Type make to download some corpus and train a model:
$ make
...Generate some pseudo-words:
$ _venv/bin/python -m wordgen.cli generate --group Simple --count 4 _data/models.h5
unusabby
distified
shrify
eate
$ _venv/bin/python -m wordgen.cli generate --group Universe --count 4 _data/models.h5
uncorphilticlasily
dipothyrinum
unremonin
flavory
$ _venv/bin/python -m wordgen.cli generate --group Name --count 4 _data/models.h5
phill
rin
anate
bellYou can train the model using your own corpus and with different parameter:
$ make _init.ok
$ _venv/bin/python -m wordgen.cli train --token-size 4 --group MyModel models.h5 < my_corpus.txtThis example extracts ASCII words from my_corpus.txt, trains a 4-gram MCMC
transition matrix and stores it in MyModel group defined in models.h5 HDF5
file.