面向多模态大语言模型的接地思维链
Grounded Chain-of-Thought for Multimodal Large Language Models
AI总结：
针对多模态大语言模型的视觉幻觉问题，提出接地思维链任务与MM-GCoT数据集，构建一致性评估体系，实验发现视觉幻觉与模型规模无直接关联，且该数据集可提升模型相关能力并泛化至多模态任务。
摘要：
尽管多模态大语言模型（MLLMs）已取得长足进展，但现有模型仍易出现视觉幻觉问题，严重阻碍了其可信应用。本文从视觉空间推理的视角研究该问题，并为MLLMs提出了一项名为接地思维链（Grounded Chain-of-Thought, GCoT）的全新学习任务。与近期更多聚焦于视觉知识推理的视觉思维链研究不同，GCoT致力于帮助MLLMs逐步识别并定位相关视觉线索，从而以定位坐标作为直观依据预测正确答案。为推动该任务的研究，我们还精心设计构建了一个名为多模态接地思维链（multimodal grounded chain-of-thought, MM-GCoT）的数据集，包含对应5033张图像的24022个GCoT样本。此外，本文还引入了一套全面的一致性评估体系，涵盖答案准确率、定位准确率以及答案-定位一致性三项指标。我们进一步针对12个先进MLLM设计并开展了一系列实验，揭示了若干值得关注的发现：其一，大多数MLLM在一致性评估中表现不佳，表明存在明显的视觉幻觉问题；其二，视觉幻觉与模型参数量规模及通用多模态性能无直接关联，即规模更大、性能更强的MLLM受该问题的影响并不会更小。最后，我们还验证了所提出的数据集能够帮助现有MLLM很好地培养GCoT能力，并显著降低回答不一致性。此外，这些模型的GCoT能力还可泛化到现有多模态任务中，例如开放世界问答和指代表达理解（REC）。
英文摘要：
Despite great progress, existing multimodal large language models (MLLMs) are prone to visual hallucination, greatly impeding their trustworthy applications. In this paper, we study this problem from the perspective of visual-spatial reasoning, and propose a new learning task for MLLMs, termed Grounded Chain-of-Thought (GCoT). Different from recent visual CoT studies, which focus more on visual knowledge reasoning, GCoT is keen to helping MLLMs to recognize and ground the relevant visual cues step by step, thereby predicting the correct answer with grounding coordinates as the intuitive basis. To facilitate this task, we also carefully design and construct a dataset called multimodal grounded chain-of-thought (MM-GCoT) consisting of 24,022 GCoT examples for 5,033 images. Besides, a comprehensive consistency evaluation system is also introduced, including the metrics of answer accuracy, grounding accuracy and answer-grounding consistency. We further design and conduct a bunch of experiments on 12 advanced MLLMs, and reveal some notable findings: i. most MLLMs performs poorly on the consistency evaluation, indicating obvious visual hallucination; ii. visual hallucination is not directly related to the parameter size and general multimodal performance, i.e., a larger and stronger MLLM is not less affected by this issue. Lastly, we also demonstrate that the proposed dataset can help existing MLLMs to well cultivate their GCoT capability and reduce the inconsistent answering significantly. Moreover, their GCoT can be also generalized to exiting multimodal tasks, such as open-world QA and REC.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
