The quickest way to use Joshua is to download a pre-built model and use them to start translating data.
Building your own models takes a bit more work, and requires you to supply parallel data that the models can be trained from. Information about how to do this can be found in the pipeline documentation.
Our pipeline.html">pipeline script is the quickest way to get started. For example, to train and test a complete model translating from Bengali to English:
First, download the Indian languages data:
curl -#L https://github.com/joshua-decoder/indian-parallel-corpora/tarball/master > indian-languages.tgz tar xf indian-languages.tgz ln -s joshua-decoder-indian-parallel-corpora-b71d31a input
Then, train and test a model
$JOSHUA/bin/pipeline.pl --source bn --target en \
--no-prepare --aligner berkeley \
--corpus input/bn-en/tok/training.bn-en \
--tune input/bn-en/tok/dev.bn-en \
--test input/bn-en/tok/devtest.bn-en
This will align the data with the Berkeley aligner, build a Hiero model, tune with MERT, decode the test sets, and reports results that should correspond with what you find on the Indian Parallel Corpora page. For more details, including information on the many options available with the pipeline script, please see its documentation page.
For more detail on the decoder itself, including its command-line options, see the Joshua decoder page. You can also learn more about other steps of the Joshua MT pipeline, including grammar extraction with Thrax and Joshua's efficient grammar representation.
If you have problems or issues, you might find some help on our answers page or in the mailing list archives.
A bundled configuration, which is a minimal set of configuration, resource, and script files, can be created and easily transferred and shared. (This is what we use to distribute pre-built models).