SimpleGAN

simplegan is a python library built using TensorFlow which aims to ease the training of generative models with high level APIs. These high level APIs are useful to people who are getting started with learning/training generative models and to people who are interested in running experiments by tweaking various parameters of the model to quantify the performance.

Getting Started:

• Installation
  • Pip Installation
  • Install from Source
• Dependencies
• Examples
  • Vanilla Autoencoder
  • DCGAN
  • Pix2Pix
• Motivation
• License

Documentation:

• simplegan.autoencoder
  • Vanilla Autoencoder
  • Convolutional Autoencoder
  • Variational Autoencoder
  • Vector-Quantized Variational Autoencoder
• simplegan.gan
  • Vanilla GAN
  • DCGAN
  • Wasserstein GAN
  • Conditional GAN
  • InfoGAN
  • Pix2Pix
  • CycleGAN
  • VoxelGAN(3DGAN)
  • Self Attention GAN
• simplegan.datasets
  • MNIST Dataloader
  • Cifar10 Dataloader
  • Cifar100 Dataloader
  • LSUN Dataloader
  • Custom Dataloader
  • Pix2Pix Dataloader
  • CycleGAN Dataloader
  • Voxels Dataloader
• simplegan.losses
  • VanillaAutoencoder losses
  • GAN losses
  • InfoGAN losses
  • Pix2Pix losses
  • CycleGAN losses
  • Wasserstein losses
  • Hinge losses
• simplegan.metrics
  • Inception Score
  • Frechet Inception Distance