CRAG benchmark

该数据集是一个结合了不同领域结构化和非结构化知识的全新RAG基准，旨在评估检索增强生成性能。该数据集的设计旨在推动RAG社区未来的研究进展，并支持在现实世界场景中的可复制性和适用性。其任务是对检索增强生成框架进行评估。

This dataset is a novel RAG benchmark that combines structured and unstructured knowledge across diverse domains, designed to evaluate the performance of retrieval-augmented generation (RAG). This benchmark is engineered to advance future research in the RAG community while supporting reproducibility and applicability in real-world scenarios. Its core task is to assess retrieval-augmented generation frameworks.