It seems everyone wants deep learning to work on M1 chips. At this point, I am quite sure there is enough interest for major frameworks to consider supporting it.