通过增强相关视觉特征注意力，减少大模型幻觉。
Attention to details, logits to truth: visual-aware attention and logits enhancement to mitigate hallucinations in LVLMs
- 基于图文相似性重分配注意力，聚焦关键视觉信息
- 在主流模型上显著降低幻觉率，同时保持生成质量
- 无需训练，可直接部署于现有视觉语言模型
现有大型视觉语言模型（LVLMs）存在视觉注意力不足的问题，导致幻觉现象。为缓解此问题，部分研究尝试调整并增强视觉注意力，但此类方法会无差别提升所有视觉标记的关注度，包括与任务无关的标记。为此，本文提出一种无需训练的注意力干预算法，其核心思想是：任务相关的视觉标记通常具有较高的视觉-文本相似性。具体地，从视觉-文本交叉注意力子矩阵中提取视觉-文本关联信息，构建重加权矩阵以重新分配注意力。此外，将视觉注意力值注入束搜索解码过程，以识别具有更高视觉关注度的生成路径。大量实验表明，该方法能显著降低主流LVLMs中的幻觉率，同时保持生成内容的准确性和连贯性。
原文摘要 · Abstract (English)
Existing Large Vision-Language Models (LVLMs) exhibit insufficient visual attention, leading to hallucinations. To alleviate this problem, some previous studies adjust and amplify visual attention. These methods present a limitation that boosting attention for all visual tokens inevitably increases attention to task irrelevant tokens. To tackle this challenge, we propose a training free attentional intervention algorithm to enhance the attention of task-relevant tokens based on the argument that task-relevant tokens generally demonstrate high visual-textual similarities. Specifically, the vision-text cross-attention submatrices, which represent visual-textual correlations, are extracted to construct the reweighting matrices to reallocate attention. Besides, to enhance the contribution of visual tokens, we inject visual attention values into the beam search decoding to identify solutions with higher visual attention. Extensive experiments demonstrate that this method significantly reduces hallucinations across mainstream LVLMs, while preserving the accuracy and coherence of generated content.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。