def test_tf_tensor_list_new_empty(self):
l = data_structures.tf_tensor_list_new([],
element_dtype=dtypes.int32,
element_shape=())
t = list_ops.tensor_list_stack(l, element_dtype=dtypes.int32)
with self.cached_session() as sess:
self.assertAllEqual(self.evaluate(t), [])
def tensor_list(elements,
element_dtype=None,
element_shape=None,
use_tensor_array=False):
"""Creates an tensor list and populates it with the given elements.
This function provides a more uniform access to tensor lists and tensor
arrays, and allows optional initialization.
Note: this function is a simplified wrapper. If you need greater control,
it is recommended to use the underlying implementation directly.
Args:
elements: Iterable[tf.Tensor, ...], the elements to initially fill the list
with
element_dtype: Optional[tf.DType], data type for the elements in the list;
required if the list is empty
element_shape: Optional[tf.TensorShape], shape for the elements in the list;
required if the list is empty
use_tensor_array: bool, whether to use the more compatible but restrictive
tf.TensorArray implementation
Returns:
Union[tf.Tensor, tf.TensorArray], the new list.
Raises:
ValueError: for invalid arguments
"""
_validate_list_constructor(elements, element_dtype, element_shape)
if use_tensor_array:
return data_structures.tf_tensor_array_new(elements, element_dtype,
element_shape)
else:
return data_structures.tf_tensor_list_new(elements, element_dtype,
element_shape)
def test_tf_tensor_list_new_empty(self):
l = data_structures.tf_tensor_list_new([],
element_dtype=dtypes.int32,
element_shape=())
t = list_ops.tensor_list_stack(l, element_dtype=dtypes.int32)
with self.cached_session() as sess:
self.assertAllEqual(sess.run(t), [])
def tensor_list(elements,
element_dtype=None,
element_shape=None,
use_tensor_array=False):
"""Creates an tensor list and populates it with the given elements.
This function provides a more uniform access to tensor lists and tensor
arrays, and allows optional initialization.
Note: this function is a simplified wrapper. If you need greater control,
it is recommended to use the underlying implementation directly.
Args:
elements: Iterable[tf.Tensor, ...], the elements to initially fill the list
with
element_dtype: Optional[tf.DType], data type for the elements in the list;
required if the list is empty
element_shape: Optional[tf.TensorShape], shape for the elements in the list;
required if the list is empty
use_tensor_array: bool, whether to use the more compatible but restrictive
tf.TensorArray implementation
Returns:
Union[tf.Tensor, tf.TensorArray], the new list.
Raises:
ValueError: for invalid arguments
"""
_validate_list_constructor(elements, element_dtype, element_shape)
if use_tensor_array:
return data_structures.tf_tensor_array_new(elements, element_dtype,
element_shape)
else:
return data_structures.tf_tensor_list_new(elements, element_dtype,
element_shape)
def test_tf_tensor_list_new_illegal_input(self):
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new([3, 4.0])
# TODO(mdan): It might make more sense to type cast in this case.
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new([3, 4], element_dtype=dtypes.float32)
# Tensor lists do support heterogeneous lists.
self.assertIsNot(data_structures.tf_tensor_list_new([3, [4, 5]]), None)
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new([3, 4], element_shape=(2,))
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new(
constant_op.constant([1, 2, 3]), element_shape=[1])
def test_tf_tensor_list_new_illegal_input(self):
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new([3, 4.0])
# TODO(mdan): It might make more sense to type cast in this case.
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new([3, 4], element_dtype=dtypes.float32)
# Tensor lists do support heterogeneous lists.
self.assertIsNot(data_structures.tf_tensor_list_new([3, [4, 5]]), None)
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new([3, 4], element_shape=(2,))
with self.assertRaises(ValueError):
data_structures.tf_tensor_list_new(
constant_op.constant([1, 2, 3]), element_shape=[1])
def test_len(self):
self.assertEqual(py_builtins.len_([1, 2, 3]), 3)
with self.test_session() as sess:
t = py_builtins.len_(constant_op.constant([[1], [2], [3]]))
self.assertEqual(t, 3)
ta = py_builtins.len_(tensor_array_ops.TensorArray(dtypes.int32, size=5))
self.assertEqual(sess.run(ta), 5)
tl = py_builtins.len_(data_structures.tf_tensor_list_new([3, 4, 5]))
self.assertEqual(sess.run(tl), 3)
def test_len(self):
self.assertEqual(py_builtins.len_([1, 2, 3]), 3)
with self.cached_session() as sess:
t = py_builtins.len_(constant_op.constant([[1], [2], [3]]))
self.assertEqual(t, 3)
ta = py_builtins.len_(tensor_array_ops.TensorArray(dtypes.int32, size=5))
self.assertEqual(self.evaluate(ta), 5)
tl = py_builtins.len_(data_structures.tf_tensor_list_new([3, 4, 5]))
self.assertEqual(self.evaluate(tl), 3)
def test_tf_tensor_list_new_from_tensor(self):
l = data_structures.tf_tensor_list_new(constant_op.constant([3, 4, 5]))
t = list_ops.tensor_list_stack(l, element_dtype=dtypes.int32)
with self.cached_session() as sess:
self.assertAllEqual(sess.run(t), [3, 4, 5])
def test_tf_tensor_list_new_from_tensor(self):
l = data_structures.tf_tensor_list_new(constant_op.constant([3, 4, 5]))
t = list_ops.tensor_list_stack(l, element_dtype=dtypes.int32)
with self.cached_session() as sess:
self.assertAllEqual(sess.run(t), [3, 4, 5])
def test_tf_tensor_list_new(self):
l = data_structures.tf_tensor_list_new([3, 4, 5])
t = list_ops.tensor_list_stack(l, element_dtype=dtypes.int32)
with self.cached_session() as sess:
self.assertAllEqual(self.evaluate(t), [3, 4, 5])
