ai.djl.training.metrics

Class TrainingMetric

java.lang.Object

ai.djl.training.metrics.TrainingMetric

All Implemented Interfaces:

java.lang.Cloneable

Direct Known Subclasses:

Accuracy, BoundingBoxError, Loss

public abstract class TrainingMetric
extends java.lang.Object
implements java.lang.Cloneable

Base class for all training metrics.

Training metrics can be used to evaluate the performance of a model. They can all be monitored to make an assessment about the performance of the model. However, only ones that further extend Loss are suited to being used to directly optimize a model.

Constructor Summary

Constructors

Constructor and Description
TrainingMetric(java.lang.String name) Creates a metric with abstract update methods.

Method Summary

Modifier and Type	Method and Description
protected void	checkLabelShapes(NDArray labels, NDArray predictions) Checks the length of NDArrays.
protected void	checkLabelShapes(NDArray labels, NDArray predictions, boolean checkDimOnly) Checks if the two input NDArray have the same length or shape.
TrainingMetric	duplicate() Creates and returns a copy of this object.
java.lang.String	getName() Gets the name of this TrainingMetric.
abstract float	getValue() Calculates metric values.
abstract void	reset() Resets metric values.
abstract void	update(NDList labels, NDList predictions) Computes and updates the training metrics based on the labels and predictions.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

TrainingMetric

public TrainingMetric(java.lang.String name)

Creates a metric with abstract update methods.

Parameters:

name - the name of the metric

Method Detail

duplicate

public TrainingMetric duplicate()

Creates and returns a copy of this object.

Returns:

a copy of this TrainingMetric

update

public abstract void update(NDList labels,
                            NDList predictions)

Computes and updates the training metrics based on the labels and predictions.

Parameters:

labels - a NDList of labels

predictions - a NDList of predictions

reset

public abstract void reset()

Resets metric values.

getName

public java.lang.String getName()

Gets the name of this TrainingMetric.

Returns:

the name of this TrainingMetric

getValue

public abstract float getValue()

Calculates metric values.

Returns:

a Pair of metric name and value

checkLabelShapes

protected void checkLabelShapes(NDArray labels,
                                NDArray predictions,
                                boolean checkDimOnly)

Checks if the two input NDArray have the same length or shape.

Parameters:

labels - a NDArray of labels

predictions - a NDArray of predictions

checkDimOnly - whether to check for first dimension only

checkLabelShapes

protected void checkLabelShapes(NDArray labels,
                                NDArray predictions)

Checks the length of NDArrays.

Parameters:

labels - a NDArray of labels

predictions - a NDArray of predictions