There's some documentation about "custom dialect pack", which I guess is what I want (I want to use a custom, smaller dictionary). But it's very bare and doesn't give any example... how can I have pybo tokenize according to a list of words in a txt file?