/Wasserstein-AutoEncoders

PyTorch implementation of Wasserstein Auto-Encoders

Primary LanguagePythonMIT LicenseMIT

Introduction

This is the implementation of Wasserstein Auto-Encoders paper in PyTorch.

I rewrite the model in DC-GAN for the encoder/decoder/discriminator instead of simple MLP

Requirement

  • python 3
  • PyTorch >= 0.4
  • torchvision
  • numpy
  • tqdm

Train

  • To train a WAE-GAN:
python wae_gan.py
  • To train a WAE-MMD:
python wae_mmd.py

Randomly Generated Images (100 Epochs)

WAE-MMD

WAE_MMD

WAE-GAN

WAE_GAN