pints-ai/Finetune-RAG

Finetune-RAG数据集是针对减少检索增强大型语言模型中的虚构内容问题而创建的，由合成策划和处理的RAG文档组成，可用于大型语言模型的微调。每个数据条目包含正确的内容块、原始文档文件名、两个误导性的虚假文档文件名和内容块、用户查询、基于正确内容的GPT-4o答案，以及可选的前后内容。

The Finetune-RAG dataset is created to address the issue of hallucination in retrieval-augmented large language models, consisting of synthetically curated and processed RAG documents that can be used for fine-tuning language models. Each entry includes a correct content chunk, the original document filename, two fictitious document filenames and content chunks, a user query, a GPT-4o answer based on the correct content, and optional preceding and succeeding content.