indo Use v2_lang.py In split=['train[:80%]', 'train[80%:]']), change :80% to :100%, to use all data as training data.