Fetching the paper…

ELIP: Efficient Discriminative Language-Image Pre-training with Fewer Vision Tokens · Around