advertorch v0.1: An Adversarial Robustness Toolbox based on PyTorch

20 Feb 2019Gavin Weiguang DingLuyu WangXiaomeng Jin

advertorch is a toolbox for adversarial robustness research. It contains various implementations for attacks, defenses and robust training methods... (read more)

PDF Abstract

Evaluation Results from the Paper


 SOTA for Adversarial Attack on 1B Words (using extra training data)

     Get a GitHub badge
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK USES EXTRA
TRAINING DATA
COMPARE
Adversarial Attack 1B Words xyz 14 gestures accuracy 0.9 # 1