oseoko/citetrace-vm

CiteTrace 是一个包含 761,495 个可评估的（查询、来源、答案）三元组的数据集，这些数据来自 10 个搜索增强的大型语言模型（LLM）对 28 个站点（其中技术和学术社区被过度代表）的 11,200 个 Stack Exchange 问题的回答。A citation-level dataset of 761,495 evaluable (query, source, answer) triples from 10 search-augmented LLMs answering 11,200 Stack Exchange questions across 28 sites (technical and academic communities are over-represented).