def call(self, inputs, training=None, mask=None): kwargs = {} if generic_utils.has_arg(self.layer.call, 'training'): kwargs['training'] = training uses_learning_phase = False # pylint: disable=redefined-outer-name input_shape = K.int_shape(inputs) if input_shape[0]: # batch size matters, use rnn-based implementation def step(x, _): global uses_learning_phase # pylint: disable=global-variable-undefined output = self.layer.call(x, **kwargs) if hasattr(output, '_uses_learning_phase'): uses_learning_phase = (output._uses_learning_phase or uses_learning_phase) return output, [] _, outputs, _ = K.rnn( step, inputs, initial_states=[], input_length=input_shape[1], unroll=False) y = outputs else: # No batch size specified, therefore the layer will be able # to process batches of any size. # We can go with reshape-based implementation for performance. input_length = input_shape[1] if not input_length: input_length = array_ops.shape(inputs)[1] inner_input_shape = self._get_shape_tuple((-1,), inputs, 2) # Shape: (num_samples * timesteps, ...). And track the # transformation in self._input_map. input_uid = generic_utils.object_list_uid(inputs) inputs = array_ops.reshape(inputs, inner_input_shape) self._input_map[input_uid] = inputs # (num_samples * timesteps, ...) if generic_utils.has_arg(self.layer.call, 'mask') and mask is not None: inner_mask_shape = self._get_shape_tuple((-1,), mask, 2) kwargs['mask'] = K.reshape(mask, inner_mask_shape) y = self.layer.call(inputs, **kwargs) if hasattr(y, '_uses_learning_phase'): uses_learning_phase = y._uses_learning_phase # Shape: (num_samples, timesteps, ...) output_shape = self.compute_output_shape(input_shape).as_list() output_shape = self._get_shape_tuple( (-1, input_length), y, 1, output_shape[2:]) y = array_ops.reshape(y, output_shape) # Apply activity regularizer if any: if (hasattr(self.layer, 'activity_regularizer') and self.layer.activity_regularizer is not None): regularization_loss = self.layer.activity_regularizer(y) self.add_loss(regularization_loss, inputs) if uses_learning_phase: y._uses_learning_phase = True return y
def call(self, inputs, training=None, mask=None): kwargs = {} if generic_utils.has_arg(self.layer.call, 'training'): kwargs['training'] = training input_shape = K.int_shape(inputs) if input_shape[0] and not self._always_use_reshape: inputs, row_lengths = K.convert_inputs_if_ragged(inputs) is_ragged_input = row_lengths is not None # batch size matters, use rnn-based implementation def step(x, _): output = self.layer(x, **kwargs) return output, [] _, outputs, _ = K.rnn(step, inputs, initial_states=[], input_length=row_lengths[0] if is_ragged_input else input_shape[1], mask=mask, unroll=False) y = K.maybe_convert_to_ragged(is_ragged_input, outputs, row_lengths) else: # No batch size specified, therefore the layer will be able # to process batches of any size. # We can go with reshape-based implementation for performance. if isinstance(inputs, ragged_tensor.RaggedTensor): y = self.layer(inputs.values, **kwargs) y = ragged_tensor.RaggedTensor.from_row_lengths( y, inputs.nested_row_lengths()[0]) else: input_length = input_shape[1] if not input_length: input_length = array_ops.shape(inputs)[1] inner_input_shape = self._get_shape_tuple((-1, ), inputs, 2) # Shape: (num_samples * timesteps, ...). And track the # transformation in self._input_map. input_uid = generic_utils.object_list_uid(inputs) inputs = array_ops.reshape(inputs, inner_input_shape) self._input_map[input_uid] = inputs # (num_samples * timesteps, ...) if generic_utils.has_arg(self.layer.call, 'mask') and mask is not None: inner_mask_shape = self._get_shape_tuple((-1, ), mask, 2) kwargs['mask'] = K.reshape(mask, inner_mask_shape) y = self.layer(inputs, **kwargs) # Shape: (num_samples, timesteps, ...) output_shape = self.compute_output_shape(input_shape).as_list() output_shape = self._get_shape_tuple((-1, input_length), y, 1, output_shape[2:]) y = array_ops.reshape(y, output_shape) return y
def call(self, inputs, training=None, mask=None): kwargs = {} if generic_utils.has_arg(self.layer.call, 'training'): kwargs['training'] = training input_shape = K.int_shape(inputs) if input_shape[0] and not self._always_use_reshape: # batch size matters, use rnn-based implementation def step(x, _): output = self.layer.call(x, **kwargs) return output, [] _, outputs, _ = K.rnn(step, inputs, initial_states=[], input_length=input_shape[1], unroll=False) y = outputs else: # No batch size specified, therefore the layer will be able # to process batches of any size. # We can go with reshape-based implementation for performance. input_length = input_shape[1] if not input_length: input_length = array_ops.shape(inputs)[1] inner_input_shape = self._get_shape_tuple((-1, ), inputs, 2) # Shape: (num_samples * timesteps, ...). And track the # transformation in self._input_map. input_uid = generic_utils.object_list_uid(inputs) inputs = array_ops.reshape(inputs, inner_input_shape) self._input_map[input_uid] = inputs # (num_samples * timesteps, ...) if generic_utils.has_arg(self.layer.call, 'mask') and mask is not None: inner_mask_shape = self._get_shape_tuple((-1, ), mask, 2) kwargs['mask'] = K.reshape(mask, inner_mask_shape) y = self.layer.call(inputs, **kwargs) # Shape: (num_samples, timesteps, ...) output_shape = self.compute_output_shape(input_shape).as_list() output_shape = self._get_shape_tuple((-1, input_length), y, 1, output_shape[2:]) y = array_ops.reshape(y, output_shape) # Apply activity regularizer if any: if (hasattr(self.layer, 'activity_regularizer') and self.layer.activity_regularizer is not None): regularization_loss = self.layer.activity_regularizer(y) base_layer_utils.check_graph_consistency( regularization_loss, method='activity_regularizer') self.add_loss(regularization_loss, inputs) return y
def compute_mask(self, inputs, mask=None): """Computes an output mask tensor for Embedding layer. This is based on the inputs, mask, and the inner layer. If batch size is specified: Simply return the input `mask`. (An rnn-based implementation with more than one rnn inputs is required but not supported in tf.keras yet.) Otherwise we call `compute_mask` of the inner layer at each time step. If the output mask at each time step is not `None`: (E.g., inner layer is Masking or RNN) Concatenate all of them and return the concatenation. If the output mask at each time step is `None` and the input mask is not `None`:(E.g., inner layer is Dense) Reduce the input_mask to 2 dimensions and return it. Otherwise (both the output mask and the input mask are `None`): (E.g., `mask` is not used at all) Return `None`. Arguments: inputs: Tensor with shape [batch size, timesteps, ...] indicating the input to TimeDistributed. If static shape information is available for "batch size", `mask` is returned unmodified. mask: Either None (indicating no masking) or a Tensor indicating the input mask for TimeDistributed. The shape can be static or dynamic. Returns: Either None (no masking), or a [batch size, timesteps, ...] Tensor with an output mask for the TimeDistributed layer with the shape beyond the second dimension being the value of the input mask shape(if the computed output mask is none), an output mask with the shape beyond the first dimension being the value of the mask shape(if mask is not None) or output mask with the shape beyond the first dimension being the value of the computed output shape. """ # cases need to call the layer.compute_mask when input_mask is None: # Masking layer and Embedding layer with mask_zero input_shape = K.int_shape(inputs) if input_shape[0]: # batch size matters, we currently do not handle mask explicitly return mask inner_mask = mask if inner_mask is not None: inner_mask_shape = self._get_shape_tuple((-1, ), mask, 2) inner_mask = K.reshape(inner_mask, inner_mask_shape) input_uid = generic_utils.object_list_uid(inputs) inner_inputs = self._input_map.get(input_uid, inputs) output_mask = self.layer.compute_mask(inner_inputs, inner_mask) if output_mask is None: if mask is None: return None # input_mask is not None, and output_mask is None: # we should return a not-None mask output_mask = mask for _ in range(2, len(K.int_shape(mask))): output_mask = K.any(output_mask, axis=-1) else: # output_mask is not None. We need to reshape it input_length = input_shape[1] if not input_length: input_length = K.shape(inputs)[1] output_mask_int_shape = K.int_shape(output_mask) if output_mask_int_shape is None: # if the output_mask does not have a static shape, # its shape must be the same as mask's if mask is not None: output_mask_int_shape = K.int_shape(mask) else: output_mask_int_shape = K.compute_output_shape( input_shape)[:-1] output_mask_shape = self._get_shape_tuple( (-1, input_length), output_mask, 1, output_mask_int_shape[1:]) output_mask = K.reshape(output_mask, output_mask_shape) return output_mask
def compute_mask(self, inputs, mask=None): """Computes an output mask tensor for Embedding layer. This is based on the inputs, mask, and the inner layer. If batch size is specified: Simply return the input `mask`. (An rnn-based implementation with more than one rnn inputs is required but not supported in tf.keras yet.) Otherwise we call `compute_mask` of the inner layer at each time step. If the output mask at each time step is not `None`: (E.g., inner layer is Masking or RNN) Concatenate all of them and return the concatenation. If the output mask at each time step is `None` and the input mask is not `None`:(E.g., inner layer is Dense) Reduce the input_mask to 2 dimensions and return it. Otherwise (both the output mask and the input mask are `None`): (E.g., `mask` is not used at all) Return `None`. Arguments: inputs: Tensor with shape [batch size, timesteps, ...] indicating the input to TimeDistributed. If static shape information is available for "batch size", `mask` is returned unmodified. mask: Either None (indicating no masking) or a Tensor indicating the input mask for TimeDistributed. The shape can be static or dynamic. Returns: Either None (no masking), or a [batch size, timesteps, ...] Tensor with an output mask for the TimeDistributed layer with the shape beyond the second dimension being the value of the input mask shape(if the computed output mask is none), an output mask with the shape beyond the first dimension being the value of the mask shape(if mask is not None) or output mask with the shape beyond the first dimension being the value of the computed output shape. """ # cases need to call the layer.compute_mask when input_mask is None: # Masking layer and Embedding layer with mask_zero input_shape = K.int_shape(inputs) if input_shape[0]: # batch size matters, we currently do not handle mask explicitly return mask inner_mask = mask if inner_mask is not None: inner_mask_shape = self._get_shape_tuple((-1,), mask, 2) inner_mask = K.reshape(inner_mask, inner_mask_shape) input_uid = generic_utils.object_list_uid(inputs) inner_inputs = self._input_map.get(input_uid, inputs) output_mask = self.layer.compute_mask(inner_inputs, inner_mask) if output_mask is None: if mask is None: return None # input_mask is not None, and output_mask is None: # we should return a not-None mask output_mask = mask for _ in range(2, len(K.int_shape(mask))): output_mask = K.any(output_mask, axis=-1) else: # output_mask is not None. We need to reshape it input_length = input_shape[1] if not input_length: input_length = K.shape(inputs)[1] output_mask_int_shape = K.int_shape(output_mask) if output_mask_int_shape is None: # if the output_mask does not have a static shape, # its shape must be the same as mask's if mask is not None: output_mask_int_shape = K.int_shape(mask) else: output_mask_int_shape = K.compute_output_shape(input_shape)[:-1] output_mask_shape = self._get_shape_tuple( (-1, input_length), output_mask, 1, output_mask_int_shape[1:]) output_mask = K.reshape(output_mask, output_mask_shape) return output_mask