Magic-MM-Embedding: Towards Visual-Token-Efficient Universal Multimodal Embedding with MLLMs Paper • 2602.05275 • Published Feb 5 • 1
