There are several implementations of wideResNet in pytorch, but they all seem to crash when used on Imagenet. This is because they expect in input a 32s32 image, and imagenet has images in the form 224x224.
From what I can see, even on the github repo of the original authors of the wideResNet paper (https://github.com/szagoruyko/wide-residual-networks/tree/master/pytorch) there is this error, i.e. it crashed on imagenet.
Has anyone managed to make it work? Do you know where I can find an implementation that works on imagenet?