def model_fn(input_features, output_sizes):
"""Applies model to input features and produces output of given sizes."""
if is_training:
# Flatten all the model-irrelevant dimensions, i.e., dimensions that
# precede the sequence / feature channel dimensions). Note that we only do
# this for training, for which the batch size is known.
num_last_dims_to_keep = 2 if sequential_inputs else 1
flattened_input_features = data_utils.flatten_first_dims(
input_features, num_last_dims_to_keep=num_last_dims_to_keep)
flattened_shape = data_utils.get_shape_by_first_dims(
input_features, num_last_dims=num_last_dims_to_keep)
outputs, activations = base_model_fn(flattened_input_features,
output_sizes)
# Unflatten back all the model-irrelevant dimensions.
for key, output in outputs.items():
outputs[key] = data_utils.unflatten_first_dim(
output, shape_to_unflatten=flattened_shape)
for key, activation in activations.items():
activations[key] = data_utils.unflatten_first_dim(
activation, shape_to_unflatten=flattened_shape)
else:
outputs, activations = base_model_fn(input_features, output_sizes)
return outputs, activations
def test_unflatten_first_dim(self):
# Shape = [6, 2].
x = tf.constant([[1, 2], [3, 4], [5, 6], [7, 8], [9, 10], [11, 12]])
unflattened_x = data_utils.unflatten_first_dim(
x, shape_to_unflatten=tf.constant([2, 3]))
self.assertAllEqual(unflattened_x,
[[[1, 2], [3, 4], [5, 6]], [[7, 8], [9, 10], [11, 12]]])