leeroy-jankins/NIST-GenAI-Profile

该数据集包含从美国国家标准与技术研究院（NIST）出版物《NIST AI 600-1，人工智能风险管理框架：生成式人工智能概况》中提取的问题-答案记录。源出版物是NIST人工智能风险管理框架的跨行业配套资源，重点关注生成式人工智能系统所特有或加剧的风险。数据集设计用于监督微调、检索增强生成评估、指令调优、文档基础问答以及人工智能治理培训等用例。每条记录包含一个用户风格的问题和一个基于源文档的助理风格答案。答案以专业解释性风格撰写，旨在鼓励健壮、上下文相关的回答，而非简短的提取式答案。

This dataset contains question-and-answer records derived from the National Institute of Standards and Technology publication **NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile**. The source publication is a cross-sectoral companion resource to the NIST AI Risk Management Framework and focuses on risks that are unique to, or exacerbated by, generative artificial intelligence systems. The dataset is designed for supervised fine-tuning, retrieval-augmented generation evaluation, instruction tuning, document-grounded question answering, and AI governance training use cases. Each record contains a user-style question and an assistant-style answer grounded in the source document. The answers are written in a professional explanatory style and are intended to encourage robust, contextual responses rather than short extractive answers.