Adversarial Robustness Python Packages

robustbench

RobustBench: a standardized adversarial robustness benchmark [NeurIPS 2021 Benchmarks and Datasets Track]

pyautoattack

Code relative to "Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks"

protransformers

[NeurIPS 2024] ProTransformer: Robustify Transformers via Plug-and-Play Paradigm. A 4-line robust attention layer that plugs into any pretrained transformer (BERT, ViT, LLaMA, Vicuna, T5, GAT) without retraining.

visprobe

Compositional robustness testing for vision models