tfl.layers.RTL

Layer which includes a random ensemble of lattices.

RTL (Random Tiny Lattices) is an ensemble of tfl.layers.Lattice layers that takes in a collection of monotonic and unconstrained features and randomly arranges them into lattices of a given rank. The input is taken as "groups", and inputs from the same group will not be used in the same lattice. E.g. the input can be the output of a calibration layer with multiple units applied to the same input feature. If there are more slots in the RTL than the number of inputs, inputs will be repeatedly used. Repeats will be approximately uniform across all inputs.

Input shape:

  • A dict with keys in ['unconstrained', 'increasing'], and the values either a list of tensors of shape (batch_size, D_i), or a single tensor of shape (batch_size, D) that will be conceptually split into a list of D tensors of size (batch_size, 1). Each tensor in the list is considered a "group" of features that the RTL layer should try not to use in the same lattice.
  • A single tensor of shape (batch_size, D), which is considered to be unconstrained and will be conceptually split into a list of D tensors of size (batch_size, 1).

Output shape:

If separate_outputs == True, the output will be in the same format as the input and can be passed to follow on RTL layers: {'unconstrained': unconstrained_out, 'increasing': mon_out} where unconstrained_out and mon_out are of (batch_size, num_unconstrained_out) and (batch_size, num_mon_out) respectively, and num_unconstrained_out + num_mon_out == num_lattices. If separate_outputs == False the output will be a rank-2 tensor with shape: (batch_size, num_lattices) if average_outputs is False, or (batch_size, 1) if True.

Example:

a = tf.keras.Input(shape=(1,))
b = tf.keras.Input(shape=(1,))
c = tf.keras.Input(shape=(1,))
d = tf.keras.Input(shape=(1,))
cal_a = tfl.layers.CategoricalCalibration(
    units=10, output_min=0, output_max=1, ...)(a)
cal_b = tfl.layers.PWLCalibration(
    units=20, output_min=0, output_max=1, ...)(b)
cal_c = tfl.layers.PWLCalibration(
    units=10, output_min=0, output_max=1, monotonicity='increasing', ...)(c)
cal_d = tfl.layers.PWLCalibration(
    units=20, output_min=0, output_max=1, monotonicity='decreasing', ...)(d)
rtl_0 = RTL(
    num_lattices=20,
    lattice_rank=3,
    output_min=0,
    output_max=1,
    separate_outputs=True,
)({
    'unconstrained': [cal_a, cal_b],
    'increasing': [cal_c, cal_d],
})
rtl_1 = RTL(num_lattices=5, lattice_rank=4)(rtl_0)
outputs = tfl.layers.Linear(
    num_input_dims=5,
    monotonicities=['increasing'] * 5,
)(rtl_1)
model = tf.keras.Model(inputs=[a, b, c, d], outputs=outputs)

num_lattices Number of lattices in the ensemble.
lattice_rank Number of features used in each lattice.
lattice_size Number of lattice vertices per dimension (minimum is 2).
output_min None or lower bound of the output.
output_max None or upper bound of the output.
init_min None or lower bound of lattice kernel initialization.
init_max None or upper bound of lattice kernel initialization.
separate_outputs If set to true, the output will be a dict in the same format as the input to the layer, ready to be passed to another RTL layer. If false, the output will be a single tensor of shape (batch_size, num_lattices). See output shape for details.
random_seed Random seed for the randomized feature arrangement in the ensemble. Also used for initialization of lattices using 'kronecker_factored' parameterization.
num_projection_iterations Number of iterations of Dykstra projections algorithm. Projection updates will be closer to a true projection (with respect to the L2 norm) with higher number of iterations. Increasing this number has diminishing return on projection precsion. Infinite number of iterations would yield perfect projection. Increasing this number might slightly improve convergence by cost of slightly increasing running time. Most likely you want this number to be proportional to number of lattice vertices in largest constrained dimension.
monotonic_at_every_step Whether to strictly enforce monotonicity and trust constraints after every gradient update by applying a final imprecise projection. Setting this parameter to True together with small num_projection_iterations parameter is likely to hurt convergence.
clip_inputs If inputs should be clipped to the input range of the lattice.
interpolation One of 'hypercube' or 'simplex' interpolation. For a d-dimensional lattice, 'hypercube' interpolates 2^d parameters, whereas 'simplex' uses d+1 parameters and thus scales better. For details see tfl.lattice_lib.evaluate_with_simplex_interpolation and tfl.lattice_lib.evaluate_with_hypercube_interpolation.
parameterization The parameterization of the lattice function class to use. A lattice function is uniquely determined by specifying its value on every lattice vertex. A parameterization scheme is a mapping from a vector of parameters to a multidimensional array of lattice vertex values. It can be one of:

  • String 'all_vertices': This is the "traditional" parameterization that keeps one scalar parameter per lattice vertex where the mapping is essentially the identity map. With this scheme, the number of parameters scales exponentially with the number of inputs to the lattice. The underlying lattices used will be tfl.layers.Lattice layers.
  • String 'kronecker_factored': With this parameterization, for each lattice input i we keep a collection of num_terms vectors each having feature_configs[0].lattice_size entries (note that the lattice size of the first feature will be used as the lattice size for all other features as well). To obtain the tensor of lattice vertex values, for t=1,2,...,num_terms we compute the outer product of the t'th vector in each collection, multiply by a per-term scale, and sum the resulting tensors. Finally, we add a single shared bias parameter to each entry in the sum. With this scheme, the number of parameters grows linearly with lattice_rank (assuming lattice sizes and num_terms are held constant). Currently, only monotonicity shape constraint and bound constraint are supported for this scheme. Regularization is not currently supported. The underlying lattices used will be tfl.layers.KroneckerFactoredLattice layers.
num_terms The number of terms in a lattice using 'kronecker_factored' parameterization. Ignored if parameterization is set to 'all_vertices'.
avoid_intragroup_interaction If set to true, the RTL algorithm will try to avoid having inputs from the same group in the same lattice.
kernel_initializer One of:
  • 'linear_initializer': initialize parameters to form a linear function with positive and equal coefficients for monotonic dimensions and 0.0 coefficients for other dimensions. Linear function is such that minimum possible output is equal to output_min and maximum possible output is equal to output_max. See tfl.lattice_layer.LinearInitializer class docstring for more details. This initialization is not supported when using the 'kronecker_factored' parameterization.
  • 'random_monotonic_initializer': initialize parameters uniformly at random such that all parameters are monotonically increasing for each input. Parameters will be sampled uniformly at random from the range [init_min, init_max] if specified, otherwise [output_min, output_max]. See tfl.lattice_layer.RandomMonotonicInitializer class docstring for more details. This initialization is not supported when using the 'kronecker_factored' parameterization.
  • 'kfl_random_monotonic_initializer': initialize parameters uniformly at random such that all parameters are monotonically increasing for each monotonic input. Parameters will be sampled uniformly at random from the range [init_min, init_max] if specified. Otherwise, the initialization range will be algorithmically determined depending on output_{min/max}. See tfl.layers.KroneckerFactoredLattice and tfl.kronecker_factored_lattice.KFLRandomMonotonicInitializer class docstrings for more details. This initialization is not supported when using 'all_vertices' parameterization.
  • kernel_regularizer None or a single element or a list of following:
  • Tuple ('torsion', l1, l2) or List ['torsion', l1, l2] where l1 and l2 represent corresponding regularization amount for graph Torsion regularizer. l1 and l2 must be single floats. Lists of floats to specify different regularization amount for every dimension is not currently supported.
  • Tuple ('laplacian', l1, l2) or List ['laplacian', l1, l2] where l1 and l2 represent corresponding regularization amount for graph Laplacian regularizer. l1 and l2 must be single floats. Lists of floats to specify different regularization amount for every dimension is not currently supported.
  • average_outputs Whether to average the outputs of this layer. Ignored when separate_outputs is True.
    **kwargs Other args passed to tf.keras.layers.Layer initializer.

    ValueError If layer hyperparameters are invalid.
    ValueError If parameterization is not one of 'all_vertices' or 'kronecker_factored'.

    • All __init__arguments.
    activity_regularizer Optional regularizer function for the output of this layer.
    compute_dtype The dtype of the layer's computations.

    This is equivalent to Layer.dtype_policy.compute_dtype. Unless mixed precision is used, this is the same as Layer.dtype, the dtype of the weights.

    Layers automatically cast their inputs to the compute dtype, which causes computations and the output to be in the compute dtype as well. This is done by the base Layer class in Layer.call, so you do not have to insert these casts if implementing your own layer.

    Layers often perform certain internal computations in higher precision when compute_dtype is float16 or bfloat16 for numeric stability. The output will still typically be float16 or bfloat16 in such cases.

    dtype The dtype of the layer weights.

    This is equivalent to Layer.dtype_policy.variable_dtype. Unless mixed precision is used, this is the same as Layer.compute_dtype, the dtype of the layer's computations.

    dtype_policy The dtype policy associated with this layer.

    This is an instance of a tf.keras.mixed_precision.Policy.

    dynamic Whether the layer is dynamic (eager-only); set in the constructor.
    input Retrieves the input tensor(s) of a layer.

    Only applicable if the layer has exactly one input, i.e. if it is connected to one incoming layer.

    input_spec InputSpec instance(s) describing the input format for this layer.

    When you create a layer subclass, you can set self.input_spec to enable the layer to run input compatibility checks when it is called. Consider a Conv2D layer: it can only be called on a single input tensor of rank 4. As such, you can set, in __init__():

    self.input_spec = tf.keras.layers.InputSpec(ndim=4)
    

    Now, if you try to call the layer on an input that isn't rank 4 (for instance, an input of shape (2,), it will raise a nicely-formatted error:

    ValueError: Input 0 of layer conv2d is incompatible with the layer:
    expected ndim=4, found ndim=1. Full shape received: [2]
    

    Input checks that can be specified via input_spec include:

    • Structure (e.g. a single input, a list of 2 inputs, etc)
    • Shape
    • Rank (ndim)
    • Dtype

    For more information, see tf.keras.layers.InputSpec.

    losses List of losses added using the add_loss() API.

    Variable regularization tensors are created when this property is accessed, so it is eager safe: accessing losses under a tf.GradientTape will propagate gradients back to the corresponding variables.

    class MyLayer(tf.keras.layers.Layer):
      def call(self, inputs):
        self.add_loss(tf.abs(tf.reduce_mean(inputs)))
        return inputs
    l = MyLayer()
    l(np.ones((10, 1)))
    l.losses
    [1.0]
    
    inputs = tf.keras.Input(shape=(10,))
    x = tf.keras.layers.Dense(10)(inputs)
    outputs = tf.keras.layers.Dense(1)(x)
    model = tf.keras.Model(inputs, outputs)
    # Activity regularization.
    len(model.losses)
    0
    model.add_loss(tf.abs(tf.reduce_mean(x)))
    len(model.losses)
    1
    
    inputs = tf.keras.Input(shape=(10,))
    d = tf.keras.layers.Dense(10, kernel_initializer='ones')
    x = d(inputs)
    outputs = tf.keras.layers.Dense(1)(x)
    model = tf.keras.Model(inputs, outputs)
    # Weight regularization.
    model.add_loss(lambda: tf.reduce_mean(d.kernel))
    model.losses
    [<tf.Tensor: shape=(), dtype=float32, numpy=1.0>]
    

    metrics List of metrics attached to the layer.
    name Name of the layer (string), set in the constructor.
    name_scope Returns a tf.name_scope instance for this class.
    non_trainable_weights List of all non-trainable weights tracked by this layer.

    Non-trainable weights are not updated during training. They are expected to be updated manually in call().

    output Retrieves the output tensor(s) of a layer.

    Only applicable if the layer has exactly one output, i.e. if it is connected to one incoming layer.

    submodules Sequence of all sub-modules.

    Submodules are modules which are properties of this module, or found as properties of modules which are properties of this module (and so on).

    a = tf.Module()
    b = tf.Module()
    c = tf.Module()
    a.b = b
    b.c = c
    list(a.submodules) == [b, c]
    True
    list(b.submodules) == [c]
    True
    list(c.submodules) == []
    True
    

    supports_masking Whether this layer supports computing a mask using compute_mask.
    trainable

    trainable_weights List of all trainable weights tracked by this layer.

    Trainable weights are updated via gradient descent during training.

    variable_dtype Alias of Layer.dtype, the dtype of the weights.
    weights Returns the list of all layer variables/weights.

    Methods

    add_loss

    Add loss tensor(s), potentially dependent on layer inputs.

    Some losses (for instance, activity regularization losses) may be dependent on the inputs passed when calling a layer. Hence, when reusing the same layer on different inputs a and b, some entries in layer.losses may be dependent on a and some on b. This method automatically keeps track of dependencies.

    This method can be used inside a subclassed layer or model's call function, in which case losses should be a Tensor or list of Tensors.

    Example:

    class MyLayer(tf.keras.layers.Layer):
      def call(self, inputs):
        self.add_loss(tf.abs(tf.reduce_mean(inputs)))
        return inputs
    

    The same code works in distributed training: the input to add_loss() is treated like a regularization loss and averaged across replicas by the training loop (both built-in Model.fit() and compliant custom training loops).

    The add_loss method can also be called directly on a Functional Model during construction. In this case, any loss Tensors passed to this Model must be symbolic and be able to be traced back to the model's Inputs. These losses become part of the model's topology and are tracked in get_config.

    Example:

    inputs = tf.keras.Input(shape=(10,))
    x = tf.keras.layers.Dense(10)(inputs)
    outputs = tf.keras.layers.Dense(1)(x)
    model = tf.keras.Model(inputs, outputs)
    # Activity regularization.
    model.add_loss(tf.abs(tf.reduce_mean(x)))
    

    If this is not the case for your loss (if, for example, your loss references a Variable of one of the model's layers), you can wrap your loss in a zero-argument lambda. These losses are not tracked as part of the model's topology since they can't be serialized.

    Example:

    inputs = tf.keras.Input(shape=(10,))
    d = tf.keras.layers.Dense(10)
    x = d(inputs)
    outputs = tf.keras.layers.Dense(1)(x)
    model = tf.keras.Model(inputs, outputs)
    # Weight regularization.
    model.add_loss(lambda: tf.reduce_mean(d.kernel))
    

    Args
    losses Loss tensor, or list/tuple of tensors. Rather than tensors, losses may also be zero-argument callables which create a loss tensor.
    **kwargs Used for backwards compatibility only.

    assert_constraints

    View source

    Asserts that weights satisfy all constraints.

    In graph mode builds and returns a list of assertion ops. In eager mode directly executes assertions.

    Args
    eps allowed constraints violation.

    Returns
    List of assertion ops in graph mode or immediately asserts in eager mode.

    build

    View source

    Standard Keras build() method.

    build_from_config

    Builds the layer's states with the supplied config dict.

    By default, this method calls the build(config["input_shape"]) method, which creates weights based on the layer's input shape in the supplied config. If your config contains other information needed to load the layer's state, you should override this method.

    Args
    config Dict containing the input shape associated with this layer.

    compute_mask

    Computes an output mask tensor.

    Args
    inputs Tensor or list of tensors.
    mask Tensor or list of tensors.

    Returns
    None or a tensor (or list of tensors, one per output tensor of the layer).

    compute_output_shape

    View source

    Standard Keras compute_output_shape() method.

    count_params

    Count the total number of scalars composing the weights.

    Returns
    An integer count.

    Raises
    ValueError if the layer isn't yet built (in which case its weights aren't yet defined).

    finalize_constraints

    View source

    Ensures layers weights strictly satisfy constraints.

    Applies approximate projection to strictly satisfy specified constraints. If monotonic_at_every_step == True there is no need to call this function.

    Returns
    In eager mode directly updates weights and returns variable which stores them. In graph mode returns a list of assign_add op which has to be executed to updates weights.

    from_config

    Creates a layer from its config.

    This method is the reverse of get_config, capable of instantiating the same layer from the config dictionary. It does not handle layer connectivity (handled by Network), nor weights (handled by set_weights).

    Args
    config A Python dictionary, typically the output of get_config.

    Returns
    A layer instance.

    get_build_config

    Returns a dictionary with the layer's input shape.

    This method returns a config dict that can be used by build_from_config(config) to create all states (e.g. Variables and Lookup tables) needed by the layer.

    By default, the config only contains the input shape that the layer was built with. If you're writing a custom layer that creates state in an unusual way, you should override this method to make sure this state is already created when Keras attempts to load its value upon model loading.

    Returns
    A dict containing the input shape associated with the layer.

    get_config

    View source

    Standard Keras get_config() method.

    get_weights

    Returns the current weights of the layer, as NumPy arrays.

    The weights of a layer represent the state of the layer. This function returns both trainable and non-trainable weight values associated with this layer as a list of NumPy arrays, which can in turn be used to load state into similarly parameterized layers.

    For example, a Dense layer returns a list of two values: the kernel matrix and the bias vector. These can be used to set the weights of another Dense layer:

    layer_a = tf.keras.layers.Dense(1,
      kernel_initializer=tf.constant_initializer(1.))
    a_out = layer_a(tf.convert_to_tensor([[1., 2., 3.]]))
    layer_a.get_weights()
    [array([[1.],
           [1.],
           [1.]], dtype=float32), array([0.], dtype=float32)]
    layer_b = tf.keras.layers.Dense(1,
      kernel_initializer=tf.constant_initializer(2.))
    b_out = layer_b(tf.convert_to_tensor([[10., 20., 30.]]))
    layer_b.get_weights()
    [array([[2.],
           [2.],
           [2.]], dtype=float32), array([0.], dtype=float32)]
    layer_b.set_weights(layer_a.get_weights())
    layer_b.get_weights()
    [array([[1.],
           [1.],
           [1.]], dtype=float32), array([0.], dtype=float32)]
    

    Returns
    Weights values as a list of NumPy arrays.

    load_own_variables

    Loads the state of the layer.

    You can override this method to take full control of how the state of the layer is loaded upon calling keras.models.load_model().

    Args
    store Dict from which the state of the model will be loaded.

    save_own_variables

    Saves the state of the layer.

    You can override this method to take full control of how the state of the layer is saved upon calling model.save().

    Args
    store Dict where the state of the model will be saved.

    set_weights

    Sets the weights of the layer, from NumPy arrays.

    The weights of a layer represent the state of the layer. This function sets the weight values from numpy arrays. The weight values should be passed in the order they are created by the layer. Note that the layer's weights must be instantiated before calling this function, by calling the layer.

    For example, a Dense layer returns a list of two values: the kernel matrix and the bias vector. These can be used to set the weights of another Dense layer:

    layer_a = tf.keras.layers.Dense(1,
      kernel_initializer=tf.constant_initializer(1.))
    a_out = layer_a(tf.convert_to_tensor([[1., 2., 3.]]))
    layer_a.get_weights()
    [array([[1.],
           [1.],
           [1.]], dtype=float32), array([0.], dtype=float32)]
    layer_b = tf.keras.layers.Dense(1,
      kernel_initializer=tf.constant_initializer(2.))
    b_out = layer_b(tf.convert_to_tensor([[10., 20., 30.]]))
    layer_b.get_weights()
    [array([[2.],
           [2.],
           [2.]], dtype=float32), array([0.], dtype=float32)]
    layer_b.set_weights(layer_a.get_weights())
    layer_b.get_weights()
    [array([[1.],
           [1.],
           [1.]], dtype=float32), array([0.], dtype=float32)]
    

    Args
    weights a list of NumPy arrays. The number of arrays and their shape must match number of the dimensions of the weights of the layer (i.e. it should match the output of get_weights).

    Raises
    ValueError If the provided weights list does not match the layer's specifications.

    with_name_scope

    Decorator to automatically enter the module name scope.

    class MyModule(tf.Module):
      @tf.Module.with_name_scope
      def __call__(self, x):
        if not hasattr(self, 'w'):
          self.w = tf.Variable(tf.random.normal([x.shape[1], 3]))
        return tf.matmul(x, self.w)
    

    Using the above module would produce tf.Variables and tf.Tensors whose names included the module name:

    mod = MyModule()
    mod(tf.ones([1, 2]))
    <tf.Tensor: shape=(1, 3), dtype=float32, numpy=..., dtype=float32)>
    mod.w
    <tf.Variable 'my_module/Variable:0' shape=(2, 3) dtype=float32,
    numpy=..., dtype=float32)>
    

    Args
    method The method to wrap.

    Returns
    The original method wrapped such that it enters the module's name scope.

    __call__

    Wraps call, applying pre- and post-processing steps.

    Args
    *args Positional arguments to be passed to self.call.
    **kwargs Keyword arguments to be passed to self.call.

    Returns
    Output tensor(s).

    Note

    • The following optional keyword arguments are reserved for specific uses:
      • training: Boolean scalar tensor of Python boolean indicating whether the call is meant for training or inference.
      • mask: Boolean input mask.
    • If the layer's call method takes a mask argument (as some Keras layers do), its default value will be set to the mask generated for inputs by the previous layer (if input did come from a layer that generated a corresponding mask, i.e. if it came from a Keras layer with masking support.
    • If the layer is not built, the method will call build.

    Raises
    ValueError if the layer's call method returns None (an invalid value).
    RuntimeError if super().__init__() was not called in the constructor.