lianghsun/RAG-Instruct

RAG-Instruct是一个基于Wikipedia语料库的RAG数据集，旨在全面增强大型语言模型(LLM)的RAG能力。该数据集利用GPT-4o合成，并具有查询-文档场景和任务多样性，能够显著提高LLM在各种任务中的RAG性能。RAG-Instruct is a RAG dataset based on the Wikipedia corpus, designed to comprehensively enhance the RAG capabilities of large language models (LLMs). synthesized using GPT-4o, the dataset offers diversity in both query-document scenarios and tasks, significantly improving the RAG performance of LLMs across various tasks.