from __future__ import unicode_literals from ...language import Language def test_simple_train(): nlp = Language() nlp.add_pipe(nlp.create_pipe('textcat')) nlp.get_pipe('textcat').add_label('is_good') nlp.begin_training() for i in range(5): for text, answer in [('aaaa', 1.), ('bbbb', 0), ('aa', 1.), ('bbbbbbbbb', 0.), ('aaaaaa', 1)]: nlp.update([text], [{'cats': {'answer': answer}}]) doc = nlp(u'aaa') assert 'is_good' in doc.cats assert doc.cats['is_good'] >= 0.5