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

tfl.visualization.plot_feature_calibrator

View source on GitHub

Plots feature calibrator(s) extracted from a TFL canned estimator.

tfl.visualization.plot_feature_calibrator(
    model_graph, feature_name, plot_submodel_calibration=True, font_size=12,
    axis_label_font_size=14, figsize=None
)

Used in the notebooks

Used in the tutorials
model_graph = estimators.get_model_graph(saved_model_path)
visualization.plot_feature_calibrator(model_graph, "feature_name")

Args:

  • model_graph: model_info.ModelGraph object that includes model nodes.
  • feature_name: Name of the feature to plot the calibrator for.
  • plot_submodel_calibration: If submodel calibrators should be included in the output plot, when more than one calibration node is provided. These are individual calibration layers for each lattice in a lattice ensemble constructed from configs.CalibratedLatticeEnsembleConfig.
  • font_size: Font size for values and labels on the plot.
  • axis_label_font_size: Font size for axis labels.
  • figsize: The figsize parameter passed to pyplot.figure().

Returns:

Pyplot figure object containing the visualisation.