GitLab

* Add ShuffleNet v2

Added 4 configurations: x0.5, x1, x1.5, x2
Add 2 pretrained models: x0.5, x1

* fix lint

* Change globalpool to torch.mean() call