NIST Cybersecurity White Papers for LLM Fine-Tuning

An open-source collection of NIST cybersecurity training documents, including a CSWP series of 23 white papers. The dataset has undergone extensive link validation, with 124,946 links fixed and 72,698 broken links cataloged. It is designed for fine-tuning large language models on cybersecurity topics.

Use Cases

• Fine-tune a language model on the CSWP series of 23 cybersecurity white papers for domain-specific text generation.
• Use the validated document corpus to train a model for cybersecurity question-answering or summarization.
• Analyze the structure and content of NIST training materials to identify key technical themes and concepts.

Strengths

• Contains a specific CSWP series of 23 new cybersecurity white papers.
• Extensive data cleaning performed, including fixing 124,946 total links and cataloging 72,698 broken links.
• Version 1.1 includes fixes for 6,150 broken DOI links and removal of 202 malformed DOIs.

Limitations

• The total number of documents, rows, or overall corpus size is unknown.
• Specific column structure and file formats are not provided, limiting direct analytical use.
• The dataset's scope is limited to NIST materials, which may not cover broader cybersecurity domains.

Provenance

Source

NIST (National Institute of Standards and Technology) cybersecurity training materials.

Collection Method

Curated and cleaned from source documents, with link validation and normalization.

Freshness

Last updated on 2025-10-22.