Progressive Growing of GANs implemented with chainer
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
LICENSE first commit Nov 7, 2017


Progressive Growing of GANs implemented with chainer

python 3.5.2 + chainer 3.0.0



$ python3 ./ -g 0 --dir ./train_images/ --epoch 100 --depth 0 

You can train models with ./

When depth = n, generated images are 2^{n+2} x 2^{n+2} size.

$ ./ 100 automatically trains models gradually (through 4 x 4 to 256 x 256).

You should tune delta and epoch when it changes too quickly or too slowly.


$ python3 ./ --gen results/gen --depth 0

You can generate images with

$ wget
$ python3 ./ --gen gen_yui_model --depth 6

You can use the pre-trained model.

It generates 256 x 256 size Ichii Yui's images.



The original paper


WGAN-GP implemented with chainer.


My Blog post related to this repository.