Module: tfl.model_info

Classes defining trained TFL model structure and parameter information.

This package provides representations and tools for analysis of a trained TF Lattice model, e.g. a canned estimator in saved model format.

Classes

class CategoricalCalibrationNode: Represetns a categorical calibration layer.

class InputFeatureNode: Input features to the model.

class LatticeNode: Represetns a lattice layer.

class LinearNode: Represents a linear layer.

class MeanNode: Represents an averaging layer.

class ModelGraph: Model info and parameter as a graph.

class PWLCalibrationNode: Represetns a PWL calibration layer.