def _attach_with_training_details_list()

in src/sagemaker/tuner.py [0:0]

• 43 lines of code
• 5 McCabe index (conditional complexity)

    def _attach_with_training_details_list(cls, sagemaker_session, estimator_cls, job_details):
        """Create a HyperparameterTuner bound to an existing hyperparameter tuning job.

        The tuning job has the ``TrainingJobDefinitions`` field set in this case.
        """
        estimator_names = sorted(
            [
                training_details["DefinitionName"]
                for training_details in job_details["TrainingJobDefinitions"]
            ]
        )
        cls._validate_dict_argument(
            name="estimator_cls", value=estimator_cls, allowed_keys=estimator_names
        )

        estimator_dict = {}
        objective_metric_name_dict = {}
        hyperparameter_ranges_dict = {}
        metric_definitions_dict = {}

        for training_details in job_details["TrainingJobDefinitions"]:
            estimator_name = training_details["DefinitionName"]

            estimator_dict[estimator_name] = cls._prepare_estimator(
                estimator_cls=estimator_cls.get(estimator_name) if estimator_cls else None,
                training_details=training_details,
                parameter_ranges=training_details["HyperParameterRanges"],
                sagemaker_session=sagemaker_session,
            )

            objective_metric_name_dict[estimator_name] = training_details["TuningObjective"][
                "MetricName"
            ]
            hyperparameter_ranges_dict[
                estimator_name
            ] = cls._prepare_parameter_ranges_from_job_description(  # noqa: E501 # pylint: disable=line-too-long
                training_details["HyperParameterRanges"]
            )

            metric_definitions = training_details["AlgorithmSpecification"].get(
                "MetricDefinitions", None
            )
            if metric_definitions is not None:
                metric_definitions_dict[estimator_name] = metric_definitions

        init_params = cls._prepare_init_params_from_job_description(job_details)

        return HyperparameterTuner.create(
            estimator_dict=estimator_dict,
            objective_metric_name_dict=objective_metric_name_dict,
            hyperparameter_ranges_dict=hyperparameter_ranges_dict,
            metric_definitions_dict=metric_definitions_dict,
            **init_params
        )