PrimeVul

Evals

Benchmark / dataset

Clean vulnerability-detection dataset; shows models are weak at it.

Category
Secure code & vuln detection
Insurance exposure
Breach/cyber
Publisher
Columbia / Google DeepMind
Publisher type
Academic
License
MIT
Status
Stale (>12mo)
Last commit
2024-09-07
Stars
279

View source on GitHub →

Wanich Risk Lab, 2026.