M365

该数据集包含了大约10000个问答对，其中包括这些问答对所依据的上下文信息。该数据集被用于评估在基于验证的RAG上下文生成准确回应时，可信赖的大语言模型及其幻觉纠正算法的有效性。这项任务的目的是对大语言模型的性能进行地面文本生成和评估。

This dataset comprises approximately 10,000 question-answer pairs, along with the supporting context information for each pair. It is utilized to evaluate the effectiveness of trustworthy large language models (LLMs) and their hallucination correction algorithms when generating accurate responses through validation-based Retrieval-Augmented Generation (RAG). The objective of this task is to conduct ground-truth text generation and evaluation to assess the performance of large language models.