org.deeplearning4j.nn.api

Interface TrainingConfig

All Known Implementing Classes:

AbstractLSTM, AbstractSameDiffLayer, ActivationLayer, AttentionVertex, AutoEncoder, BaseLayer, BaseOutputLayer, BasePretrainNetwork, BaseRecurrentLayer, BaseUpsamplingLayer, BaseWrapperLayer, BatchNormalization, Bidirectional, CapsuleLayer, CapsuleStrengthLayer, CenterLossOutputLayer, Cnn3DLossLayer, CnnLossLayer, Convolution1D, Convolution1DLayer, Convolution2D, Convolution3D, ConvolutionLayer, Cropping1D, Cropping2D, Cropping3D, Deconvolution2D, Deconvolution3D, DenseLayer, DepthwiseConvolution2D, DropoutLayer, DummyConfig, ElementWiseMultiplicationLayer, EmbeddingLayer, EmbeddingSequenceLayer, FeedForwardLayer, FrozenLayer, FrozenLayerWithBackprop, GlobalPoolingLayer, GravesBidirectionalLSTM, GravesLSTM, IdentityLayer, LastTimeStep, Layer, LearnedSelfAttentionLayer, LocallyConnected1D, LocallyConnected2D, LocalResponseNormalization, LossLayer, LSTM, MaskLayer, MaskZeroLayer, NoParamLayer, OCNNOutputLayer, OutputLayer, Pooling1D, Pooling2D, PReLULayer, PrimaryCapsules, RecurrentAttentionLayer, RepeatVector, RnnLossLayer, RnnOutputLayer, SameDiffLambdaLayer, SameDiffLambdaVertex, SameDiffLayer, SameDiffOutputLayer, SameDiffVertex, SelfAttentionLayer, SeparableConvolution2D, SimpleRnn, SpaceToBatchLayer, SpaceToDepthLayer, Subsampling1DLayer, Subsampling3DLayer, SubsamplingLayer, TimeDistributed, Upsampling1D, Upsampling2D, Upsampling3D, VariationalAutoencoder, Yolo2OutputLayer, ZeroPadding1DLayer, ZeroPadding3DLayer, ZeroPaddingLayer

public interface TrainingConfig

Simple interface for the training configuration (updater, L1/L2 values, etc) for trainable layers/vertices.

Author:

Alex Black

Method Summary

Modifier and Type	Method and Description
GradientNormalization	getGradientNormalization()
double	getGradientNormalizationThreshold()
String	getLayerName()
List<Regularization>	getRegularizationByParam(String paramName) Get the regularization types (l1/l2/weight decay) for the given parameter.
IUpdater	getUpdaterByParam(String paramName) Get the updater for the given parameter.
boolean	isPretrainParam(String paramName) Is the specified parameter a layerwise pretraining only parameter? For example, visible bias params in an autoencoder (or, decoder params in a variational autoencoder) aren't used during supervised backprop. Layers (like DenseLayer, etc) with no pretrainable parameters will return false for all (valid) inputs.
void	setDataType(DataType dataType)

Method Detail

getLayerName

String getLayerName()

Returns:

Name of the layer

getRegularizationByParam

List<Regularization> getRegularizationByParam(String paramName)

Get the regularization types (l1/l2/weight decay) for the given parameter. Different parameters may have different regularization types.

Parameters:

paramName - Parameter name ("W", "b" etc)

Returns:

Regularization types (if any) for the specified parameter

isPretrainParam

boolean isPretrainParam(String paramName)

Is the specified parameter a layerwise pretraining only parameter?
For example, visible bias params in an autoencoder (or, decoder params in a variational autoencoder) aren't used during supervised backprop.
Layers (like DenseLayer, etc) with no pretrainable parameters will return false for all (valid) inputs.

Parameters:

paramName - Parameter name/key

Returns:

True if the parameter is for layerwise pretraining only, false otherwise

getUpdaterByParam

IUpdater getUpdaterByParam(String paramName)

Get the updater for the given parameter. Typically the same updater will be used for all updaters, but this is not necessarily the case

Parameters:

paramName - Parameter name

Returns:

IUpdater for the parameter

getGradientNormalization

GradientNormalization getGradientNormalization()

Returns:

The gradient normalization configuration

getGradientNormalizationThreshold

double getGradientNormalizationThreshold()

Returns:

The gradient normalization threshold

setDataType

void setDataType(DataType dataType)