This repository contains the code to reproduce the results of Understanding Event-Generation Networks via Uncertainties https://arxiv.org/pdf/2104.04543.pdf
Implementation is in PyTorch 1.8.0. Features included are:
- Bayesian linear layers
- MMD loss
- Train on weighted points
# clone the repository
git clone git@github.com:marcobellagente93/Bayesian_INN.git
# generate linear and quadratic toy datasets
python ./data/generate_toy.py
# Run the code
python train.py