Very good work. Now I want to reproduce its data cleaning process. Will the fasttext model and its training data be opened in the future?