Golem is a minimalistic framework for performing Bayesian analysis. It was heavily inspired by Richard McElreath's Statistical Rethinking course, which is a lovely journey into the scientific thinking. It's built on top of PyTorch, and more specifically, it uses the torch.distributions module and the available optimization routines.
My main goal is to have a simple interface to sketch ideas and test them quickly. Caution is advise as the tool should be mainly used didactically to get a better grasp of the Bayesian paradigm.
tbd
tbd