Spaces:
Sleeping
Sleeping
| { | |
| "output_path": "/data/notebook_files/fa_run/training", | |
| "logger_uri": "None", | |
| "run_name": "GPT_XTTS_v2.0_Arman_fa", | |
| "project_name": "XTTS_trainer", | |
| "run_description": "\n GPT XTTS training\n ", | |
| "print_step": 50, | |
| "plot_step": 100, | |
| "model_param_stats": false, | |
| "wandb_entity": null, | |
| "dashboard_logger": "tensorboard", | |
| "save_on_interrupt": true, | |
| "log_model_step": 1000, | |
| "save_step": 10000, | |
| "save_n_checkpoints": 1, | |
| "save_checkpoints": true, | |
| "save_all_best": false, | |
| "save_best_after": 0, | |
| "target_loss": null, | |
| "print_eval": false, | |
| "test_delay_epochs": 0, | |
| "run_eval": true, | |
| "run_eval_steps": null, | |
| "distributed_backend": "nccl", | |
| "distributed_url": "tcp://localhost:54321", | |
| "mixed_precision": false, | |
| "precision": "fp16", | |
| "epochs": 1000, | |
| "batch_size": 3, | |
| "eval_batch_size": 3, | |
| "grad_clip": 0.0, | |
| "scheduler_after_epoch": true, | |
| "lr": 5e-06, | |
| "optimizer": "AdamW", | |
| "optimizer_params": { | |
| "betas": [ | |
| 0.9, | |
| 0.96 | |
| ], | |
| "eps": 1e-08, | |
| "weight_decay": 0.01 | |
| }, | |
| "lr_scheduler": "MultiStepLR", | |
| "lr_scheduler_params": { | |
| "milestones": [ | |
| 900000, | |
| 2700000, | |
| 5400000 | |
| ], | |
| "gamma": 0.5, | |
| "last_epoch": -1 | |
| }, | |
| "use_grad_scaler": false, | |
| "allow_tf32": false, | |
| "cudnn_enable": true, | |
| "cudnn_deterministic": false, | |
| "cudnn_benchmark": false, | |
| "training_seed": 1, | |
| "model": "xtts", | |
| "num_loader_workers": 8, | |
| "num_eval_loader_workers": 0, | |
| "use_noise_augment": false, | |
| "audio": { | |
| "sample_rate": 22050, | |
| "output_sample_rate": 24000, | |
| "dvae_sample_rate": 22050 | |
| }, | |
| "use_phonemes": false, | |
| "phonemizer": null, | |
| "phoneme_language": null, | |
| "compute_input_seq_cache": false, | |
| "text_cleaner": null, | |
| "enable_eos_bos_chars": false, | |
| "test_sentences_file": "", | |
| "phoneme_cache_path": null, | |
| "characters": null, | |
| "add_blank": false, | |
| "batch_group_size": 48, | |
| "loss_masking": null, | |
| "min_audio_len": 1, | |
| "max_audio_len": Infinity, | |
| "min_text_len": 1, | |
| "max_text_len": Infinity, | |
| "compute_f0": false, | |
| "compute_energy": false, | |
| "compute_linear_spec": false, | |
| "precompute_num_workers": 0, | |
| "start_by_longest": false, | |
| "shuffle": false, | |
| "drop_last": false, | |
| "datasets": [ | |
| { | |
| "formatter": "", | |
| "dataset_name": "", | |
| "path": "", | |
| "meta_file_train": "", | |
| "ignored_speakers": null, | |
| "language": "", | |
| "phonemizer": "", | |
| "meta_file_val": "", | |
| "meta_file_attn_mask": "" | |
| } | |
| ], | |
| "test_sentences": [ | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| }, | |
| { | |
| "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.", | |
| "speaker_wav": [ | |
| "/data/notebook_files/datasets/ZanAkhbar.wav" | |
| ], | |
| "language": "fa" | |
| } | |
| ], | |
| "eval_split_max_size": 256, | |
| "eval_split_size": 0.01, | |
| "use_speaker_weighted_sampler": false, | |
| "speaker_weighted_sampler_alpha": 1.0, | |
| "use_language_weighted_sampler": false, | |
| "language_weighted_sampler_alpha": 1.0, | |
| "use_length_weighted_sampler": false, | |
| "length_weighted_sampler_alpha": 1.0, | |
| "model_args": { | |
| "gpt_batch_size": 1, | |
| "enable_redaction": false, | |
| "kv_cache": true, | |
| "gpt_checkpoint": "", | |
| "clvp_checkpoint": null, | |
| "decoder_checkpoint": null, | |
| "num_chars": 255, | |
| "tokenizer_file": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/vocab.json", | |
| "gpt_max_audio_tokens": 605, | |
| "gpt_max_text_tokens": 402, | |
| "gpt_max_prompt_tokens": 70, | |
| "gpt_layers": 30, | |
| "gpt_n_model_channels": 1024, | |
| "gpt_n_heads": 16, | |
| "gpt_number_text_tokens": 6681, | |
| "gpt_start_text_token": 261, | |
| "gpt_stop_text_token": 0, | |
| "gpt_num_audio_tokens": 1026, | |
| "gpt_start_audio_token": 1024, | |
| "gpt_stop_audio_token": 1025, | |
| "gpt_code_stride_len": 1024, | |
| "gpt_use_masking_gt_prompt_approach": true, | |
| "gpt_use_perceiver_resampler": true, | |
| "input_sample_rate": 22050, | |
| "output_sample_rate": 24000, | |
| "output_hop_length": 256, | |
| "decoder_input_dim": 1024, | |
| "d_vector_dim": 512, | |
| "cond_d_vector_in_each_upsampling_layer": true, | |
| "duration_const": 102400, | |
| "min_conditioning_length": 66150, | |
| "max_conditioning_length": 132300, | |
| "gpt_loss_text_ce_weight": 0.01, | |
| "gpt_loss_mel_ce_weight": 1.0, | |
| "debug_loading_failures": false, | |
| "max_wav_length": 255995, | |
| "max_text_length": 200, | |
| "mel_norm_file": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/mel_stats.pth", | |
| "dvae_checkpoint": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/dvae.pth", | |
| "xtts_checkpoint": "/data/notebook_files/fa_run/training/XTTS_v2.0_original_model_files/model.pth", | |
| "vocoder": "" | |
| }, | |
| "model_dir": null, | |
| "languages": [ | |
| "en", | |
| "es", | |
| "fr", | |
| "de", | |
| "it", | |
| "pt", | |
| "pl", | |
| "tr", | |
| "ru", | |
| "nl", | |
| "cs", | |
| "ar", | |
| "zh-cn", | |
| "hu", | |
| "ko", | |
| "ja", | |
| "hi", | |
| "fa" | |
| ], | |
| "temperature": 0.85, | |
| "length_penalty": 1.0, | |
| "repetition_penalty": 2.0, | |
| "top_k": 50, | |
| "top_p": 0.85, | |
| "num_gpt_outputs": 1, | |
| "gpt_cond_len": 12, | |
| "gpt_cond_chunk_len": 4, | |
| "max_ref_len": 10, | |
| "sound_norm_refs": false, | |
| "optimizer_wd_only_on_weights": true, | |
| "weighted_loss_attrs": {}, | |
| "weighted_loss_multipliers": {}, | |
| "github_branch": "* master" | |
| } |