用弱监督方法评估大模型上下文可信度，避免错误信息传播。
CrEst: Credibility Estimation for Contexts in LLMs via Weak Supervision
- 通过文档间语义一致性自动判断上下文可信度
- 在5个数据集上最高提升26.86%准确率
- 适合需增强推理可靠性的知识密集型任务
上下文信息的融合显著提升了大语言模型在知识密集型任务上的表现。然而，现有方法常忽视一个关键挑战：上下文文档的可信度差异巨大，可能引发不可靠信息的传播。本文提出CrEst，一种无需人工标注的弱监督框架，用于在大模型推理时评估上下文文档的可信度。其核心思想是：可信文档更倾向于与其他可信文档保持高语义一致性，从而通过文档间的共识实现自动化评估。为将可信度融入大模型推理，我们设计两种策略：适用于无内部参数访问的黑盒方法，以及可直接修改注意力机制的白盒方法。在三种模型架构和五个数据集上的实验表明，CrEst持续优于强基线，准确率最高提升26.86%，F1分数提高3.49%。进一步分析显示，该方法在高噪声环境下仍保持鲁棒性能。
原文摘要 · Abstract (English)
The integration of contextual information has significantly enhanced the performance of large language models (LLMs) on knowledge-intensive tasks. However, existing methods often overlook a critical challenge: the credibility of context documents can vary widely, potentially leading to the propagation of unreliable information. In this paper, we introduce CrEst, a novel weakly supervised framework for assessing the credibility of context documents during LLM inference--without requiring manual annotations. Our approach is grounded in the insight that credible documents tend to exhibit higher semantic coherence with other credible documents, enabling automated credibility estimation through inter-document agreement. To incorporate credibility into LLM inference, we propose two integration strategies: a black-box approach for models without access to internal weights or activations, and a white-box method that directly modifies attention mechanisms. Extensive experiments across three model architectures and five datasets demonstrate that CrEst consistently outperforms strong baselines, achieving up to a 26.86% improvement in accuracy and a 3.49% increase in F1 score. Further analysis shows that CrEst maintains robust performance even under high-noise conditions.
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。