Why did you remove the max pooling layer after the initial convolution?
Zhengyu-Li opened this issue · comments
I remember there is a 3*3 max pooling layer after the initial convolution. Why remove it?
Hi!
Max pooling layer is used after 7x7 convolution layer only for the ImageNet version. I've implemented cifar/svhn version, where initially only 3x3 convolution is used (link to the authors code).