HALT-RAG可精准检测生成文本中的幻觉，支持多种任务且结果可信。
HALT-RAG: A Task-Adaptable Framework for Hallucination Detection with Calibrated NLI Ensembles and Abstention
- 用两个现成NLI模型和轻量信号构建通用特征集
- 在摘要、问答、对话任务上F1分别达0.7756、0.9786、0.7391
- 概率校准支持安全退避，适合高可靠性场景
生成语言模型输出中与源文本矛盾或无依据的内容检测，是其安全部署的关键挑战。本文提出HALT-RAG，一种后处理验证系统，用于识别检索增强生成（RAG）流水线输出中的幻觉。该灵活、任务自适应的框架采用由两个冻结的现成自然语言推理（NLI）模型及轻量级词汇信号组成的通用特征集，并训练一个简单、校准过的任务自适应元分类器。通过严格的五折留一（OOF）训练协议防止数据泄露，获得无偏估计，我们在HaluEval基准上评估系统表现。通过将通用特征集与轻量级任务自适应分类器及精度约束决策策略结合，HALT-RAG在摘要、问答、对话任务上的OOF F1分数分别为0.7756、0.9786和0.7391。系统校准的概率分布支持实用的退避机制，为平衡模型性能与安全性提供可靠工具。
原文摘要 · Abstract (English)
Detecting content that contradicts or is unsupported by a given source text is a critical challenge for the safe deployment of generative language models. We introduce HALT-RAG, a post-hoc verification system designed to identify hallucinations in the outputs of Retrieval-Augmented Generation (RAG) pipelines. Our flexible and task-adaptable framework uses a universal feature set derived from an ensemble of two frozen, off-the-shelf Natural Language Inference (NLI) models and lightweight lexical signals. These features are used to train a simple, calibrated, and task-adapted meta-classifier. Using a rigorous 5-fold out-of-fold (OOF) training protocol to prevent data leakage and produce unbiased estimates, we evaluate our system on the HaluEval benchmark. By pairing our universal feature set with a lightweight, task-adapted classifier and a precision-constrained decision policy, HALT-RAG achieves strong OOF F1-scores of 0.7756, 0.9786, and 0.7391 on the summarization, QA, and dialogue tasks, respectively. The system's well-calibrated probabilities enable a practical abstention mechanism, providing a reliable tool for balancing model performance with safety requirements.
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。