BeamSearchDecoder

Instance Constructors

new BeamSearchDecoder(cell: RNNCell[Output, core.Shape, S, SS], initialCellState: S, embeddingFn: (Output) ⇒ Output, beginTokens: Output, endToken: Output, beamWidth: Int, lengthPenalty: LengthPenalty = NoPenalty, outputLayer: (Output) ⇒ Output = (o: Output) => o, reorderTensorArrays: Boolean = true, name: String = "BeamSearchRNNDecoder")(implicit evOutput: Aux[Output, core.Shape], evS: Aux[S, SS])

cell
RNN cell to use for decoding.
initialCellState
Initial RNN cell state to use for starting the decoding process.
embeddingFn
Function that takes an INT32 vector of IDs and returns the corresponding embedded values that will be passed to the decoder input.
beginTokens
INT32 vector with length equal to the batch size, which contains the begin-of-sequence token IDs.
endToken
INT32 scalar containing the end-of-sequence token ID (i.e., token ID which marks the end of decoding).
beamWidth
Beam width to use for the beam search while decoding.
lengthPenalty
Length penalty method.
outputLayer
Output layer to use that is applied at the outputs of the provided RNN cell before returning them.
reorderTensorArrays
If true, TensorArrays' elements within the cell state will be reordered according to the beam search path. If the TensorArray can be reordered, the stacked form will be returned. Otherwise, the TensorArray will be returned as is. Set this flag to false if the cell state contains any TensorArrays that are not amenable to reordering.
name
Name prefix used for all created ops.

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
final def asInstanceOf[T0]: T0

Definition Classes
Any
val batchSize: Output

Scalar INT32 tensor representing the batch size of the input values.
Scalar INT32 tensor representing the batch size of the input values.

Definition Classes
BeamSearchDecoder → Decoder
val beamWidth: Int

Beam width to use for the beam search while decoding.
val beginTokens: Output

INT32 vector with length equal to the batch size, which contains the begin-of-sequence token IDs.
val cell: RNNCell[Output, core.Shape, S, SS]

RNN cell to use for decoding.
RNN cell to use for decoding.

Definition Classes
BeamSearchDecoder → Decoder
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
def decode(outputTimeMajor: Boolean = false, imputeFinished: Boolean = false, maximumIterations: Output = null, parallelIterations: Int = 32, swapMemory: Boolean = false, name: String = s"$name/DynamicRNNDecode"): (FinalOutput, State[S, SS], Output)

Performs dynamic decoding using this decoder.
Performs dynamic decoding using this decoder.
This method calls initialize() once and next() repeatedly.

Definition Classes
Decoder
val embeddingFn: (Output) ⇒ Output

Function that takes an INT32 vector of IDs and returns the corresponding embedded values that will be passed to the decoder input.
val endToken: Output

INT32 scalar containing the end-of-sequence token ID (i.e., token ID which marks the end of decoding).
final def eq(arg0: AnyRef): Boolean

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

Definition Classes
AnyRef → Any
def finalize(output: BeamSearchDecoder.Output, state: State[S, SS], sequenceLengths: Output): (FinalOutput, State[S, SS], Output)

Finalizes the output of the decoding process.
Finalizes the output of the decoding process.
output
Final output after decoding.
state
Final state after decoding.
sequenceLengths
Tensor containing the sequence lengths that the decoder cell outputs.
returns
Finalized output and state to return from the decoding process.

Definition Classes
BeamSearchDecoder → Decoder
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 initialCellState: S

Initial RNN cell state to use for starting the decoding process.
def initialize(): (Output, Output, State[S, SS])

This method is called before any decoding iterations.
This method is called before any decoding iterations. It computes the initial input values and the initial state.
returns
Tuple containing: (i) a scalar BOOLEAN tensor specifying whether initialization has finished, (ii) the next input, and (iii) the initial decoder state.

Definition Classes
BeamSearchDecoder → Decoder
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
val lengthPenalty: LengthPenalty

Length penalty method.
val name: String

Name prefix used for all created ops.
Name prefix used for all created ops.

Definition Classes
BeamSearchDecoder → Decoder
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
def next(time: Output, input: Output, state: State[S, SS]): (BeamSearchDecoder.Output, State[S, SS], Output, Output)

This method is called once per step of decoding (but only once for dynamic decoding).
This method is called once per step of decoding (but only once for dynamic decoding).
returns
Tuple containing: (i) the decoder output for this step, (ii) the next decoder state, (iii) the next input, and (iv) a scalar BOOLEAN tensor specifying whether decoding has finished.

Definition Classes
BeamSearchDecoder → Decoder
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
val outputLayer: (Output) ⇒ Output

Output layer to use that is applied at the outputs of the provided RNN cell before returning them.
val reorderTensorArrays: Boolean

If true, TensorArrays' elements within the cell state will be reordered according to the beam search path.
If true, TensorArrays' elements within the cell state will be reordered according to the beam search path. If the TensorArray can be reordered, the stacked form will be returned. Otherwise, the TensorArray will be returned as is. Set this flag to false if the cell state contains any TensorArrays that are not amenable to reordering.
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
def toString(): String

Definition Classes
AnyRef → Any
val tracksOwnFinished: Boolean

Describes whether the decoder keeps track of finished states.
Describes whether the decoder keeps track of finished states.
Most decoders will emit a true/false finished value independently at each time step. In this case, the dynamicDecode() function keeps track of which batch entries have already finished, and performs a logical OR to insert new batches to the finished set.
Some decoders, however, shuffle batches/beams between time steps and dynamicDecode() will mix up the finished state across these entries because it does not track the reshuffling across time steps. In this case, it is up to the decoder to declare that it will keep track of its own finished state by setting this property to true.
The beam-search decoder shuffles its beams and their finished state. For this reason, it conflicts with the dynamicDecode function's tracking of finished states. Setting this property to true avoids early stopping of decoding due to mismanagement of the finished state in dynamicDecode.

Definition Classes
BeamSearchDecoder → Decoder
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 zeroOutput(): BeamSearchDecoder.Output

Definition Classes
BeamSearchDecoder → Decoder

Inherited from Decoder[Output, core.Shape, S, SS, BeamSearchDecoder.Output, (core.Shape, core.Shape, core.Shape), State[S, SS], (SS, core.Shape, core.Shape, core.Shape), FinalOutput, State[S, SS]]

Related Docs: object BeamSearchDecoder | package decoders

class BeamSearchDecoder[S, SS] extends Decoder[Output, core.Shape, S, SS, decoders.BeamSearchDecoder.Output, (core.Shape, core.Shape, core.Shape), State[S, SS], (SS, core.Shape, core.Shape, core.Shape), FinalOutput, State[S, SS]]

Instance Constructors

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

final def asInstanceOf[T0]: T0

val batchSize: Output

val beamWidth: Int

val beginTokens: Output

val cell: RNNCell[Output, core.Shape, S, SS]

def clone(): AnyRef

def decode(outputTimeMajor: Boolean = false, imputeFinished: Boolean = false, maximumIterations: Output = null, parallelIterations: Int = 32, swapMemory: Boolean = false, name: String = s"$name/DynamicRNNDecode"): (FinalOutput, State[S, SS], Output)

val embeddingFn: (Output) ⇒ Output

val endToken: Output

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

def finalize(output: BeamSearchDecoder.Output, state: State[S, SS], sequenceLengths: Output): (FinalOutput, State[S, SS], Output)

def finalize(): Unit

final def getClass(): Class[_]

def hashCode(): Int

val initialCellState: S

def initialize(): (Output, Output, State[S, SS])

final def isInstanceOf[T0]: Boolean

val lengthPenalty: LengthPenalty

val name: String

final def ne(arg0: AnyRef): Boolean

def next(time: Output, input: Output, state: State[S, SS]): (BeamSearchDecoder.Output, State[S, SS], Output, Output)

final def notify(): Unit

final def notifyAll(): Unit

val outputLayer: (Output) ⇒ Output

val reorderTensorArrays: Boolean

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

def toString(): String

val tracksOwnFinished: Boolean

final def wait(): Unit

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

final def wait(arg0: Long): Unit

def zeroOutput(): BeamSearchDecoder.Output

Inherited from Decoder[Output, core.Shape, S, SS, BeamSearchDecoder.Output, (core.Shape, core.Shape, core.Shape), State[S, SS], (SS, core.Shape, core.Shape, core.Shape), FinalOutput, State[S, SS]]

Inherited from AnyRef

Inherited from Any

Ungrouped