CLIP: Connecting Text and Image (Learning Transferable Visual Models From Natural Language Supervision) - View it on GitHub