hasBias(boolean)
config (DenseLayer, EmbeddingLayer, OutputLayer, RnnOutputLayer, CenterLossOutputLayer, ConvolutionLayer, Convolution1DLayer). EmbeddingLayer now defaults to no bias (Link)Average
and Max
modes in addition to Add/Subtract/Product (Link).updater(IUpdater)
method.layer(String, Layer, String...)
alias for .addLayer(String, Layer, String...)
.list().layer(<layer>)
can now be used for configs) (Link)MultiLayerNetwork.summary(InputType)
and ComputationGraph.summary(InputType...)
methods (shows layer and activation size information) (Link)java -jar deeplearning4j-ui-standalone-1.0.0-alpha.jar -p 9124 -r true -f c:/UIStorage.bin
).weightInit(Distribution)
convenience/overload (previously: required .weightInit(WeightInit.DISTRIBUTION).dist(Distribution)
) (Link)Supplier<ScoreCalculator>
for use with non-serializable score calculators (Link)play.crypto.secret
system property to manually set the Play application secret; is randomly generated by default (Link).fit(DataSetIterator)
and similar methods no longer perform layerwise pretraining followed by backprop - only backprop is performed in these methods. For pretraining, use pretrain(DataSetIterator)
and pretrain(MultiDataSetIterator)
methods (Link).learningRate(double)
, .momentum(double)
etc) all removed.updater(new Adam(lr))
instead of .updater(Updater.ADAM).learningRate(lr)
.biasUpdater(IUpdater)
method.updater(new Adam(ISchedule))
and similar.updater(Updater)
) has been deprecated; use .updater(IUpdater)
.regularization(boolean)
config removed; functionality is now always equivalent to .regularization(true)
.useDropConnect(boolean)
removed; use .weightNoise(new DropConnect(double))
instead.iterations(int)
method has been removed (was rarely used and confusing to users)org.deeplearning4j.util
) have been deprecated and/or moved to nd4j-common. Use same class names in nd4j-common org.nd4j.util
instead..activation(String)
has been removed; use .activation(Activation)
or .activation(IActivation)
insteadapplyConstraints(int iteration, int epoch)
methodisWeightParam(String)
and isBiasParam(String)
methodsnew Bidirectional(Bidirectional.Mode.ADD, new GravesLSTM.Builder()....build()))
insteadModel
and ModelConfiguration
have been permanently removed. Use KerasModelImport instead, which is now the only entry point for Keras model import.doDiff
in SameDiff).org.datavec.api.util
mainly) have been deprecated or removed; use equivalently named util clases in nd4j-common module (Link)List<List<Writable>>
for batches, not List<Writable>
. See also NDArrayRecordBatch​ParameterSpace<IDropout>
, DropoutSpace introduced (Link).updater(IUpdater)
or .updater(ParameterSpace<IUpdater>)
methods insteadActorCriticLoss
and correct use of randomnessQLearning
would fail (non-flat input, incomplete serialization, incorrect normalization)HistoryProcessor
with async algorithms and failures when preprocessing imagesIterationListener
NeuralNet.getNeuralNetworks()
, Policy.getNeuralNet()
, and convenience constructors for Policy
MultiLayerNetwork
, and Model, corresponding to ComputationGraph
.