| { | |
| "model_name_or_path": "parler-tts/parler_tts_mini_v0.1", | |
| "feature_extractor_name": "parler-tts/dac_44khZ_8kbps", | |
| "description_tokenizer_name": "parler-tts/parler_tts_mini_v0.1", | |
| "prompt_tokenizer_name": "parler-tts/parler_tts_mini_v0.1", | |
| "report_to": ["wandb"], | |
| "overwrite_output_dir": true, | |
| "train_dataset_name": "ylacombe/expresso+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test", | |
| "train_metadata_dataset_name": "reach-vb/expresso-tagged-w-speech-mistral-v3+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated", | |
| "train_dataset_config_name": "read+default+clean+other", | |
| "train_split_name": "train[:-25]+train[:20%]+test.clean+test.other[:-31]", | |
| "eval_dataset_name": "ylacombe/expresso+reach-vb/jenny_tts_dataset+sanchit-gandhi/libritts_r_test+sanchit-gandhi/libritts_r_test", | |
| "eval_metadata_dataset_name": "reach-vb/expresso-tagged-w-speech-mistral-v3+ylacombe/jenny-tts-10k-tagged+parler-tts/libritts_r_tags_tagged_10k_generated+parler-tts/libritts_r_tags_tagged_10k_generated", | |
| "eval_dataset_config_name": "read+default+clean+other", | |
| "eval_split_name": "train+train[:20%]+test.clean+test.other", | |
| "max_eval_samples": 8, | |
| "per_device_eval_batch_size": 16, | |
| "target_audio_column_name": "audio", | |
| "description_column_name": "text_description", | |
| "prompt_column_name": "text", | |
| "max_duration_in_seconds": 30.0, | |
| "min_duration_in_seconds": 2.0, | |
| "max_text_length": 400, | |
| "preprocessing_num_workers": 2, | |
| "do_train": true, | |
| "num_train_epochs": 8, | |
| "max_steps": -1, | |
| "gradient_accumulation_steps": 8, | |
| "gradient_checkpointing": true, | |
| "per_device_train_batch_size": 16, | |
| "learning_rate": 0.00008, | |
| "adam_beta1": 0.9, | |
| "adam_beta2": 0.99, | |
| "weight_decay": 0.01, | |
| "lr_scheduler_type": "cosine", | |
| "warmup_steps": 250, | |
| "logging_steps": 5, | |
| "freeze_text_encoder": true, | |
| "audio_encoder_per_device_batch_size": 4, | |
| "dtype": "bfloat16", | |
| "seed": 456, | |
| "output_dir": "./", | |
| "temporary_save_to_disk": "../audio_code_tmp_constant/", | |
| "save_to_disk": "../tmp_dataset_audio_constant/", | |
| "dataloader_num_workers": 4, | |
| "do_eval": true, | |
| "predict_with_generate": true, | |
| "include_inputs_for_metrics": true, | |
| "save_strategy": "epoch", | |
| "evaluation_strategy": "epoch", | |
| "save_total_limit": 5, | |
| "group_by_length": true, | |
| "push_to_hub": true | |
| } | |