Class LinearSGDOptions

java.lang.Object

org.tribuo.classification.sgd.linear.LinearSGDOptions

All Implemented Interfaces:

com.oracle.labs.mlrg.olcut.config.Options, ClassificationOptions<LinearSGDTrainer>

public class LinearSGDOptions
extends Object
implements ClassificationOptions<LinearSGDTrainer>

CLI options for training a linear classifier.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	LinearSGDOptions.LossEnum	Available loss types.

Field Summary

Fields

Modifier and Type	Field	Description
int	sgdEpochs	Number of SGD epochs.
int	sgdLoggingInterval	Log the objective after n examples.
int	sgdMinibatchSize	Minibatch size.
LinearSGDOptions.LossEnum	sgdObjective	Loss function.
long	sgdSeed	Sets the random seed for the LinearSGDTrainer.
GradientOptimiserOptions	sgoOptions	The gradient descent optimiser options.

Fields inherited from interface com.oracle.labs.mlrg.olcut.config.Options

header

Constructor Summary

Constructors

Constructor	Description
LinearSGDOptions()

Method Summary

Modifier and Type	Method	Description
LabelObjective	getLoss()	Returns the loss function specified in the arguments.
LinearSGDTrainer	getTrainer()	Constructs the trainer based on the provided arguments.

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.oracle.labs.mlrg.olcut.config.Options

getOptionsDescription

Field Details

sgoOptions

public GradientOptimiserOptions sgoOptions

The gradient descent optimiser options.

sgdEpochs

@Option(longName="sgd-epochs",
        usage="Number of SGD epochs. Defaults to 5.")
public int sgdEpochs

Number of SGD epochs. Defaults to 5.

sgdObjective

@Option(longName="sgd-objective",
        usage="Loss function. Defaults to LOG.")
public LinearSGDOptions.LossEnum sgdObjective

Loss function. Defaults to LOG.

sgdLoggingInterval

@Option(longName="sgd-logging-interval",
        usage="Log the objective after <int> examples. Defaults to 100.")
public int sgdLoggingInterval

Log the objective after n examples. Defaults to 100.

sgdMinibatchSize

@Option(longName="sgd-minibatch-size",
        usage="Minibatch size. Defaults to 1.")
public int sgdMinibatchSize

Minibatch size. Defaults to 1.

sgdSeed

@Option(longName="sgd-seed",
        usage="Sets the random seed for the LinearSGDTrainer.")
public long sgdSeed

Sets the random seed for the LinearSGDTrainer.

Constructor Details

LinearSGDOptions

public LinearSGDOptions()

Method Details

getLoss

public LabelObjective getLoss()

Returns the loss function specified in the arguments.

Returns:

The loss function.

getTrainer

public LinearSGDTrainer getTrainer()

Description copied from interface: ClassificationOptions

Constructs the trainer based on the provided arguments.

Specified by:

getTrainer in interface ClassificationOptions<LinearSGDTrainer>

Returns:

The trainer.