| { | |
| "exp_name": "F5TTS_Base", | |
| "learning_rate": 1e-05, | |
| "batch_size_per_gpu": 4800, | |
| "batch_size_type": "frame", | |
| "max_samples": 64, | |
| "grad_accumulation_steps": 1, | |
| "max_grad_norm": 1, | |
| "epochs": 300, | |
| "num_warmup_updates": 500, | |
| "save_per_updates": 10000, | |
| "last_per_steps": 1000, | |
| "finetune": true, | |
| "file_checkpoint_train": "/notebooks/F5-TTS/data/fullspeakers_pinyin/model_last.pt", | |
| "tokenizer_type": "pinyin", | |
| "tokenizer_file": "", | |
| "mixed_precision": "fp16", | |
| "logger": "wandb" | |
| } |