通过重看视觉信息减少多模态大模型幻觉
Look Twice Before You Answer: Memory-Space Visual Retracing for Hallucination Mitigation in Multimodal Large Language Models
- 在推理不确定时重新注入视觉特征作为记忆
- 多个模型测试中幻觉率显著降低，无额外计算开销
- 适合需要高准确性的视觉问答与内容生成场景
尽管多模态大语言模型（MLLMs）能力强大，但容易产生幻觉——即生成与输入无关或不忠实的内容。与纯文本大模型不同，MLLMs的幻觉常源于文本解码器对视觉标记的敏感性，导致对视觉信息的“遗忘”。为此，我们提出MemVR，一种受人类认知启发的新解码范式：当人忘记刚看到的图像时，会再次查看以获取事实答案。基于此，我们将视觉标记视为补充证据，在中间触发层通过前馈网络（FFN）重新注入模型，形成“键值记忆”。该“看两次”机制在模型推理不确定性高时触发，有效提升内容真实性。大量实验表明，MemVR在多种MLLM上显著减少幻觉，在通用基准上表现优异，且无需额外推理时间。代码已开源：https://github.com/1zhou-Wang/MemVR
原文摘要 · Abstract (English)
Despite their impressive capabilities, multimodal large language models (MLLMs) are prone to hallucinations, i.e., the generated content that is nonsensical or unfaithful to input sources. Unlike in LLMs, hallucinations in MLLMs often stem from the sensitivity of text decoder to visual tokens, leading to a phenomenon akin to "amnesia" about visual information. To address this issue, we propose MemVR, a novel decoding paradigm inspired by common cognition: when the memory of an image seen the moment before is forgotten, people will look at it again for factual answers. Following this principle, we treat visual tokens as supplementary evidence, re-injecting them into the MLLM through Feed Forward Network (FFN) as "key-value memory" at the middle trigger layer. This "look-twice" mechanism occurs when the model exhibits high uncertainty during inference, effectively enhancing factual alignment. Comprehensive experimental evaluations demonstrate that MemVR significantly mitigates hallucination across various MLLMs and excels in general benchmarks without incurring additional time overhead. The implementation is available from https://github.com/1zhou-Wang/MemVR
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。