@datafire/azure_machinelearningservices_hyperdrive

Client library for HyperDrive

Installation and Usage

npm install --save @datafire/azure_machinelearningservices_hyperdrive

let azure_machinelearningservices_hyperdrive = require('@datafire/azure_machinelearningservices_hyperdrive').create({
  access_token: "",
  refresh_token: "",
  client_id: "",
  client_secret: "",
  redirect_uri: ""
});

.then(data => {
  console.log(data);
});

Description

HyperDrive REST API

Actions

HyperparameterTuning_CreateExperiment

Create a HyperDrive Experiment.

azure_machinelearningservices_hyperdrive.HyperparameterTuning_CreateExperiment({
  "armScope": "",
  "config": ""
}, context)

Input

• input object
  • armScope required string: The ARM scope passed in through URL with format:
  • config required string, object: The configuration file with experiment JSON content. A text file that is a JSON-serialized '#/definitions/HyperDriveCreateExperiment' object.
    • content string
    • encoding string (values: ascii, utf8, utf16le, base64, binary, hex)
    • contentType string
    • filename string

Output

• output HyperDriveExperimentResponse

HyperparameterTuning_CancelExperiment

Cancel a HyperDrive Experiment.

azure_machinelearningservices_hyperdrive.HyperparameterTuning_CancelExperiment({
  "armScope": "",
  "runId": ""
}, context)

Input

• input object
  • armScope required string: The ARM scope passed in through URL with format:
  • runId required string: Hyperdrive run id to cancel.
  • RunHistoryHost string: The host for run location.

Output

• output HyperDriveCancelExperimentResponse

Definitions

HyperDriveBanditPolicy

• HyperDriveBanditPolicy object: Bandit Policy configuration. Please refer https://docs.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive.banditpolicy?view=azure-ml-py for more information.
  • name required string (values: Default, Bandit, MedianStopping, TruncationSelection): Type of early termination policy.
  • properties object: Policy configuration properties.
    • delay_evaluation integer: Value indicating the number of sequences for which the first evaluation is delayed.
    • evaluation_interval integer: Evaluation interval of the policy.
    • slack_amount number: Slack amount.
    • slack_factor number: Slack factor.

HyperDriveCancelExperimentResponse

• HyperDriveCancelExperimentResponse object: Response for HyperDrive_CancelExperiment in case of success.
  • code required integer: Response code.
  • result required string: Message indicating operation success.

HyperDriveCreateExperiment

• HyperDriveCreateExperiment object: Part of request for HyperDrive_CreateExperiment. Contains configuration details required to create hyperdrive run.
  • description string: The description for Hyperdrive run.
  • generator_config required object: Hyperparameter space and the sampling method configuration.
    • name required string (values: RANDOM, GRID, BAYESIANOPTIMIZATION): Hyperparameter sampling method.
    • parameter_space required object: Dictionary specifying hyperparameter space.
  • max_concurrent_jobs integer: Maximum number of runs to run concurrently.
  • max_duration_minutes integer: Maximum duration of the Hyperdrive run.
  • max_total_jobs integer: Maximum number of runs.
  • name required string: Name of the Hyperdrive run.
  • platform required string (values: AML): Platform of the Hyperdrive run.
  • platform_config required object: Platform config object specifying the run definition structure.
  • policy_config required HyperDrivePolicyConfigBase
  • primary_metric_config required object: Name of the primary metric and goal of optimizing.
    • goal required string (values: MAXIMIZE, MINIMIZE): Determines if the primary metric has to be minimized/maximized.
    • name required string: Name of the primary metric reported by runs.
  • study_id integer: Study Id of the Hyperdrive run.
  • user required string: User who is creating the Hyperdrive run.

HyperDriveDefaultPolicy

• HyperDriveDefaultPolicy object: No early termination is applied in the case of DefaultPolicy
  • name required string (values: Default, Bandit, MedianStopping, TruncationSelection): Type of early termination policy.

HyperDriveErrorResponse

• HyperDriveErrorResponse object: Response in case of an error.
  • code required integer: Error response code.
  • message required string: Error message.

HyperDriveExperiment

• HyperDriveExperiment object: Part of response for HyperDrive_CreateExperiment in case of success. Contains details about the created hyperdrive run.
  • description string: The description for Hyperdrive run.
  • generator_config required object: Hyperparameter space and the sampling method configuration.
    • name required string (values: RANDOM, GRID, BAYESIANOPTIMIZATION): Hyperparameter sampling method.
    • parameter_space required object: Dictionary specifying hyperparameter space.
  • max_concurrent_jobs integer: Maximum number of runs to run concurrently.
  • max_duration_minutes integer: Maximum duration of the Hyperdrive run.
  • max_total_jobs integer: Maximum number of runs.
  • name required string: Name of the Hyperdrive run.
  • platform required string (values: AML): Platform of the Hyperdrive run.
  • platform_config required object: Platform config object specifying the run definition structure.
  • policy_config required HyperDrivePolicyConfigBase
  • primary_metric_config required object: Name of the primary metric and goal of optimizing.
    • goal required string (values: MAXIMIZE, MINIMIZE): Determines if the primary metric has to be minimized/maximized.
    • name required string: Name of the primary metric reported by runs.
  • study_id integer: Study Id of the Hyperdrive run.
  • all_jobs_generated required boolean: Indicates if all runs have been generated.
  • cancellation_requested required boolean: Indicates if cancellation has been requested for this Hyperdrive run.
  • created required string: Hyperdrive run creation time.
  • created_by_user_id required integer: Id of the user who created the Hyperdrive run.
  • experiment_id required string: Hyperdrive run id.
  • experiment_uri string: Hyperdrive run Uri.
  • hyperdrive_run_id required string: Hyperdrive run id.
  • modified required string: Hyperdrive run modification time.
  • status required string: Hyperdrive run status.
  • study_uri string: Study Uri of the Hyperdrive run.

HyperDriveExperimentBase

• HyperDriveExperimentBase object: Base object for both request and response of HyperDrive_CreateExperiment api.
  • description string: The description for Hyperdrive run.
  • generator_config required object: Hyperparameter space and the sampling method configuration.
    • name required string (values: RANDOM, GRID, BAYESIANOPTIMIZATION): Hyperparameter sampling method.
    • parameter_space required object: Dictionary specifying hyperparameter space.
  • max_concurrent_jobs integer: Maximum number of runs to run concurrently.
  • max_duration_minutes integer: Maximum duration of the Hyperdrive run.
  • max_total_jobs integer: Maximum number of runs.
  • name required string: Name of the Hyperdrive run.
  • platform required string (values: AML): Platform of the Hyperdrive run.
  • platform_config required object: Platform config object specifying the run definition structure.
  • policy_config required HyperDrivePolicyConfigBase
  • primary_metric_config required object: Name of the primary metric and goal of optimizing.
    • goal required string (values: MAXIMIZE, MINIMIZE): Determines if the primary metric has to be minimized/maximized.
    • name required string: Name of the primary metric reported by runs.
  • study_id integer: Study Id of the Hyperdrive run.

HyperDriveExperimentResponse

• HyperDriveExperimentResponse object: Response for HyperDrive_CreateExperiment in case of success.
  • code required integer: Response code.
  • result required HyperDriveExperiment

HyperDriveMedianStoppingPolicy

• HyperDriveMedianStoppingPolicy object: Median stopping policy configuration. Please refer https://docs.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive.medianstoppingpolicy?view=azure-ml-py for more information.
  • name required string (values: Default, Bandit, MedianStopping, TruncationSelection): Type of early termination policy.
  • properties object: Policy configuration properties.
    • delay_evaluation integer: Value indicating the number of sequences for which the first evaluation is delayed.
    • evaluation_interval integer: Evaluation interval of the policy.

HyperDrivePolicyConfigBase

• HyperDrivePolicyConfigBase object: Early termination policy configuration.
  • name required string (values: Default, Bandit, MedianStopping, TruncationSelection): Type of early termination policy.

HyperDriveTruncationSelectionPolicy

• HyperDriveTruncationSelectionPolicy object: Truncation selection policy configuration. Please refer https://docs.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive.truncationselectionpolicy?view=azure-ml-py for more information.
  • name required string (values: Default, Bandit, MedianStopping, TruncationSelection): Type of early termination policy.
  • properties object: Policy configuration properties.
    • delay_evaluation integer: Value indicating the number of sequences for which the first evaluation is delayed.
    • evaluation_interval integer: Evaluation interval of the policy.
    • exclude_finished_jobs boolean: Boolean indicating if metrics from finished jobs should be excluded in the policy decision process.
    • truncation_percentage integer: Truncation percentage value.