line 369：
test_minst() ——>> test_mnist()

Hi, is the black-box attack is the code corresponding to OptAttack in paper QUERY-EFFICIENT HARD-LABEL BLACK-BOX ATTACK: AN OPTIMIZATION-BASED APPROACH published on ICLR2019?

model.py

line 469:
change
test(net, test_loader) ----->test_mnist(net, test_loader).

line 369:
change
test_minst(net, test_loader) ----->test_mnist(net, test_loader).

maybe error 2 need not to change.It is just a suggestion.
Thanks very much for your sharing your code about black-box adversarial examples.you know that the sharing about black-box adversarial examples is rarely.I also have some questions about the framework Pythorch you used that is only deployed in linux？I deploy the code in my windows,and still have an error “cuda runtime error (48) : no kernel image is available for execution on the device at”，I do not find an useful suggestion for me to solve this troublesome problem.
I will look forward to your reply very much.

RuntimeError: cuda runtime error (48) : no kernel image is available for execution on the device at d:\pytorch\pytorch\torch\lib\thcunn\generic/Threshold.cu:34

I want to know,do you have met this problem?

at https://github.com/LeMinhThong/blackbox-attack/blob/57dd80c7e390baeba70105fb55a9a6e4a5b77e78/boundary_attack.py#L178

looks like distortion_fix_sample should be distortion_fixsample for line https://github.com/LeMinhThong/blackbox-attack/blob/57dd80c7e390baeba70105fb55a9a6e4a5b77e78/boundary_attack.py#L191

or vise varsa.

Hi, I found that your reimplementation for boundary-attack is quite different from the original version. For example, in your boundary_attack.py[line 72--80], you didn't use any ``orthogonal_perturbation'', which is a core idea in boundary-attack[see https://github.com/greentfrapp/boundary-attack/blob/5ce924cf62a041b218cd4a0d44a5e0d7c7619813/demo/server/controllers/boundaryattack_controller.py#L87].

Maybe there're some misunderstandings, and I would like to ask if you could possiblely explain this issue. Thank you very much :)