Abstract: Multimodal large language models (MLLMs) have demonstrated strong language understanding and generation capabilities, excelling in visual tasks like referring and grounding. However, due to ...

Abstract: Despite the recent breakthroughs achieved by Large Vision Language Models (LVLMs) in understanding and responding to complex visual-textual contexts, their inherent hallucination tendencies ...