Attention-Space Contrastive Guidance for Efficient Hallucination Mitigation in LVLMs

Yujin Jo, Sangyoon Bae, Taesup Kim

Keywords: Vision, Language, and Reasoning

Chat is not available.