def setup_mock_rnn_introspective_rationale_explainer(model_config, data):
explainer = IntrospectiveRationaleExplainer(classifier_type=CLASSIFIER_TYPE_RNN, cuda=CUDA)
preprocessor = GlovePreprocessor(TOKEN_COUNT_THRESHOLD, MAX_SENT_COUNT)
preprocessor.build_vocab(data)
explainer.set_preprocessor(preprocessor)
explainer.build_model_config(model_config)
return explainer
def test_set_custom_classifier(self):
all_data = get_ssts_dataset('train')
model_config = RNN_MODEL_CONFIG
preprocessor = GlovePreprocessor(TOKEN_COUNT_THRESHOLD, MAX_SENT_COUNT)
preprocessor.build_vocab(all_data[TEXT_COL])
explainer = IntrospectiveRationaleExplainer(
classifier_type=CLASSIFIER_TYPE_CUSTOM, cuda=CUDA)
explainer.build_model_config(model_config)
# set custom classifier
classifier = ClassifierModule(explainer.get_model_config(),
preprocessor.word_vocab)
explainer.set_preprocessor(preprocessor)
explainer.set_classifier(classifier)
assert explainer.classifier_type is CLASSIFIER_TYPE_CUSTOM
def test_rnn_explain_local(self):
train_data = get_ssts_dataset('train')
test_data = get_ssts_dataset('test')
all_data = pd.concat([train_data, test_data])
X_train = train_data[TEXT_COL]
X_test = test_data[TEXT_COL]
preprocessor = GlovePreprocessor(count_threshold=TOKEN_COUNT_THRESHOLD,
token_cutoff=MAX_SENT_COUNT)
preprocessor.build_vocab(all_data[TEXT_COL])
df_train = pd.concat(
[train_data[LABEL_COL],
preprocessor.preprocess(X_train)], axis=1)
df_test = pd.concat(
[test_data[LABEL_COL],
preprocessor.preprocess(X_test)], axis=1)
model_config = RNN_MODEL_CONFIG
explainer = IntrospectiveRationaleExplainer(
classifier_type=CLASSIFIER_TYPE_RNN, cuda=CUDA)
explainer.build_model_config(model_config)
explainer.set_preprocessor(preprocessor)
explainer.load()
explainer.fit(df_train, df_test)
local_explanation = explainer.explain_local(SENTENCE)
assert len(local_explanation.local_importance_values) == len(
SENTENCE.split())