In the root directory, start up Simple Build Tool:

sbt

Then to run on the WSJ corpus, for example, you'll need to get the WSJ corpus, and then run (within the SBT console):

run-main nlp.cb.NgramEvaluator ../data/penn-treebank3/tagged/pos/brown/ 0.1

You're gonna need to install the SBT start-script plugin real quick, first:

https://github.com/sbt/sbt-start-script

In my ~/git/nlp.git/hooks/post-receive on the server (where I can run job scripts):

#!/bin/sh
cd /u/chbrown/nlp
env -i git pull # get rid of GITDIR setting
./sbt stage

Unused java / scala command line option: -XX:+CMSClassUnloadingEnabled