VanishingGradient Rule - Amazon SageMaker

VanishingGradient Rule

This rule detects if the gradients in a trial become extremely small or drop to a zero magnitude. If the mean of the absolute values of the gradients drops below a specified threshold, the rule returns True.

Parameters Descriptions for the VanishingGradient Rule
Parameter Name Description
base_trial

The trial run using this rule. The rule inspects the tensors gathered from this trial.

Required

Valid values: String

threshold The value at which the gradient is determined to be vanishing.

Optional

Valid values: Float

Default value: 0.0000001.

For an example of how to configure and deploy a built-in rule, see How to Use Built-in Rules for Model Analysis.

Note

This rule can't be applied to the XGBoost algorithm.