def test_Identity(self):
with self.cached_session():
tensor_shape = (3, 4, 5)
with self.assertRaises(ValueError):
self._runner(init_ops.Identity(),
tensor_shape,
target_mean=1. / tensor_shape[0],
target_max=1.)
tensor_shape = (3, 3)
self._runner(init_ops.Identity(),
tensor_shape,
target_mean=1. / tensor_shape[0],
target_max=1.)
def test_Identity(self):
with self.cached_session():
shape = (3, 4, 5)
for tensor_shape in [shape, tensor_shape_lib.TensorShape(shape)]:
with self.assertRaises(ValueError):
self._runner(init_ops.Identity(),
tensor_shape,
target_mean=1. / int(tensor_shape[0]),
target_max=1.)
shape = (3, 3)
for tensor_shape in [shape, tensor_shape_lib.TensorShape(shape)]:
self._runner(init_ops.Identity(),
tensor_shape,
target_mean=1. / int(tensor_shape[0]),
target_max=1.)
def build(self, input_shape):
channel_axis = self._channel_axis()
input_shape = tensor_shape.TensorShape(input_shape)
num_channels = input_shape[channel_axis].value
if num_channels is None:
raise ValueError("The channel dimension of the inputs to `GDN` "
"must be defined.")
self._input_rank = input_shape.ndims
self.input_spec = base.InputSpec(ndim=input_shape.ndims,
axes={channel_axis: num_channels})
self.beta = self._beta_parameterization(name="beta",
shape=[num_channels],
dtype=self.dtype,
getter=self.add_variable,
initializer=init_ops.Ones())
self.gamma = self._gamma_parameterization(
name="gamma",
shape=[num_channels, num_channels],
dtype=self.dtype,
getter=self.add_variable,
initializer=init_ops.Identity(gain=self._gamma_init))
self.built = True