Package org.deeplearning4j.models.embeddings.wordvectors

Class WordVectorsImpl<T extends SequenceElement>

java.lang.Object

org.deeplearning4j.models.embeddings.wordvectors.WordVectorsImpl<T>

All Implemented Interfaces:

Serializable, WordVectors, org.deeplearning4j.nn.weights.embeddings.EmbeddingInitializer

Direct Known Subclasses:

SequenceVectors

public class WordVectorsImpl<T extends SequenceElement>
extends Object
implements WordVectors

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected int	batchSize
static String	DEFAULT_UNK
protected int	layerSize
protected org.nd4j.shade.guava.util.concurrent.AtomicDouble	learningRate
protected int	learningRateDecayWords
protected WeightLookupTable<T>	lookupTable
protected double	minLearningRate
protected int	minWordFrequency
protected ModelUtils<T>	modelUtils
protected double	negative
protected int	numEpochs
protected int	numIterations
protected boolean	resetModel
protected double	sampling
protected long	seed
protected Collection<String>	stopWords
protected boolean	trainElementsVectors
protected boolean	trainSequenceVectors
protected boolean	useAdeGrad
protected boolean	useUnknown
protected int[]	variableWindows
protected VocabCache<T>	vocab
protected int	window
protected int	workers

Constructor Summary

Constructors

Constructor	Description
WordVectorsImpl()

Method Summary

Modifier and Type	Method	Description
Map<String,Double>	accuracy(List<String> questions)	Accuracy based on questions which are a space separated list of strings where the first word is the query word, the next 2 words are negative, and the last word is the predicted word to be nearest
int	getLayerSize()	This method returns word vector size
double[]	getWordVector(String word)	Get the word vector for a given matrix
org.nd4j.linalg.api.ndarray.INDArray	getWordVectorMatrix(String word)	Get the word vector for a given matrix
org.nd4j.linalg.api.ndarray.INDArray	getWordVectorMatrixNormalized(String word)	Returns the word vector divided by the norm2 of the array
org.nd4j.linalg.api.ndarray.INDArray	getWordVectors(@NonNull Collection<String> labels)	This method returns 2D array, where each row represents corresponding label
org.nd4j.linalg.api.ndarray.INDArray	getWordVectorsMean(Collection<String> labels)	This method returns mean vector, built from words/labels passed in
boolean	hasWord(String word)	Returns true if the model has this word in the vocab
int	indexOf(String word)
boolean	jsonSerializable()
void	loadWeightsInto(org.nd4j.linalg.api.ndarray.INDArray array)
WeightLookupTable	lookupTable()	Lookup table for the vectors
boolean	outOfVocabularySupported()	Does implementation vectorize words absent in vocabulary
void	setLookupTable(@NonNull WeightLookupTable lookupTable)
void	setModelUtils(@NonNull ModelUtils modelUtils)	Specifies ModelUtils to be used to access model
void	setVocab(VocabCache vocab)
double	similarity(String word, String word2)	Returns similarity of two elements, provided by ModelUtils
List<String>	similarWordsInVocabTo(String word, double accuracy)	Find all words with a similar characters in the vocab
protected void	update()
protected void	update(org.nd4j.linalg.heartbeat.reports.Environment env, org.nd4j.linalg.heartbeat.reports.Event event)
int	vectorSize()
VocabCache<T>	vocab()	Vocab for the vectors
long	vocabSize()
Collection<String>	wordsNearest(String word, int n)	Get the top n words most similar to the given word
Collection<String>	wordsNearest(Collection<String> positive, Collection<String> negative, int top)	Words nearest based on positive and negative words
Collection<String>	wordsNearest(org.nd4j.linalg.api.ndarray.INDArray words, int top)	Words nearest based on positive and negative words * @param top the top n words
Collection<String>	wordsNearestSum(String word, int n)	Get the top n words most similar to the given word
Collection<String>	wordsNearestSum(Collection<String> positive, Collection<String> negative, int top)	Words nearest based on positive and negative words
Collection<String>	wordsNearestSum(org.nd4j.linalg.api.ndarray.INDArray words, int top)	Words nearest based on positive and negative words * @param top the top n words

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.deeplearning4j.models.embeddings.wordvectors.WordVectors

getUNK, setUNK

Field Detail

minWordFrequency

protected int minWordFrequency

lookupTable

protected WeightLookupTable<T extends SequenceElement> lookupTable

vocab

protected VocabCache<T extends SequenceElement> vocab

layerSize

protected int layerSize

modelUtils

protected transient ModelUtils<T extends SequenceElement> modelUtils

numIterations

protected int numIterations

numEpochs

protected int numEpochs

negative

protected double negative

sampling

protected double sampling

learningRate

protected org.nd4j.shade.guava.util.concurrent.AtomicDouble learningRate

minLearningRate

protected double minLearningRate

window

protected int window

batchSize

protected int batchSize

learningRateDecayWords

protected int learningRateDecayWords

resetModel

protected boolean resetModel

useAdeGrad

protected boolean useAdeGrad

workers

protected int workers

trainSequenceVectors

protected boolean trainSequenceVectors

trainElementsVectors

protected boolean trainElementsVectors

seed

protected long seed

useUnknown

protected boolean useUnknown

variableWindows

protected int[] variableWindows

DEFAULT_UNK

public static final String DEFAULT_UNK

See Also:

Constant Field Values

stopWords

protected Collection<String> stopWords

Constructor Detail

WordVectorsImpl

public WordVectorsImpl()

Method Detail

getLayerSize

public int getLayerSize()

This method returns word vector size

Returns:

hasWord

public boolean hasWord(String word)

Returns true if the model has this word in the vocab

Specified by:

hasWord in interface WordVectors

Parameters:

word - the word to test for

Returns:

true if the model has the word in the vocab

wordsNearestSum

public Collection<String> wordsNearestSum(Collection<String> positive,
                                          Collection<String> negative,
                                          int top)

Words nearest based on positive and negative words

Specified by:

wordsNearestSum in interface WordVectors

Parameters:

positive - the positive words

negative - the negative words

top - the top n words

Returns:

the words nearest the mean of the words

wordsNearestSum

public Collection<String> wordsNearestSum(org.nd4j.linalg.api.ndarray.INDArray words,
                                          int top)

Words nearest based on positive and negative words * @param top the top n words

Specified by:

wordsNearestSum in interface WordVectors

Returns:

the words nearest the mean of the words

wordsNearest

public Collection<String> wordsNearest(org.nd4j.linalg.api.ndarray.INDArray words,
                                       int top)

Words nearest based on positive and negative words * @param top the top n words

Specified by:

wordsNearest in interface WordVectors

Returns:

the words nearest the mean of the words

wordsNearestSum

public Collection<String> wordsNearestSum(String word,
                                          int n)

Get the top n words most similar to the given word

Specified by:

wordsNearestSum in interface WordVectors

Parameters:

word - the word to compare

n - the n to get

Returns:

the top n words

accuracy

public Map<String,Double> accuracy(List<String> questions)

Accuracy based on questions which are a space separated list of strings where the first word is the query word, the next 2 words are negative, and the last word is the predicted word to be nearest

Specified by:

accuracy in interface WordVectors

Parameters:

questions - the questions to ask

Returns:

the accuracy based on these questions

indexOf

public int indexOf(String word)

Specified by:

indexOf in interface WordVectors

similarWordsInVocabTo

public List<String> similarWordsInVocabTo(String word,
                                          double accuracy)

Find all words with a similar characters in the vocab

Specified by:

similarWordsInVocabTo in interface WordVectors

Parameters:

word - the word to compare

accuracy - the accuracy: 0 to 1

Returns:

the list of words that are similar in the vocab

getWordVector

public double[] getWordVector(String word)

Get the word vector for a given matrix

Specified by:

getWordVector in interface WordVectors

Parameters:

word - the word to get the matrix for

Returns:

the ndarray for this word

getWordVectorMatrixNormalized

public org.nd4j.linalg.api.ndarray.INDArray getWordVectorMatrixNormalized(String word)

Returns the word vector divided by the norm2 of the array

Specified by:

getWordVectorMatrixNormalized in interface WordVectors

Parameters:

word - the word to get the matrix for

Returns:

the looked up matrix

getWordVectorMatrix

public org.nd4j.linalg.api.ndarray.INDArray getWordVectorMatrix(String word)

Description copied from interface: WordVectors

Get the word vector for a given matrix

Specified by:

getWordVectorMatrix in interface WordVectors

Parameters:

word - the word to get the matrix for

Returns:

the ndarray for this word

wordsNearest

public Collection<String> wordsNearest(Collection<String> positive,
                                       Collection<String> negative,
                                       int top)

Words nearest based on positive and negative words

Specified by:

wordsNearest in interface WordVectors

Parameters:

positive - the positive words

negative - the negative words

top - the top n words

Returns:

the words nearest the mean of the words

getWordVectors

public org.nd4j.linalg.api.ndarray.INDArray getWordVectors(@NonNull
                                                           @NonNull Collection<String> labels)

This method returns 2D array, where each row represents corresponding label

Specified by:

getWordVectors in interface WordVectors

Parameters:

labels -

Returns:

getWordVectorsMean

public org.nd4j.linalg.api.ndarray.INDArray getWordVectorsMean(Collection<String> labels)

This method returns mean vector, built from words/labels passed in

Specified by:

getWordVectorsMean in interface WordVectors

Parameters:

labels -

Returns:

wordsNearest

public Collection<String> wordsNearest(String word,
                                       int n)

Get the top n words most similar to the given word

Specified by:

wordsNearest in interface WordVectors

Parameters:

word - the word to compare

n - the n to get

Returns:

the top n words

similarity

public double similarity(String word,
                         String word2)

Returns similarity of two elements, provided by ModelUtils

Specified by:

similarity in interface WordVectors

Parameters:

word - the first word

word2 - the second word

Returns:

a normalized similarity (cosine similarity)

vocab

public VocabCache<T> vocab()

Description copied from interface: WordVectors

Vocab for the vectors

Specified by:

vocab in interface WordVectors

Returns:

lookupTable

public WeightLookupTable lookupTable()

Description copied from interface: WordVectors

Lookup table for the vectors

Specified by:

lookupTable in interface WordVectors

Returns:

setModelUtils

public void setModelUtils(@NonNull
                          @NonNull ModelUtils modelUtils)

Description copied from interface: WordVectors

Specifies ModelUtils to be used to access model

Specified by:

setModelUtils in interface WordVectors

setLookupTable

public void setLookupTable(@NonNull
                           @NonNull WeightLookupTable lookupTable)

setVocab

public void setVocab(VocabCache vocab)

update

protected void update()

update

protected void update(org.nd4j.linalg.heartbeat.reports.Environment env,
                      org.nd4j.linalg.heartbeat.reports.Event event)

loadWeightsInto

public void loadWeightsInto(org.nd4j.linalg.api.ndarray.INDArray array)

Specified by:

loadWeightsInto in interface org.deeplearning4j.nn.weights.embeddings.EmbeddingInitializer

vocabSize

public long vocabSize()

Specified by:

vocabSize in interface org.deeplearning4j.nn.weights.embeddings.EmbeddingInitializer

vectorSize

public int vectorSize()

Specified by:

vectorSize in interface org.deeplearning4j.nn.weights.embeddings.EmbeddingInitializer

jsonSerializable

public boolean jsonSerializable()

Specified by:

jsonSerializable in interface org.deeplearning4j.nn.weights.embeddings.EmbeddingInitializer

outOfVocabularySupported

public boolean outOfVocabularySupported()

Description copied from interface: WordVectors

Does implementation vectorize words absent in vocabulary

Specified by:

outOfVocabularySupported in interface WordVectors

Returns:

boolean