Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
AdvBenchgithub.comAdvbench is a benchmark and dataset collection for security-oriented adversarial NLP, released with the EMNLP 2022 paper. It provides a standardized platform to evaluate adversarial attacks and defenses on real-world security tasks, emphasizing practical attacker goals over imperceptibility.
Parse Score
