Bayesian Neural Networks with MCMC in Pytorch

Support project for my research in Bayesian Neural Networks for modeling human behavior. The project is currently focused on Markov chain Monte Carlo (MCMC) with various sampling methods including HMC, SG-HMC, SG-LD, and cyclical learning rates. A key feature is the rapid inference through tensor batching of samples simultaneously.

Code for Variational Inference and Monte Carlo Dropout are included but not actively maintained.

Source code: Github