RetrievalQA

该数据集是由五个其他问答基准测试中筛选出的问题子集整理而成的，旨在挑战闭卷问答的表现。所选问题包含的知识点是目前大型语言模型中尚未涵盖的，因此具有一定的挑战性。该任务的目的是进行问题回答。

This dataset is compiled from subsets of questions selected from five existing question answering benchmarks, and is intended to challenge the performance of closed-book question answering systems. The knowledge points involved in the selected questions have not yet been covered by current large language models, thus rendering the dataset reasonably challenging. The core objective of this task is question answering.