test_data = test_data.sample(frac=1.0).reset_index(drop=True)
test_data = test_data.head(int(len(test_data) * args.size))
print(test_data.columns)
assert len(test_data.columns) > 1, "Labels of Test set not available."
test_label = list(test_data.SECTION)
test_data = list(test_data.STORY)
predictions = []
if args.model != 'sentence_pair':
for story in tqdm(test_data):
words = nltk.word_tokenize(story.lower())
embedding = np.array([reader.sent2vec(words)])
prediction = model.predict(embedding)[0]
predictions.append(np.argmax(prediction))
else:
for story in tqdm(test_data):
words = nltk.word_tokenize(story.lower())
embedding = np.array([reader.sent2vec(words)])
prediction = []
for i in range(4):
x2 = np.array([reader.embedding[reader.label_map[i]]])
prediction.append(model.predict([embedding, x2])[0])
predictions.append(np.argmax(prediction))
print(classification_report(test_label, predictions, labels=[0, 1, 2, 3]))
