def call(self, inputs):
if not isinstance(inputs, list):
raise ValueError('A merge layer should be called ' 'on a list of inputs.')
if self._reshape_required:
reshaped_inputs = []
input_ndims = list(map(K.ndim, inputs))
if None not in input_ndims:
# If ranks of all inputs are available,
# we simply expand each of them at axis=1
# until all of them have the same rank.
max_ndim = max(input_ndims)
for x in inputs:
x_ndim = K.ndim(x)
for _ in range(max_ndim - x_ndim):
x = K.expand_dims(x, 1)
reshaped_inputs.append(x)
return self._merge_function(reshaped_inputs)
else:
# Transpose all inputs so that batch size is the last dimension.
# (batch_size, dim1, dim2, ... ) -> (dim1, dim2, ... , batch_size)
transposed = False
for x in inputs:
x_ndim = K.ndim(x)
if x_ndim is None:
x_shape = K.shape(x)
batch_size = x_shape[0]
new_shape = K.concatenate([x_shape[1:], K.expand_dims(batch_size)])
x_transposed = K.reshape(x,
K.stack([batch_size,
K.prod(x_shape[1:])]))
x_transposed = K.permute_dimensions(x_transposed, (1, 0))
x_transposed = K.reshape(x_transposed, new_shape)
reshaped_inputs.append(x_transposed)
transposed = True
elif x_ndim > 1:
dims = list(range(1, x_ndim)) + [0]
reshaped_inputs.append(K.permute_dimensions(x, dims))
transposed = True
else:
# We don't transpose inputs if they are 1D vectors or scalars.
reshaped_inputs.append(x)
y = self._merge_function(reshaped_inputs)
y_ndim = K.ndim(y)
if transposed:
# If inputs have been transposed, we have to transpose the output too.
if y_ndim is None:
y_shape = K.shape(y)
y_ndim = K.shape(y_shape)[0]
batch_size = y_shape[y_ndim - 1]
new_shape = K.concatenate(
[K.expand_dims(batch_size), y_shape[:y_ndim - 1]])
y = K.reshape(y, (-1, batch_size))
y = K.permute_dimensions(y, (1, 0))
y = K.reshape(y, new_shape)
elif y_ndim > 1:
dims = [y_ndim - 1] + list(range(y_ndim - 1))
y = K.permute_dimensions(y, dims)
return y
else:
return self._merge_function(inputs)
def compute_mask(self, inputs, mask=None):
if mask is None:
return None
if not isinstance(mask, list):
raise ValueError('`mask` should be a list.')
if not isinstance(inputs, list):
raise ValueError('`inputs` should be a list.')
if len(mask) != len(inputs):
raise ValueError('The lists `inputs` and `mask` '
'should have the same length.')
if all([m is None for m in mask]):
return None
# Make a list of masks while making sure
# the dimensionality of each mask
# is the same as the corresponding input.
masks = []
for input_i, mask_i in zip(inputs, mask):
if mask_i is None:
# Input is unmasked. Append all 1s to masks,
masks.append(K.ones_like(input_i, dtype='bool'))
elif K.ndim(mask_i) < K.ndim(input_i):
# Mask is smaller than the input, expand it
masks.append(K.expand_dims(mask_i))
else:
masks.append(mask_i)
concatenated = K.concatenate(masks, axis=self.axis)
return K.all(concatenated, axis=-1, keepdims=False)
def compute_mask(self, inputs, mask=None):
if mask is None:
return None
if not isinstance(mask, list):
raise ValueError('`mask` should be a list.')
if not isinstance(inputs, list):
raise ValueError('`inputs` should be a list.')
if len(mask) != len(inputs):
raise ValueError('The lists `inputs` and `mask` '
'should have the same length.')
if all([m is None for m in mask]):
return None
# Make a list of masks while making sure
# the dimensionality of each mask
# is the same as the corresponding input.
masks = []
for input_i, mask_i in zip(inputs, mask):
if mask_i is None:
# Input is unmasked. Append all 1s to masks,
# but cast it to bool first
masks.append(K.cast(K.ones_like(input_i), 'bool'))
elif K.ndim(mask_i) < K.ndim(input_i):
# Mask is smaller than the input, expand it
masks.append(K.expand_dims(mask_i))
else:
masks.append(mask_i)
concatenated = K.concatenate(masks, axis=self.axis)
return K.all(concatenated, axis=-1, keepdims=False)
def _create_gates(self, inputs, memory):
"""Create input and forget gates for this step using `inputs` and `memory`.
Args:
inputs: Tensor input.
memory: The current state of memory.
Returns:
input_gate: A LSTM-like insert gate.
forget_gate: A LSTM-like forget gate.
"""
# We'll create the input and forget gates at once. Hence, calculate double
# the gate size.
memory = K.tanh(memory)
inputs = K.batch_flatten(inputs)
gate_inputs = self._linear(inputs, self.kernel_gi, self.bias_gi)
gate_inputs = K.expand_dims(gate_inputs, axis=1)
gate_memory = self._linear(memory, self.kernel_gm, self.bias_gm)
gates = tf.split(gate_memory + gate_inputs,
num_or_size_splits=2,
axis=2)
input_gate, forget_gate = gates
input_gate = tf.sigmoid(input_gate + self.input_bias)
forget_gate = tf.sigmoid(forget_gate + self.forget_bias)
return input_gate, forget_gate
def get_initial_state(self, inputs):
# build an all-zero tensor of shape (samples, output_dim)
initial_state = K.zeros_like(inputs) # (samples, timesteps, input_dim)
initial_state = K.sum(initial_state, axis=(1, 2)) # (samples,)
initial_state = K.expand_dims(initial_state) # (samples, 1)
initial_state = K.tile(initial_state, [1,
self.units]) # (samples, output_dim)
initial_state = [initial_state for _ in range(len(self.states))]
return initial_state
def compute_mask(self, inputs, mask=None):
if mask is None:
return None
if not isinstance(mask, list):
raise ValueError('`mask` should be a list.')
if not isinstance(inputs, list):
raise ValueError('`inputs` should be a list.')
if len(mask) != len(inputs):
raise ValueError('The lists `inputs` and `mask` '
'should have the same length.')
if all([m is None for m in mask]):
return None
masks = [K.expand_dims(m, 0) for m in mask if m is not None]
return K.all(K.concatenate(masks, axis=0), axis=0, keepdims=False)
def compute_mask(self, inputs, mask=None):
if mask is None:
return None
if not isinstance(mask, list):
raise ValueError('`mask` should be a list.')
if not isinstance(inputs, list):
raise ValueError('`inputs` should be a list.')
if len(mask) != len(inputs):
raise ValueError('The lists `inputs` and `mask` '
'should have the same length.')
if all([m is None for m in mask]):
return None
masks = [K.expand_dims(m, 0) for m in mask if m is not None]
return K.all(K.concatenate(masks, axis=0), axis=0, keepdims=False)
def call(self, inputs, memory, training=None):
batch_size = int(inputs.shape[0])
memory = K.reshape(memory, (batch_size, self.mem_slots, self.mem_size))
inputs = self._linear(inputs, self.kernel_in, self.bias_in)
inputs_reshape = K.expand_dims(inputs, axis=1)
memory_plus_input = K.concatenate([memory, inputs_reshape], axis=1)
next_memory = self._attend_over_memory(memory_plus_input)
n = inputs_reshape.get_shape().as_list()[1]
next_memory = next_memory[:, :-n, :]
input_gate, forget_gate = self._create_gates(inputs_reshape, memory)
next_memory = input_gate * K.tanh(next_memory)
next_memory += forget_gate * memory
next_memory = K.batch_flatten(next_memory)
return next_memory, (next_memory, )
def call(self, inputs):
if not isinstance(inputs, list):
raise ValueError('A merge layer should be called '
'on a list of inputs.')
if self._reshape_required:
reshaped_inputs = []
input_ndims = list(map(K.ndim, inputs))
if None not in input_ndims:
# If ranks of all inputs are available,
# we simply expand each of them at axis=1
# until all of them have the same rank.
max_ndim = max(input_ndims)
for x in inputs:
x_ndim = K.ndim(x)
for _ in range(max_ndim - x_ndim):
x = K.expand_dims(x, 1)
reshaped_inputs.append(x)
return self._merge_function(reshaped_inputs)
else:
# Transpose all inputs so that batch size is the last dimension.
# (batch_size, dim1, dim2, ... ) -> (dim1, dim2, ... , batch_size)
transposed = False
for x in inputs:
x_ndim = K.ndim(x)
if x_ndim is None:
x_shape = K.shape(x)
batch_size = x_shape[0]
new_shape = K.concatenate(
[x_shape[1:],
K.expand_dims(batch_size)])
x_transposed = K.reshape(
x, K.stack([batch_size,
K.prod(x_shape[1:])]))
x_transposed = K.permute_dimensions(
x_transposed, (1, 0))
x_transposed = K.reshape(x_transposed, new_shape)
reshaped_inputs.append(x_transposed)
transposed = True
elif x_ndim > 1:
dims = list(range(1, x_ndim)) + [0]
reshaped_inputs.append(K.permute_dimensions(x, dims))
transposed = True
else:
# We don't transpose inputs if they are 1D vectors or scalars.
reshaped_inputs.append(x)
y = self._merge_function(reshaped_inputs)
y_ndim = K.ndim(y)
if transposed:
# If inputs have been transposed, we have to transpose the output too.
if y_ndim is None:
y_shape = K.shape(y)
y_ndim = K.shape(y_shape)[0]
batch_size = y_shape[y_ndim - 1]
new_shape = K.concatenate(
[K.expand_dims(batch_size), y_shape[:y_ndim - 1]])
y = K.reshape(y, (-1, batch_size))
y = K.permute_dimensions(y, (1, 0))
y = K.reshape(y, new_shape)
elif y_ndim > 1:
dims = [y_ndim - 1] + list(range(y_ndim - 1))
y = K.permute_dimensions(y, dims)
return y
else:
return self._merge_function(inputs)
