Are `class transformers.CLIPModel` encoders lossy?

#18
by sinanabavi - opened

I would like to decode an embedding to an image or text:
Can I do this for CLIP text and image features?

Sign up or log in to comment