Use with a fine-tuned model

[Ar770]

Thanks for the great code!
Can you explain if and how it's possible to train a whisper-at model based on a fine-tuned whisper model?
Maybe a more general question, if we have a whisper model (transformers version), what's the process for training a model?
From what I can see, the training dataset isn't in the repo. Where can we find it?