PositionalNeuralModel

Instance Constructors

new PositionalNeuralModel(annotator: (BinarizedTree[L], IndexedSeq[W]) ⇒ BinarizedTree[IndexedSeq[L2]], constrainer: Factory[L, W], topology: RuleTopology[L], lexicon: Lexicon[L, W], refinedTopology: RuleTopology[L2], refinements: GrammarRefinements[L, L2], labelFeaturizer: RefinedFeaturizer[L, W, Feature], surfaceFeaturizer: Word2VecSurfaceFeaturizerIndexed[W], depFeaturizer: Word2VecDepFeaturizerIndexed[W], transforms: IndexedSeq[OutputTransform[Array[Int], DenseVector[Double]]], maybeSparseSurfaceFeaturizer: Option[IndexedSpanFeaturizer[L, L2, W]], depTransforms: Seq[OutputTransform[Array[Int], DenseVector[Double]]], decoupledTransforms: Seq[OutputTransform[Array[Int], DenseVector[Double]]])

Type Members

type ExpectedCounts = StandardExpectedCounts[Feature]

Definition Classes
Model → Model
type Inference = PositionalNeuralModel.Inference[L, L2, W]

Definition Classes
PositionalNeuralModel → ParserModel → Model
type Marginal = ParseMarginal[L, W]

Definition Classes
ParserModel → Model
type Scorer = GrammarAnchoring[L, W]

Definition Classes
ParserModel → Model

Value Members

final def !=(arg0: Any): Boolean

Definition Classes
AnyRef → Any
final def ##(): Int

Definition Classes
AnyRef → Any
final def ==(arg0: Any): Boolean

Definition Classes
AnyRef → Any
def accumulateCounts(inf: Inference, s: Scorer, d: TreeInstance[L, W], m: Marginal, accum: ExpectedCounts, scale: Double): Unit

Definition Classes
PositionalNeuralModel → Model
final def accumulateCounts(inf: Inference, d: TreeInstance[L, W], accum: ExpectedCounts, scale: Double): Unit

Definition Classes
Model
final def asInstanceOf[T0]: T0

Definition Classes
Any
def cacheFeatureWeights(weights: DenseVector[Double], suffix: String = ""): Unit

Caches the weights using the cache broker.
Caches the weights using the cache broker.

Definition Classes
Model
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
def cloneModelForEnsembling: PositionalNeuralModel[L, L2, W]
val constrainer: Factory[L, W]

Definition Classes
PositionalNeuralModel → ParserExtractable
val decoupledTransforms: Seq[OutputTransform[Array[Int], DenseVector[Double]]]
val depTransforms: Seq[OutputTransform[Array[Int], DenseVector[Double]]]
def emptyCounts: StandardExpectedCounts[Feature]

Definition Classes
Model → Model
final def eq(arg0: AnyRef): Boolean

Definition Classes
AnyRef
def equals(arg0: Any): Boolean

Definition Classes
AnyRef → Any
final def expectedCounts(inf: Inference, d: TreeInstance[L, W], scale: Double = 1.0): ExpectedCounts

Definition Classes
Model
def expectedCountsToObjective(ecounts: ExpectedCounts): (Double, DenseVector[Double])

Definition Classes
Model → Model
def extractParser(weights: DenseVector[Double], trainExs: Seq[TreeInstance[L, W]])(implicit deb: Debinarizer[L]): Parser[L, W]

When doing batch normalization, we need to normalize the test network
def extractParser(weights: DenseVector[Double])(implicit deb: Debinarizer[L]): Parser[L, W]

Definition Classes
ParserModel → ParserExtractable
def featureIndex: Index[Feature]

Models have features, and this defines the mapping from indices in the weight vector to features.
Models have features, and this defines the mapping from indices in the weight vector to features.

Definition Classes
PositionalNeuralModel → Model
def finalize(): Unit

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( classOf[java.lang.Throwable] )
final def getClass(): Class[_]

Definition Classes
AnyRef → Any
def hashCode(): Int

Definition Classes
AnyRef → Any
val index: SegmentedIndex[Feature, Index[Feature]]

Models have features, and this defines the mapping from indices in the weight vector to features.
def inferenceFromWeights(weights: DenseVector[Double], forTrain: Boolean): Inference
def inferenceFromWeights(weights: DenseVector[Double]): Inference

Definition Classes
PositionalNeuralModel → Model
def initialValueForFeature(f: Feature): Double

Definition Classes
PositionalNeuralModel → Model
def initialWeightVector(initWeightsScale: Double, initializerSpec: String, trulyRandom: Boolean = false): DenseVector[Double]
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
val lexicon: Lexicon[L, W]

Definition Classes
PositionalNeuralModel → ParserExtractable
def logger: Logger

Definition Classes
SafeLogging
val maybeSparseSurfaceFeaturizer: Option[IndexedSpanFeaturizer[L, L2, W]]
def mergeWeightsForEnsembling(x1: DenseVector[Double], x2: DenseVector[Double]): DenseVector[Double]
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
def numFeatures: Int

Definition Classes
Model
def readCachedFeatureWeights(suffix: String = ""): Option[DenseVector[Double]]

just saves feature weights to disk as a serialized counter.
just saves feature weights to disk as a serialized counter. The file is prefix.ser.gz

Definition Classes
Model
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
def toString(): String

Definition Classes
AnyRef → Any
val topology: RuleTopology[L]

Definition Classes
PositionalNeuralModel → ParserExtractable
val transforms: IndexedSeq[OutputTransform[Array[Int], DenseVector[Double]]]
final def wait(): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long, arg1: Int): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
def weightsCacheName: String

Attributes
protected
Definition Classes
Model

Related Docs: object PositionalNeuralModel | package models

class PositionalNeuralModel[L, L2, W] extends ParserModel[L, W] with Serializable

Instance Constructors

Type Members

type ExpectedCounts = StandardExpectedCounts[Feature]

type Inference = PositionalNeuralModel.Inference[L, L2, W]

type Marginal = ParseMarginal[L, W]

type Scorer = GrammarAnchoring[L, W]

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

def accumulateCounts(inf: Inference, s: Scorer, d: TreeInstance[L, W], m: Marginal, accum: ExpectedCounts, scale: Double): Unit

final def accumulateCounts(inf: Inference, d: TreeInstance[L, W], accum: ExpectedCounts, scale: Double): Unit

final def asInstanceOf[T0]: T0

def cacheFeatureWeights(weights: DenseVector[Double], suffix: String = ""): Unit

def clone(): AnyRef

def cloneModelForEnsembling: PositionalNeuralModel[L, L2, W]

val constrainer: Factory[L, W]

val decoupledTransforms: Seq[OutputTransform[Array[Int], DenseVector[Double]]]

val depTransforms: Seq[OutputTransform[Array[Int], DenseVector[Double]]]

def emptyCounts: StandardExpectedCounts[Feature]

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

final def expectedCounts(inf: Inference, d: TreeInstance[L, W], scale: Double = 1.0): ExpectedCounts

def expectedCountsToObjective(ecounts: ExpectedCounts): (Double, DenseVector[Double])

def extractParser(weights: DenseVector[Double], trainExs: Seq[TreeInstance[L, W]])(implicit deb: Debinarizer[L]): Parser[L, W]

def extractParser(weights: DenseVector[Double])(implicit deb: Debinarizer[L]): Parser[L, W]

def featureIndex: Index[Feature]

def finalize(): Unit

final def getClass(): Class[_]

def hashCode(): Int

val index: SegmentedIndex[Feature, Index[Feature]]

def inferenceFromWeights(weights: DenseVector[Double], forTrain: Boolean): Inference

def inferenceFromWeights(weights: DenseVector[Double]): Inference

def initialValueForFeature(f: Feature): Double

def initialWeightVector(initWeightsScale: Double, initializerSpec: String, trulyRandom: Boolean = false): DenseVector[Double]

final def isInstanceOf[T0]: Boolean

val lexicon: Lexicon[L, W]

def logger: Logger

val maybeSparseSurfaceFeaturizer: Option[IndexedSpanFeaturizer[L, L2, W]]

def mergeWeightsForEnsembling(x1: DenseVector[Double], x2: DenseVector[Double]): DenseVector[Double]

final def ne(arg0: AnyRef): Boolean

final def notify(): Unit

final def notifyAll(): Unit

def numFeatures: Int

def readCachedFeatureWeights(suffix: String = ""): Option[DenseVector[Double]]

final def synchronized[T0](arg0: ⇒ T0): T0

def toString(): String

val topology: RuleTopology[L]

val transforms: IndexedSeq[OutputTransform[Array[Int], DenseVector[Double]]]

final def wait(): Unit

final def wait(arg0: Long, arg1: Int): Unit

final def wait(arg0: Long): Unit

def weightsCacheName: String

Inherited from Serializable

Inherited from Serializable

Inherited from ParserModel[L, W]

Inherited from ParserExtractable[L, W]

Inherited from Model[TreeInstance[L, W]]

Inherited from Model[TreeInstance[L, W]]

Inherited from SafeLogging

Inherited from AnyRef

Inherited from Any

Ungrouped