MOLBIND: Multimodal Alignment for Drug Discovery

MOLBIND proposes a framework for multi-modal alignment in drug discovery, enhancing zero-shot learning performance across various tasks by mapping different modalities to a shared feature space through contrastive learning.