Evaluating Hallucinations in Chinese Large Language Models

Anonymous

Abstract: In this paper, we establish a benchmark named HalluQA (Chinese Hallucination Question-Answering) to measure the hallucination phenomenon in Chinese large language models. HalluQA contains 450 meticulously designed adversarial questions, spanning multiple domains, and takes into account Chinese historical culture, customs, and social phenomena. During the construction of HalluQA, we consider two types of hallucinations: imitative falsehoods and factual errors, and we construct adversarial samples based on GLM-130B and ChatGPT.For evaluation, we design an automated evaluation method using GPT-4 to judge whether a model output is hallucinated.We conduct extensive experiments on 24 large language models, including ERNIE-Bot, Baichuan2, ChatGLM, Qwen, SparkDesk and etc. Out of the 24 models, 18 achieved non-hallucination rates lower than 50\%. This indicates that HalluQA is highly challenging.We analyze the primary types of hallucinations in different types of models and their causes. Additionally, we discuss which types of hallucinations should be prioritized for different types of models.

Paper Type: long

Research Area: Resources and Evaluation

Contribution Types: Model analysis & interpretability, Data resources

Languages Studied: Chinese