Chenxi

@Vaibhavs10 Sure it's added! (btw GMT here) :D

Thanks! Yes pushing the update is necessary since we are changing the server this weekend and pre-pydantic Predictor will no longer be supported by Replicate ~

same request please!

Hey I have added most of the args Whisper accepts, the demo and the API will be updated accordingly, thanks for the suggestion!

just pushed a version supporting this :)

yes now it is set to use `large-v1` for the original and `large-v2` for the new weights

I don't think it implements multi-gpu training? So essentially just one gpu is used

Hi thanks for the suggestions, had to make a bit more changes for the multiple output to work and now they are updated on github and on replicate.

it seems to be a huggingface hub change issue, in the main repo I have temp fix before maybe HF fix the `cache_dir`. See thread here https://github.com/replicate/cog-stable-diffusion/issues/74#issuecomment-1462616896, you can adapt...