Prines opened this issue 4 years ago · 0 comments
i think the num of params for cifar10 residual network is incorrect, i find that it is much bigger than the num in paper