Deconv should be conv?

[sjosund]

A detail about the generator:
The last layer of the generation in this implementation is a deconv layer, while in the original implementation and as written in the paper they use a conv layer.

Line in this repo: https://github.com/goldkim92/StarGAN-tensorflow/blob/master/module.py#L28
Original: https://github.com/yunjey/StarGAN/blob/master/model.py#L51

Thanks for a nice implementation!