Open source code

MALLET. The MALLET toolkit implements classification, sequence tagging, and topic modeling algorithms. I wrote the topic model package. Development site

Sparse stochastic inference. This class trains a topic model with online variational inference, but with miniature Gibbs samplers in the inner loop. For details see this paper. This class depends on MALLET and some of the code below. Java README

Unicode barplots. This class produces low-res barplots that can be printed in a terminal window using Unicode characters. I use it for visualizing probability distributions from the command line. Java