Missed TensorFlow Dev Summit? Check out the video playlist. Watch recordings

Module: tfl

View source on GitHub

Tensorflow Lattice Library.

This package provides functions and classes for lattice modeling.

Modules

categorical_calibration_layer module: Categorical calibration layer with monotonicity and bound constraints.

categorical_calibration_lib module: Helpers and computations of categorical calibration layer.

configs module: TFL model configuration library for canned estimators.

estimators module: TF Lattice canned estimators implement typical monotonic model architectures.

lattice_layer module: Lattice layer with monotonicity, unimodality, trust and bound constraints.

lattice_lib module: Implementation of algorithms required for Lattice layer.

layers module: 'layers' namespace for TFL layers.

linear_layer module: Layer which represents linear function. See class level comment.

linear_lib module: Implementation of algorithms required for Linear layer.

model_info module: Classes defining trained TFL model structure and parameter information.

parallel_combination_layer module: ParallelCombination layer for combining several parallel calibration layers.

premade module: TF Lattice premade models implement typical monotonic model architectures.

premade_lib module: Implementation of algorithms required for premade models.

pwl_calibration_layer module: Piecewise linear calibration layer.

pwl_calibration_lib module: Implementation of algorithms required for PWL calibration layer.

test_utils module: Helpers to train simple model for tests and print debug output.

visualization module: Tools to analyse and plot TFL models.