_381_f7405ef7b645b3265477.wav filter=lfs diff=lfs merge=lfs -text
43
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_09db3d4c95c46a3bae02.wav filter=lfs diff=lfs merge=lfs -text
44
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_3349ee3cb71e3d789f00.wav filter=lfs diff=lfs merge=lfs -text
45
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_63964662df5bc7731b7a.wav filter=lfs diff=lfs merge=lfs -text
46
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_7ab4dededd2d3a052def.wav filter=lfs diff=lfs merge=lfs -text
47
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_b8be5a72ea66619af007.wav filter=lfs diff=lfs merge=lfs -text
48
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_d1c21fd700287f9f3109.wav filter=lfs diff=lfs merge=lfs -text
49
- wandb/run-20240513_205249-qaoje1x9/files/media/audio/Speech[[:space:]]samples/eval_762_e8feadb7228a5b136ce8.wav filter=lfs diff=lfs merge=lfs -text
50
- wandb/run-20240513_205249-qaoje1x9/run-qaoje1x9.wandb filter=lfs diff=lfs merge=lfs -text
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ wandb
checkpoint-1022-epoch-1/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-1022-epoch-1/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-1022-epoch-1/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78d33dad3796bd4b6db702a974f4884cd226c97de0ac5dfc991f5d71fc773d8e
3
+ size 2588462170
checkpoint-1022-epoch-1/random_states_0.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9de7a3178f60fc1dcd22d8eb17eea1898161f2c178c851bb85dcfdecc7c3b237
3
+ size 14408
checkpoint-1533-epoch-2/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-1533-epoch-2/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-1533-epoch-2/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a284790562434becee241ef9c77144317612d1af981342dfe871c99a864ad59
3
+ size 2588462170
checkpoint-1533-epoch-2/random_states_0.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c04544dbbe1a0d1c5f5e2ad843fcc8cfe3929791a91d7977da432179b81bfff3
3
+ size 14408
checkpoint-2044-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-2044-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-2044-epoch-3/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b008dd2c3d8c37830fe88b4d67a71b11dd6e73a3e6e4074d5c900771ab119037
3
+ size 2588462170
checkpoint-2044-epoch-3/random_states_0.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d64372b61b34e033f9049f7382e17cf33dd59798d9059fb52b733fd37c03078
3
+ size 14408
checkpoint-2555-epoch-4/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-2555-epoch-4/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-2555-epoch-4/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f267899224976ba420425d9d1be5d25735e6fb51c2eb22cc389dfba597dcfb0
3
+ size 2588462170
checkpoint-2555-epoch-4/random_states_0.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61c8b7024de5c903e71a8ecf4b2c760bce62e891f3f7e7549197d81c0e5eef5b
3
+ size 14408
checkpoint-3066-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-3066-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-3066-epoch-5/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:743b9735451c3f434bbf26c12eea90aaaee58ca80beefc8f57e7b9cb578d1b35
3
+ size 2588462170
checkpoint-3066-epoch-5/random_states_0.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:befeb6210ed1910169503a3d4ff1711d1dfe66fe77bda76f31bf2e2704433240
3
+ size 14408
finetuning_concatenated_config.json CHANGED
@@ -8,7 +8,7 @@
8
  "train_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test",
9
  "train_metadata_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal-tags-mistral+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated",
10
  "train_dataset_config_name": "read+default+clean+other",
11
- "train_split_name": "train+train[:20%]+test.clean+test.other",
12
  "eval_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test",
13
  "eval_metadata_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal-tags-mistral+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated",
14
  "eval_dataset_config_name": "read+default+clean+other",
@@ -49,6 +49,7 @@
49
  "save_strategy": "epoch",
50
  "evaluation_strategy": "epoch",
51
  "save_total_limit": 5,
52
- "group_by_length": true
 
53
  }
54
 
 
8
  "train_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test",
9
  "train_metadata_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal-tags-mistral+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated",
10
  "train_dataset_config_name": "read+default+clean+other",
11
+ "train_split_name": "train+train[:20%]+test.clean+test.other[:-7]",
12
  "eval_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test",
13
  "eval_metadata_dataset_name": "sanchit-gandhi/expresso-concatenated-half-normal-tags-mistral+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated",
14
  "eval_dataset_config_name": "read+default+clean+other",
 
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 sha256:6396d1eeba66874a8a1745723484840747515c11cffe72fa122feba5dff56c32
3
  size 2588215392
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a3b4ea5b20a3c4549980d89ef63fa62af70a137ef1926bb1004e65c4a352713
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
+