barghavani commited on
Commit
778e287
·
verified ·
1 Parent(s): 433fa1c

Upload 2 files

Browse files
checkpoints/config (5).json ADDED
@@ -0,0 +1,290 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "output_path": "/data/notebook_files/fa_run/training",
3
+ "logger_uri": "None",
4
+ "run_name": "GPT_XTTS_v2.0_Arman_fa",
5
+ "project_name": "XTTS_trainer",
6
+ "run_description": "\n GPT XTTS training\n ",
7
+ "print_step": 50,
8
+ "plot_step": 100,
9
+ "model_param_stats": false,
10
+ "wandb_entity": null,
11
+ "dashboard_logger": "tensorboard",
12
+ "save_on_interrupt": true,
13
+ "log_model_step": 1000,
14
+ "save_step": 10000,
15
+ "save_n_checkpoints": 1,
16
+ "save_checkpoints": true,
17
+ "save_all_best": false,
18
+ "save_best_after": 0,
19
+ "target_loss": null,
20
+ "print_eval": false,
21
+ "test_delay_epochs": 0,
22
+ "run_eval": true,
23
+ "run_eval_steps": null,
24
+ "distributed_backend": "nccl",
25
+ "distributed_url": "tcp://localhost:54321",
26
+ "mixed_precision": false,
27
+ "precision": "fp16",
28
+ "epochs": 1000,
29
+ "batch_size": 3,
30
+ "eval_batch_size": 3,
31
+ "grad_clip": 0.0,
32
+ "scheduler_after_epoch": true,
33
+ "lr": 5e-06,
34
+ "optimizer": "AdamW",
35
+ "optimizer_params": {
36
+ "betas": [
37
+ 0.9,
38
+ 0.96
39
+ ],
40
+ "eps": 1e-08,
41
+ "weight_decay": 0.01
42
+ },
43
+ "lr_scheduler": "MultiStepLR",
44
+ "lr_scheduler_params": {
45
+ "milestones": [
46
+ 900000,
47
+ 2700000,
48
+ 5400000
49
+ ],
50
+ "gamma": 0.5,
51
+ "last_epoch": -1
52
+ },
53
+ "use_grad_scaler": false,
54
+ "allow_tf32": false,
55
+ "cudnn_enable": true,
56
+ "cudnn_deterministic": false,
57
+ "cudnn_benchmark": false,
58
+ "training_seed": 1,
59
+ "model": "xtts",
60
+ "num_loader_workers": 8,
61
+ "num_eval_loader_workers": 0,
62
+ "use_noise_augment": false,
63
+ "audio": {
64
+ "sample_rate": 22050,
65
+ "output_sample_rate": 24000,
66
+ "dvae_sample_rate": 22050
67
+ },
68
+ "use_phonemes": false,
69
+ "phonemizer": null,
70
+ "phoneme_language": null,
71
+ "compute_input_seq_cache": false,
72
+ "text_cleaner": null,
73
+ "enable_eos_bos_chars": false,
74
+ "test_sentences_file": "",
75
+ "phoneme_cache_path": null,
76
+ "characters": null,
77
+ "add_blank": false,
78
+ "batch_group_size": 48,
79
+ "loss_masking": null,
80
+ "min_audio_len": 1,
81
+ "max_audio_len": Infinity,
82
+ "min_text_len": 1,
83
+ "max_text_len": Infinity,
84
+ "compute_f0": false,
85
+ "compute_energy": false,
86
+ "compute_linear_spec": false,
87
+ "precompute_num_workers": 0,
88
+ "start_by_longest": false,
89
+ "shuffle": false,
90
+ "drop_last": false,
91
+ "datasets": [
92
+ {
93
+ "formatter": "",
94
+ "dataset_name": "",
95
+ "path": "",
96
+ "meta_file_train": "",
97
+ "ignored_speakers": null,
98
+ "language": "",
99
+ "phonemizer": "",
100
+ "meta_file_val": "",
101
+ "meta_file_attn_mask": ""
102
+ }
103
+ ],
104
+ "test_sentences": [
105
+ {
106
+ "text": "\u0632\u06cc\u0646 \u0647\u0645\u0631\u0647\u0627\u0646 \u0633\u0633\u062a \u0639\u0646\u0627\u0635\u0631\u060c \u062f\u0644\u0645 \u06af\u0631\u0641\u062a.",
107
+ "speaker_wav": [
108
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
109
+ ],
110
+ "language": "fa"
111
+ },
112
+ {
113
+ "text": "\u0622\u0646\u06a9\u0647 \u0645\u062f\u0627\u0645 \u0628\u0647 \u06a9\u0627\u0631 \u062f\u06cc\u06af\u0631\u0627\u0646 \u0633\u0631\u06a9 \u0645\u06cc \u06a9\u0634\u062f \u0648 \u06a9\u0646\u062c\u06a9\u0627\u0648 \u0627\u0633\u062a \u062a\u0627 \u0628\u0628\u06cc\u0646\u062f \u0622\u0646\u0647\u0627 \u0686\u0647 \u0645\u06cc \u06a9\u0646\u0646\u062f \u0645\u0627\u0646\u0646\u062f \u0633\u0627\u06cc\u0647 \u0627\u06cc\u06cc \u0628\u0631 \u062f\u06cc\u0648\u0627\u0631 \u0627\u0633\u062a \u06a9\u0647 \u0645\u062f\u0627\u0645 \u0628\u062f\u0646\u0628\u0627\u0644 \u0645\u0627 \u0645\u06cc \u062f\u0648\u062f \u0628\u062f\u0648\u0646 \u0622\u0646\u06a9\u0647 \u0627\u0632 \u062e\u0648\u062f \u0627\u062e\u062a\u06cc\u0627\u0631\u06cc \u062f\u0627\u0634\u062a\u0647 \u0628\u0627\u0634\u062f",
114
+ "speaker_wav": [
115
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
116
+ ],
117
+ "language": "fa"
118
+ },
119
+ {
120
+ "text": "\u0628\u06cc\u0627 \u062a\u0627 \u06af\u0644 \u0628\u0631\u0627\u0641\u0634\u0627\u0646\u06cc\u0645 \u0648 \u0645\u06cc \u062f\u0631 \u0633\u0627\u063a\u0631 \u0627\u0646\u062f\u0627\u0632\u06cc\u0645.",
121
+ "speaker_wav": [
122
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
123
+ ],
124
+ "language": "fa"
125
+ },
126
+ {
127
+ "text": "\u0631\u0627\u0647 \u0622\u0634\u062a\u06cc \u0631\u0627 \u06a9\u0633\u06cc \u0628\u0627\u06cc\u062f \u0628\u06cc\u0627\u0628\u062f \u06a9\u0647 \u062e\u0648\u062f \u0633\u0628\u0628 \u062c\u062f\u0627\u06cc\u06cc \u0634\u062f\u0647 \u0627\u0633\u062a",
128
+ "speaker_wav": [
129
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
130
+ ],
131
+ "language": "fa"
132
+ },
133
+ {
134
+ "text": "\u0628\u0646\u06cc \u0622\u062f\u0645 \u0627\u0639\u0636\u0627\u06cc \u06cc\u06a9 \u067e\u06cc\u06a9\u0631\u0646\u062f, \u06a9\u0647 \u062f\u0631 \u0622\u0641\u0631\u06cc\u0646\u0634 \u0632 \u06cc\u06a9 \u06af\u0648\u0647\u0631\u0646\u062f.",
135
+ "speaker_wav": [
136
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
137
+ ],
138
+ "language": "fa"
139
+ },
140
+ {
141
+ "text": "\u0633\u0647\u0627\u0645 \u0632\u0646\u062f\u06af\u06cc \u0628\u0647 10 \u062f\u0631\u0635\u062f \u0648 \u0633\u0647\u0627\u0645 \u0628\u06cc\u062a\u06a9\u0648\u06cc\u0646 \u06af\u0648\u06af\u0644 \u0628\u0647 33 \u062f\u0631\u0635\u062f \u0627\u0641\u0632\u0627\u06cc\u0634 \u06cc\u0627\u0641\u062a.",
142
+ "speaker_wav": [
143
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
144
+ ],
145
+ "language": "fa"
146
+ },
147
+ {
148
+ "text": "\u06cc\u0647 \u062f\u0648 \u062f\u0642\u0647 \u0647\u0645 \u0628\u0647 \u062d\u0631\u0641\u0645 \u06af\u0648\u0634 \u06a9\u0646\u060c \u0646\u06af\u0648 \u0646\u06af\u0648\u0634\u06cc\u062f\u0645 \u0648 \u0646\u062d\u0631\u0641\u06cc\u062f\u06cc.",
149
+ "speaker_wav": [
150
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
151
+ ],
152
+ "language": "fa"
153
+ },
154
+ {
155
+ "text": "\u0645\u0637\u0627\u0644\u0639\u0627\u062a \u0646\u0634\u0627\u0646 \u062f\u0627\u062f\u0647\u200c\u0627\u0646\u062f \u06a9\u0647 40 \u062f\u0631\u0635\u062f \u0627\u0632 \u062f\u0627\u0646\u0634\u200c\u0622\u0645\u0648\u0632\u0627\u0646\u060c \u062f\u0631 \u0633\u0627\u0644 \u062a\u062d\u0635\u06cc\u0644\u06cc 2024-2023\u060c \u062f\u0631 \u062f\u0648\u0631\u0647\u200c\u0647\u0627\u06cc \u0622\u0646\u0644\u0627\u06cc\u0646 \u0634\u0631\u06a9\u062a \u06a9\u0631\u062f\u0647\u200c\u0627\u0646\u062f\u060c \u06a9\u0647 \u0627\u06cc\u0646 \u0627\u0645\u0631 \u0646\u0634\u0627\u0646\u200c\u062f\u0647\u0646\u062f\u0647 \u062a\u063a\u06cc\u06cc\u0631 \u0686\u0634\u0645\u06af\u06cc\u0631\u06cc \u062f\u0631 \u0631\u0648\u0634\u200c\u0647\u0627\u06cc \u0622\u0645\u0648\u0632\u0634\u06cc \u062f\u0631 \u0627\u06cc\u0631\u0627\u0646 \u0645\u06cc\u200c\u0628\u0627\u0634\u062f.",
156
+ "speaker_wav": [
157
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
158
+ ],
159
+ "language": "fa"
160
+ },
161
+ {
162
+ "text": "\u0645\u0646 \u0628\u0648\u062f\u0645 \u0648 \u0622\u0628\u062c\u06cc \u0641\u0648\u062a\u06cc\u0646\u0627\u060c \u0648 \u062d\u0627\u0644\u0627 \u0631\u067e\u062a\u06cc \u067e\u062a\u06cc\u0646\u0627. \u0627\u06cc\u0646 \u0634\u0639\u0631 \u06cc\u06a9\u06cc \u0627\u0632 \u0627\u0634\u0639\u0627\u0631 \u0645\u0639\u0631\u0648\u0641 \u0631\u0648 \u062d\u0648\u0636\u06cc \u0627\u0633\u062a \u06a9\u0647 \u062f\u0631 \u06a9\u0648\u0686\u0647 \u0628\u0627\u0632\u0627\u0631 \u062a\u0647\u0631\u0627\u0646 \u0632\u0645\u0632\u0645\u0647 \u0645\u06cc \u0634\u062f\u0647 \u0627\u0633\u062a.",
163
+ "speaker_wav": [
164
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
165
+ ],
166
+ "language": "fa"
167
+ },
168
+ {
169
+ "text": "\u062f\u0631 \u0628\u0648\u062f\u062c\u0647 \u0633\u0627\u0644 2025\u060c \u062f\u0648\u0644\u062a \u0627\u06cc\u0631\u0627\u0646 \u0627\u0639\u0644\u0627\u0645 \u06a9\u0631\u062f \u06a9\u0647 15 \u062f\u0631\u0635\u062f \u0627\u0632 \u06a9\u0644 \u0628\u0648\u062f\u062c\u0647\u060c \u0645\u0639\u0627\u062f\u0644 \u062a\u0642\u0631\u06cc\u0628\u0627\u064b 30 \u0645\u06cc\u0644\u06cc\u0627\u0631\u062f \u062f\u0644\u0627\u0631\u060c \u0628\u0647 \u062a\u0648\u0633\u0639\u0647 \u0632\u06cc\u0631\u0633\u0627\u062e\u062a\u200c\u0647\u0627\u06cc \u062d\u0645\u0644\u200c\u0648\u0646\u0642\u0644 \u0648 25 \u062f\u0631\u0635\u062f \u062f\u06cc\u06af\u0631\u060c \u0628\u0647 \u0628\u0647\u0628\u0648\u062f \u0628\u062e\u0634 \u0628\u0647\u062f\u0627\u0634\u062a \u0648 \u062f\u0631\u0645\u0627\u0646 \u0627\u062e\u062a\u0635\u0627\u0635 \u06cc\u0627\u0641\u062a\u0647 \u0627\u0633\u062a.",
170
+ "speaker_wav": [
171
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
172
+ ],
173
+ "language": "fa"
174
+ },
175
+ {
176
+ "text": "\u062a\u062d\u0642\u06cc\u0642\u0627\u062a \u0628\u0627\u0632\u0627\u0631 \u0646\u0634\u0627\u0646 \u0645\u06cc\u200c\u062f\u0647\u062f \u06a9\u0647 \u0641\u0631\u0648\u0634 \u06af\u0648\u0634\u06cc\u200c\u0647\u0627\u06cc \u0647\u0648\u0634\u0645\u0646\u062f \u062f\u0631 \u0627\u06cc\u0631\u0627\u0646\u060c \u062f\u0631 \u0633\u0647 \u0645\u0627\u0647\u0647 \u0627\u0648\u0644 \u0633\u0627\u0644 2024\u060c \u0646\u0633\u0628\u062a \u0628\u0647 \u0645\u062f\u062a \u0645\u0634\u0627\u0628\u0647 \u062f\u0631 \u0633\u0627\u0644 \u0642\u0628\u0644\u060c 55 \u062f\u0631\u0635\u062f \u0627\u0641\u0632\u0627\u06cc\u0634 \u06cc\u0627\u0641\u062a\u0647 \u06a9\u0647 \u0627\u06cc\u0646 \u067e\u062f\u06cc\u062f\u0647 \u0628\u06cc\u0634\u062a\u0631 \u0628\u0647 \u062f\u0644\u06cc\u0644 \u0639\u0631\u0636\u0647 \u0645\u062f\u0644\u200c\u0647\u0627\u06cc \u062c\u062f\u06cc\u062f \u0648 \u06a9\u0627\u0647\u0634 \u0642\u06cc\u0645\u062a\u200c\u0647\u0627 \u0627\u0633\u062a.",
177
+ "speaker_wav": [
178
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
179
+ ],
180
+ "language": "fa"
181
+ },
182
+ {
183
+ "text": "\u0633\u0644\u0627\u0645\u060c \u062e\u0648\u0634\u062d\u0627\u0644\u0645 \u06a9\u0647 \u062f\u0648\u0628\u0627\u0631\u0647 \u0645\u06cc\u200c\u0628\u06cc\u0646\u0645\u062a\u061b \u0627\u0645\u06cc\u062f\u0648\u0627\u0631\u0645 \u06a9\u0647 \u0627\u06cc\u0646 \u0647\u0641\u062a\u0647 \u0628\u0631\u0627\u062a \u062e\u0648\u0628 \u06af\u0630\u0634\u062a\u0647 \u0628\u0627\u0634\u0647.",
184
+ "speaker_wav": [
185
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
186
+ ],
187
+ "language": "fa"
188
+ },
189
+ {
190
+ "text": "\u0628\u0631\u0627\u06cc \u062c\u0634\u0646 \u062a\u0648\u0644\u062f \u067e\u062f\u0631\u0645 \u0642\u0635\u062f \u062f\u0627\u0631\u0645 \u06cc\u06a9 \u0645\u0647\u0645\u0627\u0646\u06cc \u06a9\u0648\u0686\u06a9 \u0628\u0627 \u062e\u0627\u0646\u0648\u0627\u062f\u0647 \u0648 \u062f\u0648\u0633\u062a\u0627\u0646 \u0646\u0632\u062f\u06cc\u06a9 \u062a\u0631\u062a\u06cc\u0628 \u0628\u062f\u0647\u0645 \u061b \u062f\u0648\u0633\u062a \u062f\u0627\u0631\u06cc \u062a\u0648 \u0647\u0645 \u0628\u06cc\u0627\u06cc\u061f",
191
+ "speaker_wav": [
192
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
193
+ ],
194
+ "language": "fa"
195
+ },
196
+ {
197
+ "text": "\u062f\u06cc\u0634\u0628 \u06cc\u06a9 \u0641\u06cc\u0644\u0645 \u062c\u0630\u0627\u0628 \u062f\u0631 \u062a\u0644\u0648\u06cc\u0632\u06cc\u0648\u0646 \u062f\u06cc\u062f\u0645 \u06a9\u0647 \u062f\u0631 \u0645\u0648\u0631\u062f \u062a\u0627\u0631\u06cc\u062e\u0686\u0647 \u062a\u0645\u062f\u0646\u200c\u0647\u0627\u06cc \u0628\u0627\u0633\u062a\u0627\u0646 \u0628\u0648\u062f\u061b \u0648\u0627\u0642\u0639\u0627\u064b \u0627\u0637\u0644\u0627\u0639\u0627\u062a \u0645\u0641\u06cc\u062f \u0648 \u062c\u0627\u0644\u0628\u06cc \u062f\u0627\u0634\u062a.",
198
+ "speaker_wav": [
199
+ "/data/notebook_files/datasets/ZanAkhbar.wav"
200
+ ],
201
+ "language": "fa"
202
+ }
203
+ ],
204
+ "eval_split_max_size": 256,
205
+ "eval_split_size": 0.01,
206
+ "use_speaker_weighted_sampler": false,
207
+ "speaker_weighted_sampler_alpha": 1.0,
208
+ "use_language_weighted_sampler": false,
209
+ "language_weighted_sampler_alpha": 1.0,
210
+ "use_length_weighted_sampler": false,
211
+ "length_weighted_sampler_alpha": 1.0,
212
+ "model_args": {
213
+ "gpt_batch_size": 1,
214
+ "enable_redaction": false,
215
+ "kv_cache": true,
216
+ "gpt_checkpoint": "",
217
+ "clvp_checkpoint": null,
218
+ "decoder_checkpoint": null,
219
+ "num_chars": 255,
220
+ "tokenizer_file": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/vocab.json",
221
+ "gpt_max_audio_tokens": 605,
222
+ "gpt_max_text_tokens": 402,
223
+ "gpt_max_prompt_tokens": 70,
224
+ "gpt_layers": 30,
225
+ "gpt_n_model_channels": 1024,
226
+ "gpt_n_heads": 16,
227
+ "gpt_number_text_tokens": 6681,
228
+ "gpt_start_text_token": 261,
229
+ "gpt_stop_text_token": 0,
230
+ "gpt_num_audio_tokens": 1026,
231
+ "gpt_start_audio_token": 1024,
232
+ "gpt_stop_audio_token": 1025,
233
+ "gpt_code_stride_len": 1024,
234
+ "gpt_use_masking_gt_prompt_approach": true,
235
+ "gpt_use_perceiver_resampler": true,
236
+ "input_sample_rate": 22050,
237
+ "output_sample_rate": 24000,
238
+ "output_hop_length": 256,
239
+ "decoder_input_dim": 1024,
240
+ "d_vector_dim": 512,
241
+ "cond_d_vector_in_each_upsampling_layer": true,
242
+ "duration_const": 102400,
243
+ "min_conditioning_length": 66150,
244
+ "max_conditioning_length": 132300,
245
+ "gpt_loss_text_ce_weight": 0.01,
246
+ "gpt_loss_mel_ce_weight": 1.0,
247
+ "debug_loading_failures": false,
248
+ "max_wav_length": 255995,
249
+ "max_text_length": 200,
250
+ "mel_norm_file": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/mel_stats.pth",
251
+ "dvae_checkpoint": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/dvae.pth",
252
+ "xtts_checkpoint": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/model.pth",
253
+ "vocoder": ""
254
+ },
255
+ "model_dir": null,
256
+ "languages": [
257
+ "en",
258
+ "es",
259
+ "fr",
260
+ "de",
261
+ "it",
262
+ "pt",
263
+ "pl",
264
+ "tr",
265
+ "ru",
266
+ "nl",
267
+ "cs",
268
+ "ar",
269
+ "zh-cn",
270
+ "hu",
271
+ "ko",
272
+ "ja",
273
+ "hi",
274
+ "fa"
275
+ ],
276
+ "temperature": 0.85,
277
+ "length_penalty": 1.0,
278
+ "repetition_penalty": 2.0,
279
+ "top_k": 50,
280
+ "top_p": 0.85,
281
+ "num_gpt_outputs": 1,
282
+ "gpt_cond_len": 12,
283
+ "gpt_cond_chunk_len": 4,
284
+ "max_ref_len": 10,
285
+ "sound_norm_refs": false,
286
+ "optimizer_wd_only_on_weights": true,
287
+ "weighted_loss_attrs": {},
288
+ "weighted_loss_multipliers": {},
289
+ "github_branch": "* master"
290
+ }
checkpoints/vocab (27).json ADDED
The diff for this file is too large to render. See raw diff