How to set the config of finetune?

[JingxuOuyang]

I followed the steps in the readme and the model I trained has very poor inference performance. Is this normal? I found that the document suggested that the training epochs are [100,600], and the configuration config.json is 10000, but I tried both parameters, and the inferred audio effect was very poor. Please tell me what configuration should I use to achieve good results?