ethanolivertroy/nist-cybersecurity-training

NIST网络安全训练数据集v1.1是一个开源的、用于微调大型语言模型的网络安全训练数据集。该数据集包含了来自596个NIST出版物的结构化训练数据，包括FIPS、SP、IR和CSWP（网络安全白皮书）系列。数据集版本1.1新增了23篇网络安全白皮书，修复了6,150个DOI格式错误链接，移除了202个 malformed DOI，并且为未来的恢复工作记录了72,698个断裂链接。该数据集适用于微调LLM以获得NIST网络安全专业知识，并支持RAG应用、聊天机器人、问答系统和自动化合规性检查工具。

The NIST Cybersecurity Training Dataset v1.1 is an open-source dataset designed for fine-tuning large language models on cybersecurity expertise. It contains structured training data from 596 NIST publications, including the FIPS, SP, IR, and CSWP (Cybersecurity White Papers) series. Version 1.1 adds 23 new Cybersecurity White Papers, fixes 6,150 DOI format error links, removes 202 malformed DOIs, and catalogs 72,698 broken links for future recovery. The dataset is suitable for fine-tuning LLMs to gain NIST cybersecurity expertise and supports RAG applications, chatbots, question-answering systems, and automated compliance checking tools.