def test5():
'''
===========================================================
Configuration:
1. Return only min pooling
===========================================================
'''
print('running vectorizer test 5 ...')
model_file = 'model.bin'
model = KeyedVectors.load_word2vec_format(model_file, binary=True)
config = vectorizer.Config('gensim',
model=model,
average_type=None,
min_pooling=True,
max_pooling=False,
concatenation=None,
return_type='only_min_pooling')
pipeline = vectorizer.Vectorizer(config)
case = [
['отвергнуть_VERB', 'щедрый_ADJ', 'дар_NOUN'],
['покупать_VERB', 'преданность_NOUN', 'дар_NOUN', 'награда_NOUN'],
['яд_NOUN', 'последний_ADJ', 'дар_NOUN', 'мой_DET', 'изора_NOUN']
]
expected = pickle.load(open('./sanity_checks_data/vectorizer5.p',
'rb'))
res = pipeline.fit(case).transform(case)
np.testing.assert_allclose(expected, res)
print('vectorizer test 5 is passed')
def test6():
'''
===========================================================
Configuration:
1. Full
2. One word in line
===========================================================
'''
print('running vectorizer test 6 ...')
model_file = 'model.bin'
model = KeyedVectors.load_word2vec_format(model_file, binary=True)
config = vectorizer.Config('gensim',
model=model,
average_type='simple_average',
min_pooling=True,
max_pooling=True,
concatenation='full',
return_type=None)
pipeline = vectorizer.Vectorizer(config)
case = [['отвергнуть_VERB', 'щедрый_ADJ'], ['покупать_VERB']]
expected = pickle.load(open('./sanity_checks_data/vectorizer6.p',
'rb'))
res = pipeline.fit(case).transform(case)
np.testing.assert_allclose(expected, res)
print('vectorizer test 6 is passed')
def test1():
'''
===========================================================
Full house:
1. Simple average
2. Min pooling
3. Max pooling
4. Full concatenation
5. Full return
===========================================================
'''
print('running vectorizer test 1 ...')
model_file = 'model.bin'
model = KeyedVectors.load_word2vec_format(model_file, binary=True)
config = vectorizer.Config('gensim',
model=model,
average_type='simple_average',
min_pooling=True,
max_pooling=True,
concatenation='full',
return_type='full')
pipeline = vectorizer.Vectorizer(config)
case = [
['отвергнуть_VERB', 'щедрый_ADJ', 'дар_NOUN'],
['покупать_VERB', 'преданность_NOUN', 'дар_NOUN', 'награда_NOUN'],
['яд_NOUN', 'последний_ADJ', 'дар_NOUN', 'мой_DET', 'изора_NOUN']
]
expected = pickle.load(open('./sanity_checks_data/vectorizer1.p',
'rb'))
res = pipeline.fit(case).transform(case)
np.testing.assert_allclose(expected, res)
print('vectorizer test 1 is passed')