Skip to content

Simple Pytorch implementations of most used Generative Adversarial Network (GAN) varieties.

Notifications You must be signed in to change notification settings

wangguanan/Pytorch-Basic-GANs

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GANs

Simple Pytorch implementations of most used Generative Adversarial Network (GAN) varieties.

GPU or CPU

Support both GPU and CPU.

Dependencies

Table of Contents

Experiment Results

Vanilla GAN (GAN)

epoch 0 epoch 10 epoch 20 epoch 30 epoch 40
xxx xxx xxx xxx xxx
epoch 50 epoch 100 epoch 150 epoch 199 -
xxx xxx xxx xxx -

Conditional GAN (cGAN)

epoch 0 epoch 10 epoch 20 epoch 30 epoch 40
xxx xxx xxx xxx xxx
epoch 50 epoch 100 epoch 150 epoch 199 -
xxx xxx xxx xxx -

Improved Conditional GAN (Improved cGAN)

epoch 0 epoch 10 epoch 20 epoch 30 epoch 40
xxx xxx xxx xxx xxx
epoch 50 epoch 100 epoch 150 epoch 199 -
xxx xxx xxx xxx -

Deep Convolutional GAN (DCGAN)

epoch 0 epoch 10 epoch 20 epoch 30 epoch 40
xxx xxx xxx xxx xxx
epoch 50 epoch 60 epoch 70 epoch 80 epoch 90
xxx xxx xxx xxx xxx

Wasserstein GAN (WGAN)

epoch 0 epoch 10 epoch 20 epoch 30 epoch 40
xxx xxx xxx xxx xxx
epoch 50 epoch 100 epoch 150 epoch 199 -
xxx xxx xxx xxx -

Wasserstein GAN with Gradient Plenty (WGAN-GP)

epoch 0 epoch 10 epoch 20 epoch 30 epoch 40
xxx xxx xxx xxx xxx
epoch 50 epoch 100 epoch 150 epoch 199 -
xxx xxx xxx xxx -

Acknowledgement

This project is going with the GAN Theory and Practice part of the Deep Learning Course: from Algorithm to Practice.

Contacts

If you have any question about the project, please feel free to contact with me.

E-mail: guan.wang0706@gmail.com

About

Simple Pytorch implementations of most used Generative Adversarial Network (GAN) varieties.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages