MadryLab / cifar10_challenge

A challenge to explore adversarial robustness of neural networks on CIFAR10.
MIT License
488 stars 133 forks source link

About the network architecture #23

Closed dongyp13 closed 4 years ago

dongyp13 commented 4 years ago

Hi,

I have a question about the network architecture. As you commented in https://github.com/MadryLab/cifar10_challenge/blob/master/model.py#L50, it is a w28-10 wide residual network. However, when I looked at the original implementation of wide resnet at https://github.com/tensorflow/models/blob/master/research/resnet/resnet_model.py#L87-L92, the hps.num_residual_units needs to be set as 4. But it is set as 5 in your code, which corresponds to a w34-10 network?

dtsip commented 4 years ago

Thank for catching this! We probably missed this part of the code when making the model wider. I am not super familiar with the wresnet terminology so this might indeed not be a w28-10 model. I will update the comment to avoid further confusion.