Bayes-stack is a framework for parallel probabilistic inference on
graphical models. The framework provides infrastructure for easily
implementing MCMC/Gibbs sampling methods capable of scaling to dozens
of cores.
Along with the framework itself, several models using blocked Gibbs
sampling are provided in network-topic-models/. See documentation: blob/stable/doc/usage.markdown