imrahulr / adversarial_robustness_pytorch Goto Github PK

Unofficial implementation of the DeepMind papers "Uncovering the Limits of Adversarial Training against Norm-Bounded Adversarial Examples" & "Fixing Data Augmentation to Improve Adversarial Robustness" in PyTorch

License: MIT License

Python 100.00%

adversarial-robustness adversarial-training deep-learning pytorch

adversarial_robustness_pytorch's People

Contributors

Stargazers

Watchers

Forkers

austinyi shwinshaker nicowangdev paulgavrikov meowrowan adverml deepaksingh98 shihuahuang95 byerose changx03 chenglinyu

adversarial_robustness_pytorch's Issues

cifar10s

Means the 's' semi supervised?
Where do get this dataset?

Value of `Tau` Differs from the actual paper.

Hi there,
I was reading through the paper and I came across page 5, in the outer optimization part, it says that the value of tau=0.999, while in the repo it is set to 0.995. Is there a reason or is it just a mistake?

Adding resume options may be more perfect.

Great implementation, it will be even more perfect if adding resume options. haha

Missing args.txt

Dear author, thanks for your PyTorch implementation. However, the missing args.txt prevents me from testing the pre-trained model provided by https://github.com/deepmind/deepmind-research/tree/master/adversarial_robustness. Thank you for your help!

adversarial_robustness_pytorch/eval-aa.py

Line 35 in 6df6a8f