How to Resume Training in Gradio Interface?

[hamreen]

I'm using the Gradio interface for training F5-TTS, and I want to resume training after stopping it. For example, if I stop after 20 epochs, I want to continue from the same parameters to improve the model further.

I tried simply going to the training tab and starting again, but I got this error:
Fatal Python error: config_init_hash_seed: PYTHONHASHSEED must be "random" or an integer in range [0; 4294967295]
Python runtime state: preinitialized

I also tried enabling the Finetune checkbox and setting the last model's .pt file in Path to the Pretrained Checkpoint, but I wasn’t sure what to put in Tokenizer.

Do I need to prepare the data again, or can I continue without reprocessing it?

Note: I'm not changing the dataset or any parameters; I just want to resume training from where I left off.

Can someone guide me on the correct steps to resume training using the Gradio interface?

[h408]

Try deleting (or just move) the most recent partially completed checkpoint file from the directory and then rerun training.

[hamreen]

@h408 Thank you for your replay, You mean i simply delete "model_last.pt" and rerun the training?

[zeng16107]

Try deleting (or just move) the most recent partially completed checkpoint file from the directory and then rerun training.

i met this issue too. can you explain detailed about how to fix this issue?