/Pytorch-WGANGP

Pytorch implementation of Improved Training of Wasserstein GANs or WGAN-GP (Wasserstein GAN with Gradient Penalty) using DCGAN architecture for generating 64x64 images.

Primary LanguagePython

Pytorch-WGANGP

Pytorch implementation of Improved Training of Wasserstein GANs or WGAN-GP (Wasserstein GAN with Gradient Penalty) using DCGAN architecture for generating 64x64 images.

LSUN Dataset

To download LSUN dataset follow the steps at https://github.com/fyu/lsun


Change the DB variable to change the dataset.

For using the saved model to generate images, set LOAD_MODEL to True and EPOCHS to 0.

Generated Samples

LSUN-Bedroom

LSUN-Church

CelebA