org.deeplearning4j.optimize.listeners

Class ParamAndGradientIterationListener

java.lang.Object

org.deeplearning4j.optimize.listeners.ParamAndGradientIterationListener

All Implemented Interfaces:

java.io.Serializable, IterationListener

public class ParamAndGradientIterationListener
extends java.lang.Object
implements IterationListener

An iteration listener that provides details on parameters and gradients at each iteration during traning. Attempts to provide much of the same information as the UI histogram iteration listener, but in a text-based format (for example, when learning on a system accessed via SSH etc). i.e., is intended to aid network tuning and debugging
This iteration listener is set up to calculate mean, min, max, and mean absolute value of each type of parameter and gradient in the network at each iteration.
These

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
ParamAndGradientIterationListener() Default constructor for output to console only every iteration, tab delimited
ParamAndGradientIterationListener(int iterations, boolean printHeader, boolean printMean, boolean printMinMax, boolean printMeanAbsValue, boolean outputToConsole, boolean outputToFile, boolean outputToLogger, java.io.File file, java.lang.String delimiter) Full constructor with all options.

Method Summary

Modifier and Type	Method and Description
void	invoke() Change invoke to true
boolean	invoked() Get if listener invoked
void	iterationDone(Model model, int iteration) Event listener for each iteration

Methods inherited from class java.lang.Object

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

Constructor Detail

ParamAndGradientIterationListener

public ParamAndGradientIterationListener()

Default constructor for output to console only every iteration, tab delimited

ParamAndGradientIterationListener

public ParamAndGradientIterationListener(int iterations,
                                         boolean printHeader,
                                         boolean printMean,
                                         boolean printMinMax,
                                         boolean printMeanAbsValue,
                                         boolean outputToConsole,
                                         boolean outputToFile,
                                         boolean outputToLogger,
                                         java.io.File file,
                                         java.lang.String delimiter)

Full constructor with all options. Note also: ParamAndGradientIterationListener.builder() can be used instead of this constructor.

Parameters:

iterations - calculate and report values every 'iterations' iterations

printHeader - Whether to output a header row (i.e., names for each column)

printMean - Calculate and display the mean of parameters and gradients

printMinMax - Calculate and display the min/max of the parameters and gradients

printMeanAbsValue - Calculate and display the mean absolute value

outputToConsole - If true, display the values to the console (System.out.println())

outputToFile - If true, write the values to a file, one per line

outputToLogger - If true, log the values

file - File to write values to. May be null, not used if outputToFile == false

delimiter - delimiter (for example, "\t" or "," etc)

Method Detail

invoked

public boolean invoked()

Description copied from interface: IterationListener

Get if listener invoked

Specified by:

invoked in interface IterationListener

invoke

public void invoke()

Description copied from interface: IterationListener

Change invoke to true

Specified by:

invoke in interface IterationListener

iterationDone

public void iterationDone(Model model,
                          int iteration)

Description copied from interface: IterationListener

Event listener for each iteration

Specified by:

iterationDone in interface IterationListener

Parameters:

model - the model iterating

iteration - the iteration