Qin Jian

Results 3 comments of Qin Jian

I think so. Especially, you set shuffle dataset to True. In this case, every epoch is different, even if you do not use data augmentation.

Hi, I've ran the experiment with abs op. But found out the ranking prefer to prune the shallow layer of net. Maybe there are mistakes in my experiment settings. After...

Hi, I have my own implementation. It also prefers shallow layer. Maybe we need contact the paper's author for more details. Seems you speak Chinese. I think we can keep...