tfr.keras.layers.create_tower

Creates a feed-forward network as tf.keras.Sequential.

tfr.keras.layers.create_tower(
    hidden_layer_dims: List[int],
    output_units: int,
    activation: Optional[Callable[..., tf.Tensor]] = None,
    input_batch_norm: bool = False,
    use_batch_norm: bool = True,
    batch_norm_moment: float = 0.999,
    dropout: float = 0.5,
    name: Optional[str] = None,
    **kwargs
)

It creates a feed-forward network with batch normalization and dropout, and optionally applies batch normalization on inputs.

Example usage:

tower = create_tower(hidden_layer_dims=[64, 32, 16], output_units=1)
inputs = tf.ones([2, 3, 1])
tower_logits = tower(inputs)

Args
hidden_layer_dims	Iterable of number hidden units per layer. All layers are fully connected. Ex. [64, 32] means first layer has 64 nodes and second one has 32.
output_units	Size of output logits from this tower.
activation	Activation function applied to each layer. If None, will use an identity activation.
input_batch_norm	Whether to use batch normalization for input layer.
use_batch_norm	Whether to use batch normalization after each hidden layer.
batch_norm_moment	Momentum for the moving average in batch normalization.
dropout	When not None, the probability we will drop out a given coordinate.
name	Name of the Keras layer.
**kwargs	Keyword arguments for every tf.keras.Dense layers.

Returns
A tf.keras.Sequential object.