fix sample usage code

>>> inputs.to(torch.bfloat16)
Attempting to cast a BatchEncoding to type torch.bfloat16. This is not supported.
Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment