Skip to content

An Invertible Neural Network using Variational-Inference to estimate the model uncertainty

Notifications You must be signed in to change notification settings

marcobellagente93/Bayesian_INN

Repository files navigation

Bayesian_INN

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

Usage

# 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

About

An Invertible Neural Network using Variational-Inference to estimate the model uncertainty

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages