BayesDLL: Bayesian Deep Learning Library

TL;DR

BayesDLL is a PyTorch library enabling scalable Bayesian deep learning with minimal code changes, supporting large models like Vision Transformers and incorporating pre-trained weights as priors.

Contribution

It introduces a scalable Bayesian neural network library for large models, with easy integration and support for pre-trained weights as priors.

Findings

Supports large-scale models including ViTs

Requires minimal code modifications

Allows use of pre-trained weights as priors

Abstract

We release a new Bayesian neural network library for PyTorch for large-scale deep networks. Our library implements mainstream approximate Bayesian inference algorithms: variational inference, MC-dropout, stochastic-gradient MCMC, and Laplace approximation. The main differences from other existing Bayesian neural network libraries are as follows: 1) Our library can deal with very large-scale deep networks including Vision Transformers (ViTs). 2) We need virtually zero code modifications for users (e.g., the backbone network definition codes do not neet to be modified at all). 3) Our library also allows the pre-trained model weights to serve as a prior mean, which is very useful for performing Bayesian inference with the large-scale foundation models like ViTs that are hard to optimise from scratch with the downstream data alone. Our code is publicly available at: \url{https://github.com/SamsungLabs/BayesDLL}\footnote{A mirror repository is also available at: \url{https://github.com/minyoungkim21/BayesDLL}.}.