https://github.com/fabriziomusacchio/wasserstein_gan

This repository contains some code for demonstrating the application of Wasserstein GANs (WGANs)
https://github.com/fabriziomusacchio/wasserstein_gan

gan gans generative-adversarial-network wasserstein-distance wasserstein-gan wasserstein-gans wgan wgans

Last synced: 24 days ago
JSON representation

This repository contains some code for demonstrating the application of Wasserstein GANs (WGANs)

• Host: GitHub
• URL: https://github.com/fabriziomusacchio/wasserstein_gan
• Owner: FabrizioMusacchio
• Created: 2023-07-30T00:16:06.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-07-30T21:52:05.000Z (over 1 year ago)
• Last Synced: 2024-11-13T09:12:06.883Z (3 months ago)
• Topics: gan, gans, generative-adversarial-network, wasserstein-distance, wasserstein-gan, wasserstein-gans, wgan, wgans
• Language: Python
• Homepage: https://www.fabriziomusacchio.com/blog/2023-07-29-wgan/
• Size: 18.6 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Wasserstein GANs

This repository contains some code for demonstrating the application of Wasserstein GANs (WGANs). The code is used in the following blog posts:

* [Wasserstein GANs](https://www.fabriziomusacchio.com/blog/2023-07-29-wgan/)

* [Eliminating the middleman: Direct Wasserstein distance computation in WGANs without discriminator](https://www.fabriziomusacchio.com/blog/2023-07-30-wgan_with_direct_wasserstein_distance/)

* [Conditional GANs](https://www.fabriziomusacchio.com/blog/2023-07-30-cgan/)

For further details, please refer to these posts.

Results of training a default GAN on the MNIST dataset for 50 epochs:

![gif](GAN_images/depp_conv_gan.gif)

Results of training a Wasserstein GAN on the same dataset:

![gif](WGAN_images/depp_conv_wgan.gif)

Results of training a Wasserstein GAN using the POT library, avoiding the necessity of a discriminator:

![gif](GAN_demo_images/cross_animation.gif)

Results of training a conditional GAN:

![gif](cGAN_demo_images/cGAN_animation_edited.gif)

For reproducibility:

```powershell

conda create -n gan -y python=3.9

conda activate gan

conda install mamba -y

mamba install -y numpy matplotlib scikit-learn scipy pot tensorflow imageio pillow ipykernel

mamba install -y pytorch torchvision -c pytorch

pip install POT

```

If you want to run the code on a Mac with Apple Silicon (M1, M2), install tensorflow and pytorch as described here:

* [How to run TensorFlow on the M1 Mac GPU](https://www.fabriziomusacchio.com/blog/2022-11-10-apple_silicon_and_tensorflow/)

* [How to run PyTorch on the M1 Mac GPU](https://www.fabriziomusacchio.com/blog/2022-11-18-apple_silicon_and_pytorch/)