Totally Versatile Miscellania for PyTorch

Thomas Viehmann tv@lernapparat.de

This repository collects various things I have implmented for PyTorch

Layers, autograd functions and calculations

• 2d Wavelet Transformation using PyTorch convolutions and coefficents from PyWavelet
• Implicit Function Theorem and Implicit Functions in PyTorch
• Wasserstein Distance of Histograms/Distributions

Learning approaches

• Gaussian Process Regression see also candlegp, a more complete GP library for PyTorch
• 1d Gaussian Mixture Density Networks

Generative Adversarial Networks

Wasserstein GAN - See also my two blog posts on the subject

• Semi-Improved Training of Wasserstein GAN
• Improved Training of Wasserstein GAN