Pytorch model #6

Activity

1. pinyuchen commented on Jun 9, 2021

   

   pinyuchen

   on Jun 9, 2021

   Member

   More actions

   Hello @kfarivar
   You can use the general wrapper from IBM Adversarial Robustness Toolbox: https://adversarial-robustness-toolbox.readthedocs.io/en/stable/modules/metrics.html
2. huanzhang12 commented on Jun 10, 2021

   

   huanzhang12

   on Jun 10, 2021

   Contributor

   More actions

   Thank you @pinyuchen for the answer. Let us know if you have any further questions.
3. kfarivar commented on Jun 12, 2021

   

   kfarivar

   on Jun 12, 2021

   Author

   More actions

   I have seen the wrapper. It has 2 major problems. 1. The implementation is not efficient. The original implementation uses multitasking library while theirs is a serial implementation.
   2. The implementation is incomplete. The original code estimates the c_init while that's not implemented in the wrapper.
4. huanzhang12 commented on Jun 12, 2021

   

   huanzhang12

   on Jun 12, 2021

   Contributor

   More actions

   @kfarivar I agree with you that the original implementation is more efficient. I did a lot of multiprocessing optimizations to fully utilize the GPU. The new implementation was done by an independent team which focuses more on functionality rather than performance.

   It is possible to change the original implementation to use a PyTorch model. CLEVER is a sampling based method so as long as you can integrate pytorch model inference into the code, everything else should work. More precisely, you will need to replace this tensorflow sess.run with a pytorch model inference:

   https://github.com/huanzhang12/CLEVER/blob/master/estimate_gradient_norm.py#L461

   Most other things can be kept unchanged, including the multitasking part. However it does need some efforts if you plan to take this route.
5. liuyiming199721 commented on Sep 16, 2023

   

   liuyiming199721

   on Sep 16, 2023

   More actions

   Hello, is there an easy way to use your pytorch models?