ai.djl.mxnet.engine

Class MxTrainer

java.lang.Object

ai.djl.mxnet.engine.MxTrainer

All Implemented Interfaces:

ai.djl.training.Trainer, java.lang.AutoCloseable

public class MxTrainer
extends java.lang.Object
implements ai.djl.training.Trainer

MxTrainer is the MXNet implementation of the Trainer.

Method Summary

Modifier and Type	Method and Description
void	close()
protected void	finalize()
ai.djl.ndarray.NDList	forward(ai.djl.ndarray.NDList input)
ai.djl.training.loss.Loss	getLoss()
ai.djl.ndarray.NDManager	getManager()
ai.djl.metric.Metrics	getMetrics() Gets the Metrics associated with this Trainer.
ai.djl.Model	getModel()
<T extends ai.djl.training.metrics.TrainingMetric> T	getTrainingMetric(java.lang.Class<T> clazz)
ai.djl.training.loss.Loss	getValidationLoss()
<T extends ai.djl.training.metrics.TrainingMetric> T	getValidationMetric(java.lang.Class<T> clazz)
void	initialize(ai.djl.ndarray.types.Shape... shapes)
ai.djl.training.GradientCollector	newGradientCollector()
void	resetTrainingMetrics()
void	setMetrics(ai.djl.metric.Metrics metrics)
void	setTrainingListener(ai.djl.training.TrainingListener listener)
void	step()
void	trainBatch(ai.djl.training.dataset.Batch batch)
void	validateBatch(ai.djl.training.dataset.Batch batch)

Methods inherited from class java.lang.Object

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

Methods inherited from interface ai.djl.training.Trainer

iterateDataset

Method Detail

initialize

public void initialize(ai.djl.ndarray.types.Shape... shapes)

Specified by:

initialize in interface ai.djl.training.Trainer

newGradientCollector

public ai.djl.training.GradientCollector newGradientCollector()

Specified by:

newGradientCollector in interface ai.djl.training.Trainer

trainBatch

public void trainBatch(ai.djl.training.dataset.Batch batch)

Specified by:

trainBatch in interface ai.djl.training.Trainer

forward

public ai.djl.ndarray.NDList forward(ai.djl.ndarray.NDList input)

Specified by:

forward in interface ai.djl.training.Trainer

validateBatch

public void validateBatch(ai.djl.training.dataset.Batch batch)

Specified by:

validateBatch in interface ai.djl.training.Trainer

step

public void step()

Specified by:

step in interface ai.djl.training.Trainer

setMetrics

public void setMetrics(ai.djl.metric.Metrics metrics)

Specified by:

setMetrics in interface ai.djl.training.Trainer

setTrainingListener

public void setTrainingListener(ai.djl.training.TrainingListener listener)

Specified by:

setTrainingListener in interface ai.djl.training.Trainer

resetTrainingMetrics

public void resetTrainingMetrics()

Specified by:

resetTrainingMetrics in interface ai.djl.training.Trainer

getLoss

public ai.djl.training.loss.Loss getLoss()

Specified by:

getLoss in interface ai.djl.training.Trainer

getValidationLoss

public ai.djl.training.loss.Loss getValidationLoss()

Specified by:

getValidationLoss in interface ai.djl.training.Trainer

getModel

public ai.djl.Model getModel()

Specified by:

getModel in interface ai.djl.training.Trainer

getMetrics

public ai.djl.metric.Metrics getMetrics()

Gets the Metrics associated with this Trainer.

Returns:

the Metrics associated with this Trainer

getTrainingMetric

public final <T extends ai.djl.training.metrics.TrainingMetric> T getTrainingMetric(java.lang.Class<T> clazz)

Specified by:

getTrainingMetric in interface ai.djl.training.Trainer

getValidationMetric

public <T extends ai.djl.training.metrics.TrainingMetric> T getValidationMetric(java.lang.Class<T> clazz)

Specified by:

getValidationMetric in interface ai.djl.training.Trainer

getManager

public ai.djl.ndarray.NDManager getManager()

Specified by:

getManager in interface ai.djl.training.Trainer

finalize

protected void finalize()
                 throws java.lang.Throwable

Overrides:

finalize in class java.lang.Object

Throws:

java.lang.Throwable

close

public void close()

Specified by:

close in interface ai.djl.training.Trainer

Specified by:

close in interface java.lang.AutoCloseable