通过跨模态注意力模式检测大模型幻觉，无需训练或额外推理。
DHCP: Detecting Hallucinations by Cross-modal Attention Pattern in Large Vision-Language Models
- 分析视觉语言模型中幻觉与正常状态的跨模态注意力差异。
- 在多个数据集上实现高精度幻觉检测，性能优于现有方法。
- 轻量级设计，适合部署于实际应用，尤其适合关注模型可信性的研究者。
大型视觉语言模型（LVLM）在复杂多模态任务中表现出色，但仍存在显著的幻觉问题，包括物体、属性和关系幻觉。为准确检测这些幻觉，我们研究了幻觉与非幻觉状态下跨模态注意力模式的差异。基于这些差异，我们提出了一种轻量级检测器——基于跨模态注意力模式的幻觉检测（DHCP）。该方法简单有效，无需额外的LVLM训练或推理步骤。实验结果表明，DHCP在幻觉检测任务中表现卓越，为理解并分析LVLM中的幻觉现象提供了新视角，有助于提升模型的可靠性与可信度。代码已开源：https://github.com/btzyd/DHCP。
原文摘要 · Abstract (English)
Large vision-language models (LVLMs) have demonstrated exceptional performance on complex multimodal tasks. However, they continue to suffer from significant hallucination issues, including object, attribute, and relational hallucinations. To accurately detect these hallucinations, we investigated the variations in cross-modal attention patterns between hallucination and non-hallucination states. Leveraging these distinctions, we developed a lightweight detector capable of identifying hallucinations. Our proposed method, Detecting Hallucinations by Cross-modal Attention Patterns (DHCP), is straightforward and does not require additional LVLM training or extra LVLM inference steps. Experimental results show that DHCP achieves remarkable performance in hallucination detection. By offering novel insights into the identification and analysis of hallucinations in LVLMs, DHCP contributes to advancing the reliability and trustworthiness of these models. The code is available at https://github.com/btzyd/DHCP.
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。