风险感知的选择性提示用于大型视觉-语言模型中的幻觉缓解
Risk-aware Selective Prompting for Hallucination Mitigation in Large Vision-Language Models
AI总结：
本文系统研究提示验证在大型视觉-语言模型中的风险，发现其效果依赖输入难度，并提出基于预生成不确定性信号的选择性提示方法RSP以平衡性能。
摘要：
基于提示的验证被广泛用于缓解大型视觉-语言模型（LVLMs）中的幻觉，但其何时有效仍不清楚。我们系统研究了两种代表性LVLM架构和幻觉基准上的验证提示，发现它是一种有风险的干预：其纠正随输入难度增加，而新引入的错误在不同难度级别持续存在。因此，始终开启的提示在困难输入上有帮助，但在简单输入上益处甚微甚至有害。我们的分析进一步表明，这种行为与保守的输出偏移相关。验证提示将注意力从视觉令牌重新分配到指令令牌，并诱导出中性提示控制中不存在的中层熵模式，这表明是指令条件化的注意力重新分配而非统一的视觉基础改善。受这种输入依赖风险的启发，我们提出了风险感知的选择性提示（RSP），一种无需训练的方法，利用预生成不确定性信号选择性地触发验证。RSP减轻了始终开启提示的性能下降，同时保持基线性能，并揭示了有效的选择信号因架构而异。
英文摘要：
Prompt-based verification is widely used to mitigate hallucinations in large vision-language models (LVLMs), yet when it helps remains poorly understood. We systematically study verification prompting across two representative LVLM architectures and hallucination benchmarks, and find that it is a risk-bearing intervention: its corrections increase with input difficulty, while newly introduced errors persist across difficulty levels. As a result, always-on prompting helps on hard inputs but offers little benefit -- and can harm -- easier ones. Our analysis further shows that this behavior is associated with a conservative output shift. Verification prompts redistribute attention from visual tokens toward instruction tokens and induce a distinct middle-layer entropy pattern absent in a neutral-prompt control, suggesting instruction-conditioned attention redistribution rather than uniformly improved visual grounding. Motivated by this input-dependent risk, we propose Risk-aware Selective Prompting (RSP), a training-free approach that uses pre-generation uncertainty signals to trigger verification selectively. RSP mitigates the degradation of always-on prompting while preserving baseline performance, and reveals that effective selection signals vary across architectures.
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。
