convex_adversarial
convex_adversarial copied to clipboard
Support in dual_layers for MaxPool2d
Wondering if anyone has built (or has plans to) build support for nn.MaxPool2d?
I would use this if it were an option.