VieCRF: A Fast Toolkit For Factorial Conditional Random Fields

Abstract

The VieCRF package allows for training of factorial (i.e. "grid-structured") conditional random fields. Major features: uses belief propagation to approximate marginals or optimizes pseudolikelihood instead; parallelized training (uses all your CPU cores to speed up the process); various batch and online training algorithms; well-documented C++ and Perl APIs.