How to train with additional corpora

[TheFausap]

Hello,
In the training directory, there's some brief explanation about training.
the README file is pointing to a not existing ngram file (there's no ngram_frequencies file in the repo).
Is it possible to describe a quick procedure how to perform the training to get the final ngram pickle file, and what are the requirements about the corpora file format?
thanks
Fausto