org.deeplearning4j.nn.conf

Class MultiLayerConfiguration.Builder

java.lang.Object

org.deeplearning4j.nn.conf.MultiLayerConfiguration.Builder

Direct Known Subclasses:

NeuralNetConfiguration.ListBuilder

Enclosing class:

MultiLayerConfiguration

public static class MultiLayerConfiguration.Builder
extends Object

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	backprop Deprecated.
protected BackpropType	backpropType
protected CacheMode	cacheMode
protected List<NeuralNetConfiguration>	confs
protected double	dampingFactor
protected WorkspaceMode	inferenceWorkspaceMode
protected Map<Integer,InputPreProcessor>	inputPreProcessors
protected InputType	inputType
protected boolean	legacyBatchScaledL2
protected boolean	pretrain Deprecated.
protected int	tbpttBackLength
protected int	tbpttFwdLength
protected WorkspaceMode	trainingWorkspaceMode
protected boolean	validateOutputConfig

Constructor Summary

Constructors

Constructor and Description
Builder()

Method Summary

Modifier and Type	Method and Description
MultiLayerConfiguration.Builder	backprop(boolean backprop) Deprecated. doesn't affect training any more. Use MultiLayerNetwork.fit(DataSetIterator) when training for backprop.
MultiLayerConfiguration.Builder	backpropType(BackpropType type) The type of backprop.
MultiLayerConfiguration	build()
MultiLayerConfiguration.Builder	cacheMode(CacheMode cacheMode) This method defines how/if preOutput cache is handled: NONE: cache disabled (default value) HOST: Host memory will be used DEVICE: GPU memory will be used (on CPU backends effect will be the same as for HOST)
MultiLayerConfiguration.Builder	confs(List<NeuralNetConfiguration> confs)
MultiLayerConfiguration.Builder	inferenceWorkspaceMode(WorkspaceMode workspaceMode) Deprecated. Use NeuralNetConfiguration.Builder.inferenceWorkspaceMode(WorkspaceMode)
MultiLayerConfiguration.Builder	inputPreProcessor(Integer layer, InputPreProcessor processor) Specify the processors.
MultiLayerConfiguration.Builder	inputPreProcessors(Map<Integer,InputPreProcessor> processors)
MultiLayerConfiguration.Builder	legacyBatchScaledL2(boolean legacyBatchScaledL2)
MultiLayerConfiguration.Builder	pretrain(boolean pretrain) Deprecated. doesn't affect training any more. Use MultiLayerNetwork.pretrain(DataSetIterator) when training for layerwise pretraining.
MultiLayerConfiguration.Builder	setInputType(InputType inputType)
MultiLayerConfiguration.Builder	tBPTTBackwardLength(int backwardLength) When doing truncated BPTT: how many steps of backward should we do? Only applicable when doing backpropType(BackpropType.TruncatedBPTT) This is the k2 parameter on pg23 of http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf
MultiLayerConfiguration.Builder	tBPTTForwardLength(int forwardLength) When doing truncated BPTT: how many steps of forward pass should we do before doing (truncated) backprop? Only applicable when doing backpropType(BackpropType.TruncatedBPTT) Typically tBPTTForwardLength parameter is same as the tBPTTBackwardLength parameter, but may be larger than it in some circumstances (but never smaller) Ideally your training data time series length should be divisible by this This is the k1 parameter on pg23 of http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf
MultiLayerConfiguration.Builder	tBPTTLength(int bpttLength) When doing truncated BPTT: how many steps should we do? Only applicable when doing backpropType(BackpropType.TruncatedBPTT) See: http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf
MultiLayerConfiguration.Builder	trainingWorkspaceMode(WorkspaceMode workspaceMode) Deprecated. Use NeuralNetConfiguration.Builder.trainingWorkspaceMode(WorkspaceMode)
MultiLayerConfiguration.Builder	validateOutputLayerConfig(boolean validate) Enabled by default.

Methods inherited from class java.lang.Object

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

Field Detail

confs

protected List<NeuralNetConfiguration> confs

dampingFactor

protected double dampingFactor

inputPreProcessors

protected Map<Integer,InputPreProcessor> inputPreProcessors

pretrain

@Deprecated
protected boolean pretrain

Deprecated.

backprop

@Deprecated
protected boolean backprop

Deprecated.

backpropType

protected BackpropType backpropType

tbpttFwdLength

protected int tbpttFwdLength

tbpttBackLength

protected int tbpttBackLength

inputType

protected InputType inputType

trainingWorkspaceMode

protected WorkspaceMode trainingWorkspaceMode

inferenceWorkspaceMode

protected WorkspaceMode inferenceWorkspaceMode

cacheMode

protected CacheMode cacheMode

validateOutputConfig

protected boolean validateOutputConfig

legacyBatchScaledL2

protected boolean legacyBatchScaledL2

Constructor Detail

Builder

public Builder()

Method Detail

inputPreProcessor

public MultiLayerConfiguration.Builder inputPreProcessor(Integer layer,
                                                         InputPreProcessor processor)

Specify the processors. These are used at each layer for doing things like normalization and shaping of input.

Parameters:

processor - what to use to preProcess the data.

Returns:

builder pattern

inputPreProcessors

public MultiLayerConfiguration.Builder inputPreProcessors(Map<Integer,InputPreProcessor> processors)

backprop

@Deprecated
public MultiLayerConfiguration.Builder backprop(boolean backprop)

Deprecated. doesn't affect training any more. Use MultiLayerNetwork.fit(DataSetIterator) when training for backprop.

Whether to do back prop or not

Parameters:

backprop - whether to do back prop or not

Returns:

trainingWorkspaceMode

@Deprecated
public MultiLayerConfiguration.Builder trainingWorkspaceMode(@NonNull
                                                                         WorkspaceMode workspaceMode)

Deprecated. Use NeuralNetConfiguration.Builder.trainingWorkspaceMode(WorkspaceMode)

inferenceWorkspaceMode

@Deprecated
public MultiLayerConfiguration.Builder inferenceWorkspaceMode(@NonNull
                                                                          WorkspaceMode workspaceMode)

Deprecated. Use NeuralNetConfiguration.Builder.inferenceWorkspaceMode(WorkspaceMode)

cacheMode

public MultiLayerConfiguration.Builder cacheMode(@NonNull
                                                 CacheMode cacheMode)

This method defines how/if preOutput cache is handled: NONE: cache disabled (default value) HOST: Host memory will be used DEVICE: GPU memory will be used (on CPU backends effect will be the same as for HOST)

Parameters:

cacheMode -

Returns:

backpropType

public MultiLayerConfiguration.Builder backpropType(@NonNull
                                                    BackpropType type)

The type of backprop. Default setting is used for most networks (MLP, CNN etc), but optionally truncated BPTT can be used for training recurrent neural networks. If using TruncatedBPTT make sure you set both tBPTTForwardLength() and tBPTTBackwardLength()

tBPTTLength

public MultiLayerConfiguration.Builder tBPTTLength(int bpttLength)

When doing truncated BPTT: how many steps should we do?
Only applicable when doing backpropType(BackpropType.TruncatedBPTT)
See: http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf

Parameters:

bpttLength - length > 0

tBPTTForwardLength

public MultiLayerConfiguration.Builder tBPTTForwardLength(int forwardLength)

When doing truncated BPTT: how many steps of forward pass should we do before doing (truncated) backprop?
Only applicable when doing backpropType(BackpropType.TruncatedBPTT)
Typically tBPTTForwardLength parameter is same as the tBPTTBackwardLength parameter, but may be larger than it in some circumstances (but never smaller)
Ideally your training data time series length should be divisible by this This is the k1 parameter on pg23 of http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf

Parameters:

forwardLength - Forward length > 0, >= backwardLength

tBPTTBackwardLength

public MultiLayerConfiguration.Builder tBPTTBackwardLength(int backwardLength)

When doing truncated BPTT: how many steps of backward should we do?
Only applicable when doing backpropType(BackpropType.TruncatedBPTT)
This is the k2 parameter on pg23 of http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf

Parameters:

backwardLength - <= forwardLength

pretrain

@Deprecated
public MultiLayerConfiguration.Builder pretrain(boolean pretrain)

Deprecated. doesn't affect training any more. Use MultiLayerNetwork.pretrain(DataSetIterator) when training for layerwise pretraining.

Whether to do pre train or not

Parameters:

pretrain - whether to do pre train or not

Returns:

builder pattern

confs

public MultiLayerConfiguration.Builder confs(List<NeuralNetConfiguration> confs)

setInputType

public MultiLayerConfiguration.Builder setInputType(InputType inputType)

validateOutputLayerConfig

public MultiLayerConfiguration.Builder validateOutputLayerConfig(boolean validate)

Enabled by default. If enabled, the output layer configuration will be validated, to throw an exception on likely invalid outputs - such as softmax + nOut=1, or LossMCXENT + Tanh.
If disabled (false) no output layer validation will be performed.
Disabling this validation is not recommended, as the configurations that fail validation usually will not be able to learn correctly. However, the option to disable this validation is provided for advanced users when creating non-standard architectures.

Parameters:

validate - If true: validate output layer configuration. False: don't validate

legacyBatchScaledL2

public MultiLayerConfiguration.Builder legacyBatchScaledL2(boolean legacyBatchScaledL2)

build

public MultiLayerConfiguration build()