Xuanqing Liu

What's your lambda value? It seems that you are not using the real data to train. And is it adversarial training?

But this is the co-training part (step 1), not fine-tuning. For # training epochs required in ImageNet-143 data, you can refer to Figure 8 in our paper. It takes ~60...

> i feel sorry for my understanding. i have read your readme file. > > 1. con train with (real_adv+fake) Co-train is to train a good GAN model > 2.finetune(real_adv+fake_adv)...

I didn't test L2 norm attacks, but I think the results should be the same. The only change we need to make is to replace linf_sgd.py in /attacker directory with...

@FlashTek To your first question, [Line 33](https://github.com/xuanqing94/BayesianDefense/blob/master/models/vgg_vi.py#L33) adds all KL terms. I will try your new PGD-attack later. I noticed the potential problem in the attack a month ago, but...

Oh sorry, that's a bug, I will fix it soon. It is supposed to be something in aaron_vi.py Overall, there are two major problems in this repo. First I will...

@FlashTek I have some initial experiments after fixing the bugs, the original hyperparameters are no longer suitable. So I am still finding better ones (specifically the original alpha value is...

Hi @sz85512678 , I am sorry about the missing file. This file is basically variable-by-variable copy from the trained transformer to the newly architected transformer. Note that all the layers...