del model
# Dataset Partition
# Make all(100%, 400000 rows) data from amazon_test.csv as train_test data to test
train_size = int(1 * (len(review_test)))
max_no_tokens = 15
indexes = set(
np.random.choice(len(review2), train_size + test_size, replace=False))
x_train2 = np.zeros((train_size, max_no_tokens, vector_size), dtype=K.floatx())
y_train2 = np.zeros((train_size, 2), dtype=np.int32)
for i, index in enumerate(indexes):
for t, token in enumerate(review2[index]):
if t >= max_no_tokens:
break
if token not in x_vectors:
continue
if i < train_size:
x_train2[i, t, :] = x_vectors[token]
if i < train_size:
y_train2[i, :] = [1.0, 0.0] if label2[index] == 1 else [0.0, 1.0]
print(x_train2.shape, y_train2.shape)
# Evaluate by 100% test data
model = model_final
model.evaluate(x=x_train2, y=y_train2, batch_size=32, verbose=1)
