def test_to_batch(self):
r = Resource(DATA_ROOT, ["sentiment", "text", "score"], "sentiment")
X, y = r.to_batch("train")
self.assertEqual(X.shape, (4, 2))
self.assertEqual(y.shape, (4, 1))
r.make_vocab()
r.column("text").as_word_seq(fixed_len=5)
X, y = r.to_batch("train", columns=("sentiment", "text"))
self.assertEqual(X.shape, (4, 5, len(r.vocab)))
def test_to_batch_iter(self):
r = Resource(DATA_ROOT, ["sentiment", "text", "score"], "sentiment")
r.make_vocab()
batch_size = 2
fixed_len = 5
r.column("text").as_word_seq(fixed_len=fixed_len)
iterator, count = r.to_batch_iter("train",
columns=("sentiment", "text"),
batch_size=batch_size)
self.assertEqual(count, batch_size)
for i in range(4):
X, y = next(iterator)
self.assertEqual(y.shape, (batch_size, 1))
self.assertEqual(X.shape, (batch_size, fixed_len, len(r.vocab)))
print(r.column("text").back(X))