def test_make_dummy_element_NamedTupleType(self):
tensor1 = computation_types.TensorType(tf.float32,
[None, 10, None, 10, 10])
tensor2 = computation_types.TensorType(tf.int32, [10, None, 10])
namedtuple = computation_types.NamedTupleType([('x', tensor1),
('y', tensor2)])
unnamedtuple = computation_types.NamedTupleType([('x', tensor1),
('y', tensor2)])
elem = tensorflow_utils.make_dummy_element_for_type_spec(namedtuple)
correct_list = [
np.zeros([0, 10, 0, 10, 10], np.float32),
np.zeros([10, 0, 10], np.int32)
]
self.assertEqual(len(elem), len(correct_list))
for k in range(len(elem)):
self.assertAllClose(elem[k], correct_list[k])
unnamed_elem = tensorflow_utils.make_dummy_element_for_type_spec(
unnamedtuple)
self.assertEqual(len(unnamed_elem), len(correct_list))
for k in range(len(unnamed_elem)):
self.assertAllClose(unnamed_elem[k], correct_list[k])
def test_make_dummy_element_TensorType(self):
type_spec = computation_types.TensorType(tf.float32,
[None, 10, None, 10, 10])
elem = tensorflow_utils.make_dummy_element_for_type_spec(type_spec)
correct_elem = np.zeros([0, 10, 0, 10, 10], np.float32)
self.assertAllClose(elem, correct_elem)
def test_make_dummy_element_for_type_spec_raises_negative_none_dim_replacement(
self):
with self.assertRaisesRegex(ValueError, 'nonnegative'):
tensorflow_utils.make_dummy_element_for_type_spec(tf.float32, -1)
def test_make_dummy_element_for_type_spec_raises_SequenceType(self):
type_spec = computation_types.SequenceType(tf.float32)
with self.assertRaisesRegex(ValueError,
'Cannot construct array for TFF type'):
tensorflow_utils.make_dummy_element_for_type_spec(type_spec)
