HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models

TL;DR

HaluEval is a comprehensive benchmark designed to evaluate large language models' tendency to hallucinate and their ability to recognize fabricated content, utilizing a large dataset of human-annotated hallucinated samples generated via a ChatGPT-based framework.

Contribution

The paper introduces HaluEval, a large-scale hallucination evaluation benchmark, and proposes a novel ChatGPT-based sampling-filtering framework for generating and annotating hallucinated samples.

Findings

ChatGPT generates hallucinations in about 19.5% of responses.

LLMs struggle to recognize hallucinations in texts.

External knowledge and reasoning improve hallucination detection.

Abstract

Large language models (LLMs), such as ChatGPT, are prone to generate hallucinations, i.e., content that conflicts with the source or cannot be verified by the factual knowledge. To understand what types of content and to which extent LLMs are apt to hallucinate, we introduce the Hallucination Evaluation benchmark for Large Language Models (HaluEval), a large collection of generated and human-annotated hallucinated samples for evaluating the performance of LLMs in recognizing hallucination. To generate these samples, we propose a ChatGPT-based two-step framework, i.e., sampling-then-filtering. Besides, we also hire some human labelers to annotate the hallucinations in ChatGPT responses. The empirical results suggest that ChatGPT is likely to generate hallucinated content in specific topics by fabricating unverifiable information (i.e., about $19.5\%$ responses). Moreover, existing LLMs face great challenges in recognizing the hallucinations in texts. However, our experiments also prove that providing external knowledge or adding reasoning steps can help LLMs recognize hallucinations. Our benchmark can be accessed at https://github.com/RUCAIBox/HaluEval.