Skip to content

Dataset · AI4Sec

Cyber Threat Intelligence (CTI) benchmark for LLMs

AI4Sec/cti-bench

English test suite for evaluating language models' knowledge in cybersecurity.

What's inside

An English text benchmark of 1,000 to 10,000 examples designed to test LLMs in Cyber Threat Intelligence. It includes multiple components, including multiple-choice tests (CTI-MCQ) on standards, threats, and detection methods.

What you can do with it

  • Evaluating language model competencies in information security and threat analysis
  • Comparing various LLMs in incident classification and cybersecurity question answering tasks
Tasks
Zero shot classification, Question answering, Text classification
Languages
English
Size
1K–10K rows
Downloads this month
2.2K
Likes
22

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("AI4Sec/cti-bench")