LDA using Gibbs sampling with hyperparameter optimization based on Minka's fixed-point iteration (Equation 55). The code is a modified version of GibbsLDA++, a LDA implementation in C++.
The usage is the same as in the original implementation of GibbsLDA++, except for alpha and beta values which are not parameters anymore. A detailed GibbsLDA++'s usage description can be found on its website under the Usage tab.