Halu-J：基于批判的幻觉评判器
Halu-J: Critique-Based Hallucination Judge
AI总结：
本文提出Halu-J，一个70亿参数的基于批判的幻觉评判器，通过选择相关证据和提供详细批判，在多证据幻觉检测中超越GPT-4o，并引入新数据集ME-FEVER。
摘要：
大型语言模型（LLMs）经常生成非事实内容，即所谓的幻觉。现有的基于检索增强的幻觉检测方法通常将其视为分类任务，根据与检索证据的一致性来评估幻觉。然而，这种方法通常缺乏对这些评估的详细解释，并且不评估这些解释的可靠性。此外，检索系统的缺陷可能导致检索到不相关或部分相关的证据，从而损害检测过程。而且，现实世界中的幻觉检测需要分析多条证据，而当前系统通常将所有证据统一对待，不考虑其与内容的相关性。为了解决这些挑战，我们引入了Halu-J，一个具有70亿参数的基于批判的幻觉评判器。Halu-J通过选择相关证据并提供详细的批判来增强幻觉检测。我们的实验表明，Halu-J在多证据幻觉检测中优于GPT-4o，并在批判生成和证据选择方面与其能力相当。我们还引入了ME-FEVER，一个专为多证据幻觉检测设计的新数据集。我们的代码和数据集可在https://github.com/GAIR-NLP/factool 找到。
英文摘要：
Large language models (LLMs) frequently generate non-factual content, known as hallucinations. Existing retrieval-augmented-based hallucination detection approaches typically address this by framing it as a classification task, evaluating hallucinations based on their consistency with retrieved evidence. However, this approach usually lacks detailed explanations for these evaluations and does not assess the reliability of these explanations. Furthermore, deficiencies in retrieval systems can lead to irrelevant or partially relevant evidence retrieval, impairing the detection process. Moreover, while real-world hallucination detection requires analyzing multiple pieces of evidence, current systems usually treat all evidence uniformly without considering its relevance to the content. To address these challenges, we introduce Halu-J, a critique-based hallucination judge with 7 billion parameters. Halu-J enhances hallucination detection by selecting pertinent evidence and providing detailed critiques. Our experiments indicate that Halu-J outperforms GPT-4o in multiple-evidence hallucination detection and matches its capability in critique generation and evidence selection. We also introduce ME-FEVER, a new dataset designed for multiple-evidence hallucination detection. Our code and dataset can be found in https://github.com/GAIR-NLP/factool .
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。
