Details about training vgg in the paper?

Roll920 / ThiNet

caffe model of ICCV'17 paper - ThiNet: A Filter Level Pruning Method for Deep Neural Network Compression https://arxiv.org/abs/1707.06342

144 stars 40 forks source link

Details about training vgg in the paper? #10

Closed Eric-mingjie closed 6 years ago

Eric-mingjie commented 6 years ago

Hi, thanks for the released model!
I have one question though, how do you train the original vgg model? Or do you use a pretrained model? Also, i notice that in your paper, there are some experiments on training the pruned model from scratch, could you share with me the training detail?