250k lines from MaCoCu-dataset, where we translated the original corpus from english to slovene and preprocessed the data to have a form:
id original_slovene_sentence \ttranslated_slovene_sentence