Knowledge Graph-based False Premise Questions (KG-FPQ)

该数据集名为KG-FPQ，包含了大约17.8万个由知识图谱中提取的真实三元组生成的虚假前提问题，旨在评估大型语言模型在面对事实性幻觉时的脆弱性。该数据集跨越了三个知识领域和六个混淆度级别，用于评估各种大型语言模型的表现。规模上，该数据集包含了大约17.8万个虚假前提问题，并适用于判别性和生成性任务。

This dataset, named KG-FPQ, contains approximately 178,000 false premise questions generated from true triples extracted from knowledge graphs, and is designed to evaluate the vulnerability of large language models (LLMs) to factual hallucinations. Covering three knowledge domains and six levels of confounding degree, it is utilized to assess the performance of various large language models. In terms of scale, this dataset includes around 178,000 false premise questions and is applicable to both discriminative and generative tasks.