Add an easy to use command line tool to easily train models from large amounts of files, preferably in a format like the [BBC datasets](http://mlg.ucd.ie/datasets/bbc.html).