NlpSweepSettings Class

Sweep settings for all AutoML NLP tasks.

Inheritance

azure.ai.ml.entities._mixins.RestTranslatableMixin

NlpSweepSettings

Constructor

NlpSweepSettings(*, sampling_algorithm: str | SamplingAlgorithmType, early_termination: EarlyTerminationPolicy | None = None)

Parameters

Name	Description
sampling_algorithm Required	Union[str, SamplingAlgorithmType] Required. Specifies type of hyperparameter sampling algorithm. Possible values include: "Grid", "Random", and "Bayesian".
early_termination Required	Optional[EarlyTerminationPolicy] Early termination policy to end poorly performing training candidates, defaults to None.

Keyword-Only Parameters

Name	Description
sampling_algorithm Required
early_termination Required

Examples

creating an nlp sweep settings

   from azure.ai.ml import automl
   from azure.ai.ml.sweep import BanditPolicy

   nlp_sweep_settings = automl.NlpSweepSettings(
       sampling_algorithm="Grid",
       early_termination=BanditPolicy(evaluation_interval=2, slack_factor=0.05, delay_evaluation=6),
   )