Module: tfl.kronecker_factored_lattice_lib

Algorithm implementations required for Kronecker-Factored Lattice layer.

Functions

assert_constraints(...): Asserts that weights satisfy constraints.

evaluate_with_hypercube_interpolation(...): Evaluates a Kronecker-Factored Lattice using hypercube interpolation.

finalize_constraints(...): Approximately projects weights to strictly satisfy all constraints.

random_monotonic_initializer(...): Returns a uniformly random sampled monotonic weight tensor.

scale_initializer(...): Initializes scale to alternate between 1 and -1 for each term.

verify_hyperparameters(...): Verifies that all given hyperparameters are consistent.