org.deeplearning4j.models.glove

Class GloveWeightLookupTable

java.lang.Object

org.deeplearning4j.models.embeddings.inmemory.InMemoryLookupTable

org.deeplearning4j.models.glove.GloveWeightLookupTable

All Implemented Interfaces:

Serializable, WeightLookupTable

public class GloveWeightLookupTable
extends InMemoryLookupTable

Glove lookup table

Author:

Adam Gibson

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	GloveWeightLookupTable.Builder

Nested classes/interfaces inherited from class org.deeplearning4j.models.embeddings.inmemory.InMemoryLookupTable

InMemoryLookupTable.WeightIterator

Field Summary

Fields inherited from class org.deeplearning4j.models.embeddings.inmemory.InMemoryLookupTable

codes, expTable, lr, MAX_EXP, negative, rng, seed, syn0, syn1, syn1Neg, table, useAdaGrad, vectorLength, vocab

Constructor Summary

Constructors

Constructor and Description
GloveWeightLookupTable(VocabCache vocab, int vectorLength, boolean useAdaGrad, double lr, org.nd4j.linalg.api.rng.Random gen, double negative, double xMax, double maxCount)

Method Summary

Modifier and Type	Method and Description
org.nd4j.linalg.api.ndarray.INDArray	getBias()
org.nd4j.linalg.learning.AdaGrad	getBiasAdaGrad()
double	getMaxCount()
org.nd4j.linalg.learning.AdaGrad	getWeightAdaGrad()
double	getxMax()
void	iterateSample(VocabWord w1, VocabWord w2, AtomicLong nextRandom, double alpha) Iterate on the given 2 vocab words
double	iterateSample(VocabWord w1, VocabWord w2, double score) glove iteration
static GloveWeightLookupTable	load(InputStream is, VocabCache vocab) Load a glove model from an input stream.
void	resetWeights() Reset the weights of the cache
void	resetWeights(boolean reset) Clear out all weights regardless
void	setBias(org.nd4j.linalg.api.ndarray.INDArray bias)
void	setMaxCount(double maxCount)
void	setxMax(double xMax)

Methods inherited from class org.deeplearning4j.models.embeddings.inmemory.InMemoryLookupTable

getCodes, getExpTable, getLr, getNegative, getSyn0, getSyn1, getSyn1Neg, getTable, getVectorLength, getVocab, initExpTable, initNegative, isUseAdaGrad, iterate, layerSize, loadCodes, makeTable, plotVocab, plotVocab, putCode, putVector, setCodes, setExpTable, setLearningRate, setLr, setNegative, setSyn0, setSyn1, setSyn1Neg, setTable, setUseAdaGrad, setVectorLength, setVocab, toString, vector, vectors

Methods inherited from class java.lang.Object

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

Constructor Detail

GloveWeightLookupTable

public GloveWeightLookupTable(VocabCache vocab,
                              int vectorLength,
                              boolean useAdaGrad,
                              double lr,
                              org.nd4j.linalg.api.rng.Random gen,
                              double negative,
                              double xMax,
                              double maxCount)

Method Detail

resetWeights

public void resetWeights(boolean reset)

Description copied from interface: WeightLookupTable

Clear out all weights regardless

Specified by:

resetWeights in interface WeightLookupTable

Overrides:

resetWeights in class InMemoryLookupTable

resetWeights

public void resetWeights()

Reset the weights of the cache

Specified by:

resetWeights in interface WeightLookupTable

Overrides:

resetWeights in class InMemoryLookupTable

iterateSample

public double iterateSample(VocabWord w1,
                            VocabWord w2,
                            double score)

glove iteration

Parameters:

w1 - the first word

w2 - the second word

score - the weight learned for the particular co occurrences

getWeightAdaGrad

public org.nd4j.linalg.learning.AdaGrad getWeightAdaGrad()

getBiasAdaGrad

public org.nd4j.linalg.learning.AdaGrad getBiasAdaGrad()

load

public static GloveWeightLookupTable load(InputStream is,
                                          VocabCache vocab)
                                   throws IOException

Load a glove model from an input stream. The format is: word num1 num2....

Parameters:

is - the input stream to read from for the weights

vocab - the vocab for the lookuptable

Returns:

the loaded model

Throws:

IOException - if one occurs

iterateSample

public void iterateSample(VocabWord w1,
                          VocabWord w2,
                          AtomicLong nextRandom,
                          double alpha)

Description copied from class: InMemoryLookupTable

Iterate on the given 2 vocab words

Specified by:

iterateSample in interface WeightLookupTable

Overrides:

iterateSample in class InMemoryLookupTable

Parameters:

w1 - the first word to iterate on

w2 - the second word to iterate on

nextRandom - next random for sampling

alpha - the alpha to use for learning

getxMax

public double getxMax()

setxMax

public void setxMax(double xMax)

getMaxCount

public double getMaxCount()

setMaxCount

public void setMaxCount(double maxCount)

getBias

public org.nd4j.linalg.api.ndarray.INDArray getBias()

setBias

public void setBias(org.nd4j.linalg.api.ndarray.INDArray bias)