Why using up-pooling instead of 2 stride convtranspose when UP

znxlwm / UGATIT-pytorch

Official PyTorch implementation of U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation

MIT License

2.51k stars 476 forks source link

Why using up-pooling instead of 2 stride convtranspose when UP #43

Open annihi1ation opened 4 years ago

annihi1ation commented 4 years ago

Hey professor. I remember that the 2 stride conv is more stable than up-pooling (according to DCGAN). However, in your model, the downsampling part is 2 stride conv, while up-pooling in upsampling part.

Could you give us a reason for this? Looking forward to your reply

jayzhan211 commented 4 years ago

this has been discussed in cyclegan

https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/issues/190