bityangke/sgan

Stacked Generative Adversarial Networks: https://arxiv.org/abs/1612.04357

Stacked Generative Adversarial Networks

This repository contains code for the paper "Stacked Generative Adversarial Networks". Part of the code is modified from OpenAI's implementation of Improved GAN.

Currently only the code for MNIST experiments is available. I am working hard to polish the code for SVHN/CIFAR-10 experiments and will push them to this repository as soon as possible.

Usage

run a simple experiment with independent training only:

cd mnist
THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python train_independent.py

or with joint training (from scratch):

THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python train_joint.py

Citations

If you use the code in this repository in your paper, please cite:

@article{huang2016sgan,
  title={Stacked Generative Adversarial Networks},
  author={Huang, Xun and Li, Yixuan and Poursaeed, Omid and Hopcroft, John and Belongie, Serge},
  journal={arXiv},
  year={2016}
}