org.deeplearning4j.spark.models.word2vec

Class SparkWord2Vec.Builder

java.lang.Object

org.deeplearning4j.spark.models.sequencevectors.SparkSequenceVectors.Builder<VocabWord>

org.deeplearning4j.spark.models.word2vec.SparkWord2Vec.Builder

Enclosing class:

SparkWord2Vec

public static class SparkWord2Vec.Builder
extends SparkSequenceVectors.Builder<VocabWord>

Field Summary

Fields inherited from class org.deeplearning4j.spark.models.sequencevectors.SparkSequenceVectors.Builder

configuration, modelExporter, peersConfiguration, storageLevel, workers

Constructor Summary

Constructors

Constructor and Description
Builder() Deprecated.
Builder(org.nd4j.parameterserver.distributed.conf.VoidConfiguration psConfiguration)
Builder(org.nd4j.parameterserver.distributed.conf.VoidConfiguration psConfiguration, VectorsConfiguration configuration)

Method Summary

Modifier and Type	Method and Description
SparkWord2Vec	build() This method returns you SparkWord2Vec instance ready for training
SparkWord2Vec.Builder	epochs(int numEpochs)
SparkWord2Vec.Builder	iterations(int num)
SparkWord2Vec.Builder	layerSize(int layerSize)
SparkWord2Vec.Builder	minWordFrequency(int num)
SparkWord2Vec.Builder	negativeSampling(long samples)
SparkWord2Vec.Builder	setElementsLearningAlgorithm(SparkElementsLearningAlgorithm ela)
SparkWord2Vec.Builder	setLearningAlgorithm(SparkElementsLearningAlgorithm ela) This method defines the learning algorithm that will be used during training
SparkWord2Vec.Builder	setLearningRate(double lr)
SparkWord2Vec.Builder	setModelExporter(SparkModelExporter<VocabWord> exporter) This method defines the way model will be exported after training is finished
SparkWord2Vec.Builder	setParameterServerConfiguration(org.nd4j.parameterserver.distributed.conf.VoidConfiguration configuration)
SparkWord2Vec.Builder	setSequenceLearningAlgorithm(SparkSequenceLearningAlgorithm sla)
SparkWord2Vec.Builder	setStorageLevel(org.apache.spark.storage.StorageLevel level)
SparkWord2Vec.Builder	setTokenizerFactory(TokenizerFactory tokenizerFactory) This method defines tokenizer htat will be used for corpus tokenization
SparkWord2Vec.Builder	subsampling(double rate)
SparkWord2Vec.Builder	useHierarchicSoftmax(boolean reallyUse)
SparkWord2Vec.Builder	workers(int numWorkers)

Methods inherited from class java.lang.Object

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

Constructor Detail

Builder

@Deprecated
public Builder()

Deprecated.

This method should NOT be used in real world applications

Builder

public Builder(@NonNull
               org.nd4j.parameterserver.distributed.conf.VoidConfiguration psConfiguration)

Builder

public Builder(@NonNull
               org.nd4j.parameterserver.distributed.conf.VoidConfiguration psConfiguration,
               @NonNull
               VectorsConfiguration configuration)

Method Detail

setTokenizerFactory

public SparkWord2Vec.Builder setTokenizerFactory(@NonNull
                                                 TokenizerFactory tokenizerFactory)

This method defines tokenizer htat will be used for corpus tokenization

Parameters:

tokenizerFactory -

Returns:

setLearningAlgorithm

public SparkWord2Vec.Builder setLearningAlgorithm(@NonNull
                                                  SparkElementsLearningAlgorithm ela)

This method defines the learning algorithm that will be used during training

Parameters:

ela -

Returns:

setModelExporter

public SparkWord2Vec.Builder setModelExporter(@NonNull
                                              SparkModelExporter<VocabWord> exporter)

This method defines the way model will be exported after training is finished

Overrides:

setModelExporter in class SparkSequenceVectors.Builder<VocabWord>

Parameters:

exporter -

Returns:

workers

public SparkWord2Vec.Builder workers(int numWorkers)

Overrides:

workers in class SparkSequenceVectors.Builder<VocabWord>

Parameters:

numWorkers -

Returns:

epochs

public SparkWord2Vec.Builder epochs(int numEpochs)

Overrides:

epochs in class SparkSequenceVectors.Builder<VocabWord>

Returns:

setStorageLevel

public SparkWord2Vec.Builder setStorageLevel(org.apache.spark.storage.StorageLevel level)

Overrides:

setStorageLevel in class SparkSequenceVectors.Builder<VocabWord>

Returns:

minWordFrequency

public SparkWord2Vec.Builder minWordFrequency(int num)

Overrides:

minWordFrequency in class SparkSequenceVectors.Builder<VocabWord>

Returns:

setLearningRate

public SparkWord2Vec.Builder setLearningRate(double lr)

Overrides:

setLearningRate in class SparkSequenceVectors.Builder<VocabWord>

Returns:

setParameterServerConfiguration

public SparkWord2Vec.Builder setParameterServerConfiguration(@NonNull
                                                             org.nd4j.parameterserver.distributed.conf.VoidConfiguration configuration)

Overrides:

setParameterServerConfiguration in class SparkSequenceVectors.Builder<VocabWord>

Returns:

iterations

public SparkWord2Vec.Builder iterations(int num)

Overrides:

iterations in class SparkSequenceVectors.Builder<VocabWord>

Returns:

subsampling

public SparkWord2Vec.Builder subsampling(double rate)

Overrides:

subsampling in class SparkSequenceVectors.Builder<VocabWord>

Returns:

negativeSampling

public SparkWord2Vec.Builder negativeSampling(long samples)

Overrides:

negativeSampling in class SparkSequenceVectors.Builder<VocabWord>

Returns:

setElementsLearningAlgorithm

public SparkWord2Vec.Builder setElementsLearningAlgorithm(@NonNull
                                                          SparkElementsLearningAlgorithm ela)

Overrides:

setElementsLearningAlgorithm in class SparkSequenceVectors.Builder<VocabWord>

Returns:

setSequenceLearningAlgorithm

public SparkWord2Vec.Builder setSequenceLearningAlgorithm(@NonNull
                                                          SparkSequenceLearningAlgorithm sla)

Overrides:

setSequenceLearningAlgorithm in class SparkSequenceVectors.Builder<VocabWord>

Returns:

useHierarchicSoftmax

public SparkWord2Vec.Builder useHierarchicSoftmax(boolean reallyUse)

Overrides:

useHierarchicSoftmax in class SparkSequenceVectors.Builder<VocabWord>

Returns:

layerSize

public SparkWord2Vec.Builder layerSize(int layerSize)

Overrides:

layerSize in class SparkSequenceVectors.Builder<VocabWord>

build

public SparkWord2Vec build()

This method returns you SparkWord2Vec instance ready for training

Overrides:

build in class SparkSequenceVectors.Builder<VocabWord>

Returns: