Package org.deeplearning4j.optimize.listeners

Class SleepyTrainingListener

java.lang.Object

org.deeplearning4j.optimize.api.BaseTrainingListener

org.deeplearning4j.optimize.listeners.SleepyTrainingListener

All Implemented Interfaces:

Serializable, TrainingListener

public class SleepyTrainingListener
extends BaseTrainingListener
implements Serializable

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	SleepyTrainingListener.SleepMode
static class	SleepyTrainingListener.TimeMode

Field Summary

Fields

Modifier and Type	Field	Description
protected ThreadLocal<AtomicLong>	lastBP
protected ThreadLocal<AtomicLong>	lastEE
protected ThreadLocal<AtomicLong>	lastES
protected ThreadLocal<AtomicLong>	lastFF
protected ThreadLocal<AtomicLong>	lastIteration
protected SleepyTrainingListener.SleepMode	sleepMode
protected SleepyTrainingListener.TimeMode	timeMode
protected long	timerBP
protected long	timerEE
protected long	timerES
protected long	timerFF
protected long	timerIteration

Constructor Summary

Constructors

Constructor	Description
SleepyTrainingListener()

Method Summary

Modifier and Type	Method	Description
void	iterationDone(Model model, int iteration, int epoch)	Event listener for each iteration.
void	onBackwardPass(Model model)	Called once per iteration (backward pass) after gradients have been calculated, and updated Gradients are available via Model.gradient().
void	onEpochEnd(Model model)	Called once at the end of each epoch, when using methods such as MultiLayerNetwork.fit(DataSetIterator), ComputationGraph.fit(DataSetIterator) or ComputationGraph.fit(MultiDataSetIterator)
void	onEpochStart(Model model)	Called once at the start of each epoch, when using methods such as MultiLayerNetwork.fit(DataSetIterator), ComputationGraph.fit(DataSetIterator) or ComputationGraph.fit(MultiDataSetIterator)
void	onForwardPass(Model model, List<INDArray> activations)	Called once per iteration (forward pass) for activations (usually for a MultiLayerNetwork), only at training time
void	onForwardPass(Model model, Map<String,INDArray> activations)	Called once per iteration (forward pass) for activations (usually for a ComputationGraph), only at training time
void	onGradientCalculation(Model model)	Called once per iteration (backward pass) before the gradients are updated Gradients are available via Model.gradient().
protected void	sleep(long sleepTimeMs)
protected void	sleep(AtomicLong lastTime, long sleepTime)

Methods inherited from class java.lang.Object

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

Field Detail

lastEE

protected final transient ThreadLocal<AtomicLong> lastEE

lastES

protected final transient ThreadLocal<AtomicLong> lastES

lastFF

protected final transient ThreadLocal<AtomicLong> lastFF

lastBP

protected final transient ThreadLocal<AtomicLong> lastBP

lastIteration

protected final transient ThreadLocal<AtomicLong> lastIteration

timerEE

protected long timerEE

timerES

protected long timerES

timerFF

protected long timerFF

timerBP

protected long timerBP

timerIteration

protected long timerIteration

sleepMode

protected SleepyTrainingListener.SleepMode sleepMode

timeMode

protected SleepyTrainingListener.TimeMode timeMode

Constructor Detail

SleepyTrainingListener

public SleepyTrainingListener()

Method Detail

sleep

protected void sleep(long sleepTimeMs)

sleep

protected void sleep(AtomicLong lastTime,
                     long sleepTime)

onEpochStart

public void onEpochStart(Model model)

Description copied from interface: TrainingListener

Called once at the start of each epoch, when using methods such as MultiLayerNetwork.fit(DataSetIterator), ComputationGraph.fit(DataSetIterator) or ComputationGraph.fit(MultiDataSetIterator)

Specified by:

onEpochStart in interface TrainingListener

Overrides:

onEpochStart in class BaseTrainingListener

onEpochEnd

public void onEpochEnd(Model model)

Description copied from interface: TrainingListener

Called once at the end of each epoch, when using methods such as MultiLayerNetwork.fit(DataSetIterator), ComputationGraph.fit(DataSetIterator) or ComputationGraph.fit(MultiDataSetIterator)

Specified by:

onEpochEnd in interface TrainingListener

Overrides:

onEpochEnd in class BaseTrainingListener

onForwardPass

public void onForwardPass(Model model,
                          List<INDArray> activations)

Description copied from interface: TrainingListener

Called once per iteration (forward pass) for activations (usually for a MultiLayerNetwork), only at training time

Specified by:

onForwardPass in interface TrainingListener

Overrides:

onForwardPass in class BaseTrainingListener

Parameters:

model - Model

activations - Layer activations (including input)

onForwardPass

public void onForwardPass(Model model,
                          Map<String,INDArray> activations)

Description copied from interface: TrainingListener

Called once per iteration (forward pass) for activations (usually for a ComputationGraph), only at training time

Specified by:

onForwardPass in interface TrainingListener

Overrides:

onForwardPass in class BaseTrainingListener

Parameters:

model - Model

activations - Layer activations (including input)

iterationDone

public void iterationDone(Model model,
                          int iteration,
                          int epoch)

Description copied from interface: TrainingListener

Event listener for each iteration. Called once, after each parameter update has ocurred while training the network

Specified by:

iterationDone in interface TrainingListener

Overrides:

iterationDone in class BaseTrainingListener

Parameters:

model - the model iterating

iteration - the iteration

onBackwardPass

public void onBackwardPass(Model model)

Description copied from interface: TrainingListener

Called once per iteration (backward pass) after gradients have been calculated, and updated Gradients are available via Model.gradient().

Unlike TrainingListener.onGradientCalculation(Model) the gradients at this point will be post-update, rather than raw (pre-update) gradients at that method call.

Specified by:

onBackwardPass in interface TrainingListener

Overrides:

onBackwardPass in class BaseTrainingListener

Parameters:

model - Model

onGradientCalculation

public void onGradientCalculation(Model model)

Description copied from interface: TrainingListener

Called once per iteration (backward pass) before the gradients are updated Gradients are available via Model.gradient(). Note that gradients will likely be updated in-place - thus they should be copied or processed synchronously in this method.

For updates (gradients post learning rate/momentum/rmsprop etc) see TrainingListener.onBackwardPass(Model)

Specified by:

onGradientCalculation in interface TrainingListener

Overrides:

onGradientCalculation in class BaseTrainingListener

Parameters:

model - Model