def test_should_not_retrain_intent_classifier_when_no_force_retrain(self):
# Given
dataset_stream = io.StringIO("""
---
type: intent
name: MakeTea
utterances:
- make me a [beverage_temperature:Temperature](hot) cup of tea
- make me [number_of_cups:snips/number](five) tea cups
---
type: intent
name: MakeCoffee
utterances:
- make me [number_of_cups:snips/number](one) cup of coffee please
- brew [number_of_cups] cups of coffee""")
dataset = Dataset.from_yaml_files("en", [dataset_stream]).json
parser = ProbabilisticIntentParser()
intent_classifier = LogRegIntentClassifier()
intent_classifier.fit(dataset)
parser.intent_classifier = intent_classifier
# When / Then
with patch("snips_nlu.intent_classifier.log_reg_classifier"
".LogRegIntentClassifier.fit") as mock_fit:
parser.fit(dataset, force_retrain=False)
mock_fit.assert_not_called()
def test_log_best_features(self):
# Given
dataset_stream = io.StringIO("""
---
type: intent
name: intent1
utterances:
- foo bar
---
type: intent
name: intent2
utterances:
- lorem ipsum""")
dataset = Dataset.from_yaml_files("en", [dataset_stream]).json
shared = self.get_shared_data(dataset)
intent_classifier = LogRegIntentClassifier(**shared)
# When
self.assertIsNone(intent_classifier.log_best_features(20))
intent_classifier.fit(dataset)
log = intent_classifier.log_best_features(20)
# Then
self.assertIsInstance(log, str)
self.assertIn("Top 20", log)
def test_training_should_be_reproducible(self):
# Given
random_state = 40
dataset_stream = io.StringIO("""
---
type: intent
name: MakeTea
utterances:
- make me a [beverage_temperature:Temperature](hot) cup of tea
- make me [number_of_cups:snips/number](five) tea cups
---
type: intent
name: MakeCoffee
utterances:
- make me [number_of_cups:snips/number](one) cup of coffee please
- brew [number_of_cups] cups of coffee""")
dataset = Dataset.from_yaml_files("en", [dataset_stream]).json
# When
classifier1 = LogRegIntentClassifier(random_state=random_state)
classifier1.fit(dataset)
classifier2 = LogRegIntentClassifier(random_state=random_state)
classifier2.fit(dataset)
# Then
with temp_dir() as tmp_dir:
dir_classifier1 = tmp_dir / "classifier1"
dir_classifier2 = tmp_dir / "classifier2"
classifier1.persist(dir_classifier1)
classifier2.persist(dir_classifier2)
hash1 = dirhash(str(dir_classifier1), 'sha256')
hash2 = dirhash(str(dir_classifier2), 'sha256')
self.assertEqual(hash1, hash2)
def test_log_activation_weights(self):
# Given
dataset_stream = io.StringIO("""
---
type: intent
name: intent1
utterances:
- foo bar
---
type: intent
name: intent2
utterances:
- lorem ipsum""")
dataset = Dataset.from_yaml_files("en", [dataset_stream]).json
shared = self.get_shared_data(dataset)
intent_classifier = LogRegIntentClassifier(**shared)
text = "yo"
utterances = [text_to_utterance(text)]
self.assertIsNone(intent_classifier.log_activation_weights(text, None))
# When
intent_classifier.fit(dataset)
x = intent_classifier.featurizer.transform(utterances)[0]
log = intent_classifier.log_activation_weights(text, x, top_n=42)
# Then
self.assertIsInstance(log, str)
self.assertIn("Top 42", log)
def test_should_not_retrain_intent_classifier_when_no_force_retrain(self):
# Given
parser = ProbabilisticIntentParser()
intent_classifier = LogRegIntentClassifier()
intent_classifier.fit(BEVERAGE_DATASET)
parser.intent_classifier = intent_classifier
# When / Then
with patch("snips_nlu.intent_classifier.log_reg_classifier"
".LogRegIntentClassifier.fit") as mock_fit:
parser.fit(BEVERAGE_DATASET, force_retrain=False)
mock_fit.assert_not_called()
