You need to remove the torch_dtype=torch.float16 from the model request.

Ohh, I did it for the colab. Fixing the other bit now.

Manjushri changed pull request status to closed

Sign up or log in to comment