Gibbs MedLDA is a Max-margin Topic Model framework developed under Yahoo_LDA framework. It supports large-scale multi-class and multi-label classification by utilizing multi-core and multi-machine parallelism. This project is released under the Apache License, Version 2.0.

Downloads

• Download the readme.txt
• Download the source code GibbsMedLDA.tar.gz.
• We also provide a non-parallel Gibbs MedLDA, which does not depend on 3rd party libraries. Download the source code GibbsMedLDA-NonParallel.tar.gz.

Update News

• Released the parallel Gibbs sampling code for Gibbs MedLDA topic model on March 10, 2013;
• Released the non-parallel Gibbs sampling code for Gibbs MedLDA topic model on March 10, 2013;

References

1. Jun Zhu, Ning Chen, Hugh Perkins, and Bo Zhang. Gibbs Max-margin Topic Models with Fast Sampling Algorithms, In ICML, Atlanta, USA, 2013.