Skip to content

/AWS1/CL_SGMHYPERBANDSTRAGCFG

The configuration for Hyperband, a multi-fidelity based hyperparameter tuning strategy. Hyperband uses the final and intermediate results of a training job to dynamically allocate resources to utilized hyperparameter configurations while automatically stopping under-performing configurations. This parameter should be provided only if Hyperband is selected as the StrategyConfig under the HyperParameterTuningJobConfig API.

CONSTRUCTOR

IMPORTING

Optional arguments:

iv_minresource TYPE /AWS1/SGMHYPBANDSTRAGMINRESRC /AWS1/SGMHYPBANDSTRAGMINRESRC

The minimum number of resources (such as epochs) that can be used by a training job launched by a hyperparameter tuning job. If the value for MinResource has not been reached, the training job is not stopped by Hyperband.

iv_maxresource TYPE /AWS1/SGMHYPBANDSTRAGMAXRESRC /AWS1/SGMHYPBANDSTRAGMAXRESRC

The maximum number of resources (such as epochs) that can be used by a training job launched by a hyperparameter tuning job. Once a job reaches the MaxResource value, it is stopped. If a value for MaxResource is not provided, and Hyperband is selected as the hyperparameter tuning strategy, HyperbandTraining attempts to infer MaxResource from the following keys (if present) in StaticsHyperParameters:

  • epochs

  • numepochs

  • n-epochs

  • n_epochs

  • num_epochs

If HyperbandStrategyConfig is unable to infer a value for MaxResource, it generates a validation error. The maximum value is 20,000 epochs. All metrics that correspond to an objective metric are used to derive early stopping decisions. For distributed training jobs, ensure that duplicate metrics are not printed in the logs across the individual nodes in a training job. If multiple nodes are publishing duplicate or incorrect metrics, training jobs may make an incorrect stopping decision and stop the job prematurely.

Queryable Attributes

MinResource

The minimum number of resources (such as epochs) that can be used by a training job launched by a hyperparameter tuning job. If the value for MinResource has not been reached, the training job is not stopped by Hyperband.

Accessible with the following methods

Method Description
GET_MINRESOURCE() Getter for MINRESOURCE, with configurable default
ASK_MINRESOURCE() Getter for MINRESOURCE w/ exceptions if field has no value
HAS_MINRESOURCE() Determine if MINRESOURCE has a value

MaxResource

The maximum number of resources (such as epochs) that can be used by a training job launched by a hyperparameter tuning job. Once a job reaches the MaxResource value, it is stopped. If a value for MaxResource is not provided, and Hyperband is selected as the hyperparameter tuning strategy, HyperbandTraining attempts to infer MaxResource from the following keys (if present) in StaticsHyperParameters:

  • epochs

  • numepochs

  • n-epochs

  • n_epochs

  • num_epochs

If HyperbandStrategyConfig is unable to infer a value for MaxResource, it generates a validation error. The maximum value is 20,000 epochs. All metrics that correspond to an objective metric are used to derive early stopping decisions. For distributed training jobs, ensure that duplicate metrics are not printed in the logs across the individual nodes in a training job. If multiple nodes are publishing duplicate or incorrect metrics, training jobs may make an incorrect stopping decision and stop the job prematurely.

Accessible with the following methods

Method Description
GET_MAXRESOURCE() Getter for MAXRESOURCE, with configurable default
ASK_MAXRESOURCE() Getter for MAXRESOURCE w/ exceptions if field has no value
HAS_MAXRESOURCE() Determine if MAXRESOURCE has a value