|View source on GitHub|
DEPRECATED: Please use
tfx.components.model_validator.executor.Executor( context: Optional[
tfx.dsl.components.base.base_executor.BaseExecutor.Context] = None )
The model validator helps prevent bad models from being pushed to production. It does this by validating exported models against known good models (e.g. the current production model), and marking the exported model as good ("blessing it") only if the exported model's metrics are within predefined thresholds around the good model's metrics.
The model validator will validate tf.serving format exported models produced by the Trainer component. The validator evaluates the models on examples created by the ExampleGen component. The validator will also automatically read data written by the Pusher component regarding the latest pushed models by using ml.metadata to query the previously pushed artifacts.
To include ModelValidator in a TFX pipeline, configure your pipeline similar to https://github.com/tensorflow/tfx/blob/master/tfx/examples/chicago_taxi_pipeline/taxi_pipeline_simple.py#L110
Do( input_dict: Dict[Text, List[
tfx.types.Artifact]], output_dict: Dict[Text, List[
tfx.types.Artifact]], exec_properties: Dict[Text, Any] ) -> None
Validate current model against last blessed model.
Input dict from input key to a list of Artifacts.
Output dict from output key to a list of Artifacts.
A dict of execution properties.