/catlog

Gradient estimation of discrete probabilistic models with the CatLog-Derivative trick

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

CatLog-Derivative trick

This repository contains the code and data generation for the experiments presented in the paper "Differentiable Sampling of Categorical Distributions Using the CatLog-Derivative Trick".

The two synthetic experiments can be found in synthetic.
The code for our discrete variational auto-encoders is given in DVAE.
The final neural-symbolic experiment is provided in ADDITION.

More details about the experiments and their implementation is given in their respective folders.