This repository provides PyTorch implementations of various approximate inference methods for Bayesian neural networks. It includes algorithms like Bayes by Backprop, Monte Carlo Dropout, Stochastic Gradient Langevin Dynamics, Preconditioned SGLD, Kronecker-Factorised Laplace approximation, and Stochastic Gradient Hamiltonian Monte Carlo. The code supports regression and classification experiments on toy datasets and UCI datasets, focusing on uncertainty quantification and out-of-distribution detection in deep learning models.