Store several original model of CNNS writen by Pytorch
it's a basic convolution network
here are basic block of inceptionv1:https://cn.bing.com/images/search?view=detailV2&id=45AF2681E9F03A94D36DC58E0C215595A2C90C62&thid=OIP.LLq33roQ395gfpMRjjZUjQHaEq&mediaurl=https%3A%2F%2Fcdn-images-1.medium.com%2Fmax%2F1600%2F1*U_McJnp7Fnif-lw9iIC5Bw.png&exph=518&expw=823&q=inceptionv1&selectedindex=4&ajaxhist=0&vt=0&eim=1,2,6
here are paper about inceptionv1:https://arxiv.org/abs/1409.4842
there are two different parts that between inceptionv1 and inceptionv2. 1) Batch Normalization 2) using two 33 kernel take place one 55 kernel because of using Batch Normalization dropout stage can be dropped In inceptionv1 there have one main output and two auxiliary output, I dropout one of these two auxiliary output. So in inceptionv2 there have one main output and one auxiliary output