Hallucinated References Dataset

该数据集包含了1000个由大型语言模型生成的候选论文标题，这些标题是通过查询ACM计算分类系统中的不同子概念而产生的。此外，该数据集能够评估语言模型在生成真实参考文献和检测虚构内容方面的能力。规模上，数据集包含了1000个候选标题，其任务在于识别和分类由语言模型生成的虚构参考文献。

This dataset contains 1,000 candidate paper titles generated by Large Language Models (LLMs), which were created by querying different sub-concepts within the ACM Computing Classification System. Additionally, this dataset can be utilized to evaluate the ability of language models to generate authentic references and detect fictional content. In terms of scale, the dataset includes 1,000 candidate titles, and its corresponding task is to identify and classify fictional references generated by language models.