通过多模型对比检测并修正大模型幻觉，提升生成内容准确性。
Zero-knowledge LLM hallucination detection and mitigation through fine-grained cross-model consistency
- 用不同模型对相似问题作答，比对细节差异发现幻觉
- 在GPQA-diamond上使答案准确率提升9个百分点
- 无需外部知识库，适合生产环境部署
大型语言模型在各类任务中表现出色，但仍易产生幻觉——生成看似合理却含事实错误的内容。本文提出Finch-Zk，一种黑盒框架，通过细粒度跨模型一致性来检测并缓解LLM输出中的幻觉，无需依赖外部知识源。该框架引入两项关键创新：1）通过比较不同模型对语义等价提示生成的回答，揭示细粒度错误；2）针对问题片段实施精准修正，同时保留正确内容。在FELM数据集上的实验表明，与现有方法相比，Finch-Zk将幻觉检测的F1分数提升6-39%。在GPQA-diamond数据集上，应用于Llama 4 Maverick和Claude 4 Sonnet等先进模型时，答案准确率最高提升9个绝对百分点。多数据集评估显示，Finch-Zk为提升生产级LLM系统的事实可靠性提供了实用且可部署的保障。
原文摘要 · Abstract (English)
Large language models (LLMs) have demonstrated impressive capabilities across diverse tasks, but they remain susceptible to hallucinations--generating content that appears plausible but contains factual inaccuracies. We present Finch-Zk, a black-box framework that leverages fine-grained cross-model consistency to detect and mitigate hallucinations in LLM outputs without requiring external knowledge sources. Finch-Zk introduces two key innovations: 1) a cross-model consistency checking strategy that reveals fine-grained inaccuracies by comparing responses generated by diverse models from semantically-equivalent prompts, and 2) a targeted mitigation technique that applies precise corrections to problematic segments while preserving accurate content. Experiments on the FELM dataset show Finch-Zk improves hallucination detection F1 scores by 6-39\% compared to existing approaches. For mitigation, Finch-Zk achieves up to 9 absolute percentage points improvement in answer accuracy on the GPQA-diamond dataset when applied to state-of-the-art models like Llama 4 Maverick and Claude 4 Sonnet. Extensive evaluation on multiple datasets demonstrates that Finch-Zk provides a practical, deployment-ready safeguard for enhancing factual reliability in production LLM systems.
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。