arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.19113cs.CLcs.AI
收藏

FACTOID:面向幻觉检测的事实蕴含

FACTOID: FACtual enTailment fOr hallucInation Detection

AI Institute, University of South Carolina(南卡罗来纳大学AI学院);Islamic University of Technology(伊斯兰技术大学);National Institute of Technology, Silchar(锡尔卡尔国家理工学院);Indira Gandhi Delhi Technical University for Women(英迪拉·甘地德里女子技术大学);Stanford University(斯坦福大学);Amazon AI(亚马逊人工智能)

Vipula Rawte, S. M Towhidul Islam Tonmoy, Krishnav Rajbangshi, Shravani Nag, Aman Chadha, Amit P. Sheth, Amitava Das

AI总结:

本文提出事实蕴含(FE)任务及FACTOID基准,通过多任务学习框架自动检测并定位LLM生成内容中的幻觉,并据此评估和排名15个现代LLM。

摘要:

大语言模型(LLMs)的广泛采用带来了诸多益处。然而,幻觉是一个重大关切。为此,检索增强生成(RAG)已成为一种极具前景的范式,通过将LLM输出建立在事实信息之上来改进其输出。RAG依赖文本蕴含(TE)或类似方法,来检查LLM生成的文本相对于检索到的文档是得到支持还是相互矛盾。本文认为,传统的TE方法不足以发现LLM生成内容中的幻觉。例如,考虑一个关于“美国在乌克兰战争上的立场”的提示。AI生成的文本写道:“……美国总统巴拉克·奥巴马表示美国不会向乌克兰派兵……”然而,在战争期间,美国总统是乔·拜登,这与事实现实相矛盾。此外,当前的TE系统无法准确标注给定文本并识别出被矛盾的确切部分。为解决这一问题,我们引入了一种新型TE,称为“事实蕴含(FE)”,旨在检测LLM生成内容中的事实不准确之处,同时突出与事实现实相矛盾的具体文本片段。我们提出了FACTOID(FACTual enTAILment for hallucInation Detection),一个用于FE的基准数据集。我们提出了一个用于FE的多任务学习(MTL)框架,融合了最先进的(SoTA)长文本嵌入,如e5-mistral-7b-instruct,以及GPT-3、SpanBERT和RoFormer。与SoTA TE方法相比,所提出的用于FE的MTL架构在FACTOID基准上平均实现了40%的准确率提升。由于FE能够自动检测幻觉,我们评估了15个现代LLM,并使用我们提出的自动幻觉脆弱性指数(HVI_auto)对它们进行排名。该指数量化并提供了一种比较尺度,以根据LLM的幻觉情况对其进行评估和排名。

英文摘要:

The widespread adoption of Large Language Models (LLMs) has facilitated numerous benefits. However, hallucination is a significant concern. In response, Retrieval Augmented Generation (RAG) has emerged as a highly promising paradigm to improve LLM outputs by grounding them in factual information. RAG relies on textual entailment (TE) or similar methods to check if the text produced by LLMs is supported or contradicted, compared to retrieved documents. This paper argues that conventional TE methods are inadequate for spotting hallucinations in content generated by LLMs. For instance, consider a prompt about the 'USA's stance on the Ukraine war''. The AI-generated text states, ...U.S. President Barack Obama says the U.S. will not put troops in Ukraine...'' However, during the war the U.S. president is Joe Biden which contradicts factual reality. Moreover, current TE systems are unable to accurately annotate the given text and identify the exact portion that is contradicted. To address this, we introduces a new type of TE called ``Factual Entailment (FE).'', aims to detect factual inaccuracies in content generated by LLMs while also highlighting the specific text segment that contradicts reality. We present FACTOID (FACTual enTAILment for hallucInation Detection), a benchmark dataset for FE. We propose a multi-task learning (MTL) framework for FE, incorporating state-of-the-art (SoTA) long text embeddings such as e5-mistral-7b-instruct, along with GPT-3, SpanBERT, and RoFormer. The proposed MTL architecture for FE achieves an avg. 40\% improvement in accuracy on the FACTOID benchmark compared to SoTA TE methods. As FE automatically detects hallucinations, we assessed 15 modern LLMs and ranked them using our proposed Auto Hallucination Vulnerability Index (HVI_auto). This index quantifies and offers a comparative scale to evaluate and rank LLMs according to their hallucinations.

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。

↑