用于缓解大视觉语言模型幻觉的指令对齐视觉注意力机制
Instruction-Aligned Visual Attention for Mitigating Hallucinations in Large Vision-Language Models
AI总结：
针对大视觉语言模型因过度关注无关图像令牌而产生幻觉的问题，提出指令对齐视觉注意力方法，通过双指令注意力对比识别无关令牌，结合对比解码调整logits，在多个基准上优于现有解码技术。
摘要：
尽管大视觉语言模型（LVLMs）已取得显著成效，但这类模型在描述图像时仍存在幻觉问题，会生成包含不存在物体的答案。据研究，这类模型往往过度关注某些无关的图像令牌，这些令牌不包含回答问题的关键信息，还会扭曲输出结果。为解决这一问题，我们提出了指令对齐视觉注意力（IAVA）方法，该方法通过对比两种不同指令下注意力权重的变化来识别无关令牌。通过应用对比解码，我们动态调整原始图像令牌和无关图像令牌生成的logits，减少模型对无关信息的过度关注。实验结果表明，在缓解物体幻觉方面，IAVA在MME、POPE和TextVQA等基准测试上的表现始终优于现有解码技术。我们的IAVA方法已在https://github.com/Lee-lab558/IAVA 线上开源。
英文摘要：
Despite the significant success of Large Vision-Language models(LVLMs), these models still suffer hallucinations when describing images, generating answers that include non-existent objects. It is reported that these models tend to over-focus on certain irrelevant image tokens that do not contain critical information for answering the question and distort the output. To address this, we propose an Instruction-Aligned Visual Attention(IAVA) approach, which identifies irrelevant tokens by comparing changes in attention weights under two different instructions. By applying contrastive decoding, we dynamically adjust the logits generated from original image tokens and irrelevant image tokens, reducing the model's over-attention to irrelevant information. The experimental results demonstrate that IAVA consistently outperforms existing decoding techniques on benchmarks such as MME, POPE, and TextVQA in mitigating object hallucinations. Our IAVA approach is available online at https://github.com/Lee-lab558/IAVA.
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。
