The scripts here demonstrate how to use the pipeline to build different types of
models (including GHKM, Hiero, SAMT, and traditional phrase-based models).

They all use the data that is downloaded with the ../download.sh script.
