CLIP Exhibits Improved Compositional Generalization Through Representation DisentanglementPublished in Preprint (ICLR 2024 submission), 2023PDFShare on Twitter Facebook LinkedIn Previous Next