Chinese-xingyi

@nizihan @hewumars 我的理解是这样的： 根源在于prune.py中的剪枝函数的剪枝操作。conv层删除相应的filter，next_conv也需要跟着作调整——1，next_conv输入的通道数变少；2,输入通道的序号，比如conv中6号filter被删除，8号filter被删除，那么next_conv原来的7号现在应该是6,原来的9号，现在应该是7。

I think the learning-rate and the epochs when I need to modified is specific for different models and different datasets. I have tried to do some test about vgg19_bn on...