$β$-CLIP: Text-Conditioned Contrastive Learning for Multi-Granular Vision-Language Alignment

CLIPVLMRe-assign community