edu.stanford.nlp.sentiment

Class RNNOptions

java.lang.Object

edu.stanford.nlp.sentiment.RNNOptions

All Implemented Interfaces:

Serializable

public class RNNOptions
extends Object
implements Serializable

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
boolean	combineClassification If this option is true, then the binary and unary classification matrices are combined.
TreebankLanguagePack	langpack
boolean	lowercaseWordVectors
int	numClasses Number of classes to build the RNN for
int	numHid Size of vectors to use.
int	randomSeed The random seed the random number generator is initialized with.
boolean	randomWordVectors By default, initialize random word vectors instead of reading from a file
boolean	simplifiedModel No symantic untying - use the same category for all categories.
RNNTrainOptions	trainOptions
String	unkWord In the wordVectors file, what word represents unknown?
boolean	useTensors
String	wordVectors Filename for the word vectors

Constructor Summary

Constructors

Constructor and Description
RNNOptions()

Method Summary

Methods

Modifier and Type	Method and Description
int	setOption(String[] args, int argIndex)
String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

randomSeed

public int randomSeed

The random seed the random number generator is initialized with.

wordVectors

public String wordVectors

Filename for the word vectors

unkWord

public String unkWord

In the wordVectors file, what word represents unknown?

randomWordVectors

public boolean randomWordVectors

By default, initialize random word vectors instead of reading from a file

numHid

public int numHid

Size of vectors to use. Must be at most the size of the vectors in the word vector file. If a smaller size is specified, vectors will be truncated.

numClasses

public int numClasses

Number of classes to build the RNN for

lowercaseWordVectors

public boolean lowercaseWordVectors

useTensors

public boolean useTensors

langpack

public TreebankLanguagePack langpack

simplifiedModel

public boolean simplifiedModel

No symantic untying - use the same category for all categories. This results in all nodes getting the same matrix (and tensor, where applicable)

combineClassification

public boolean combineClassification

If this option is true, then the binary and unary classification matrices are combined. Only makes sense if simplifiedModel is true. If combineClassification is set to true, simplifiedModel will also be set to true. If simplifiedModel is set to false, this will be set to false.

trainOptions

public RNNTrainOptions trainOptions

Constructor Detail

RNNOptions

public RNNOptions()

Method Detail

toString

public String toString()

Overrides:

toString in class Object

setOption

public int setOption(String[] args,
            int argIndex)