_images/logo-torch-blue-wide.svg

A PyTorch-based library for Bayesian learning and uncertainty estimation

python-version license pypi-version codecov docs joss

torch_blue provides a simple way for non-expert users to implement and train Bayesian Neural Networks (BNNs). Currently, it only supports Variational Inference (VI), but will hopefully grow and expand in the future. To make the user experience as easy as possible most components mirror components from PyTorch.

torch_blue is available on PyPI under the name torch-blue. If you need further instructions, check the installation guide.

To learn how you can use torch_blue you can read the quickstart guide. If you have implemented your model, but struggle to get it running check our debug guide for some basic approaches to debugging a BNN.

If you find any bug, have a feature request, or want to contribute visit us on GitHub.

And finally, if you are struggling to understand any part of this documentation let us know. torch_blue is explicitly intended to be accessible for users without experience with BNNs. If you are struggling someone else is probably struggling with the same thing now or at least in the future. Make an issue, let us know what is unclear or confusing so we can try to improve it and make BNNs accessible to everyone.