WGAN-GP

An pytorch implementation of Paper "Improved Training of Wasserstein GANs".

Prerequisites

Python, NumPy, SciPy, Matplotlib A recent NVIDIA GPU

A latest master version of Pytorch

Progress

• gan_toy.py : Toy datasets (8 Gaussians, 25 Gaussians, Swiss Roll).(Finished in 2017.5.8)

• gan_language.py : Character-level language model (Discriminator is using nn.Conv1d. Generator is using nn.Conv1d. Finished in 2017.6.23. And Results is coming soon.)

• gan_mnist.py : MNIST (Running Results while Finished in 2017.5.11. Discriminator is using nn.Conv1d. Generator is using nn.Conv1d. And New Results is coming soon.)

• gan_64x64.py: 64x64 architectures(Looking forward to your pull request)

• gan_cifar.py: CIFAR-10(Looking forward to your pull request)

Results

• Toy Dataset

  Some Sample Result, you can refer to the results/toy/ folder for details.

  • 8gaussians 154500 iteration

  

  • 25gaussians 48500 iteration

    

  • swissroll 69400 iteration

  

• Mnist Dataset

  Some Sample Result, you can refer to the results/mnist/ folder for details.

  

  

  

Acknowledge

Based on the implementation igul222/improved_wgan_training and martinarjovsky/WassersteinGAN