Package org.nd4j.linalg.api.ops.impl.shape.tensorops

Class TensorArray

java.lang.Object

org.nd4j.autodiff.functions.DifferentialFunction

org.nd4j.linalg.api.ops.DynamicCustomOp

org.nd4j.linalg.api.ops.impl.shape.tensorops.BaseTensorOp

org.nd4j.linalg.api.ops.impl.shape.tensorops.TensorArray

All Implemented Interfaces:

CustomOp

public class TensorArray
extends BaseTensorOp

Nested Class Summary

Nested classes/interfaces inherited from class org.nd4j.linalg.api.ops.DynamicCustomOp

DynamicCustomOp.DynamicCustomOpsBuilder

Field Summary

Fields

Modifier and Type	Field	Description
protected boolean	clearOnRead
protected SDVariable	flow
protected DataType	tensorArrayDataType

Fields inherited from class org.nd4j.linalg.api.ops.DynamicCustomOp

axis, bArguments, dArguments, iArguments, inplaceCall, inputArguments, outputArguments, outputVariables, sArguments, tArguments

Fields inherited from class org.nd4j.autodiff.functions.DifferentialFunction

dimensions, extraArgs, inPlace, ownName, ownNameSetWithDefault, sameDiff, scalarValue

Constructor Summary

Constructors

Constructor	Description
TensorArray()
TensorArray(String name, SameDiff sameDiff, DataType dataType)
TensorArray(SameDiff sameDiff, DataType dataType)
TensorArray(DataType dataType)
TensorArray(TensorArray ta)
TensorArray(TensorArray ta, SDVariable[] inputs)

Method Summary

Modifier and Type	Method	Description
List<DataType>	calculateOutputDataTypes(List<DataType> inputDataType)	Calculate the data types for the output arrays.
SDVariable	concat(SDVariable flow)
void	configureFromArguments()	This allows a custom op to configure relevant fields from its arguments.
static SDVariable	createEmpty(SameDiff sd, DataType dataType)	Create an empty sequence with the specified data type.
static SDVariable	createEmpty(SameDiff sd, DataType dataType, String outputVarName)	Create an empty sequence with the specified data type.
static SDVariable	createTensorArrayFrom(SameDiff sd, SDVariable[] inputs)	Create an TensorArray op from the given inputs, note this is the same as calling createTensorArrayFrom(SameDiff, SDVariable[],String) with null.
static SDVariable	createTensorArrayFrom(SameDiff sd, SDVariable[] inputs, String outputVarName)	Create an TensorArray op from the given inputs
SDVariable	gather(SDVariable flow, int... indices)
SDVariable	gather(SDVariable flow, SDVariable indices)
int	getNumOutputs()
SameDiff	getSameDiff()
static TensorArray	getTensorArray(SameDiff sd, SDVariable sequenceVar)	Get the associated TensorArray instance related to this op.
SDVariable	getVar()
void	initFromTensorFlow(NodeDef nodeDef, SameDiff initWith, Map<String,AttrValue> attributesForNode, GraphDef graph)	Initialize the function from the given NodeDef
static SDVariable	itemAtIndex(SameDiff sd, SDVariable[] inputs)	Returns the item at the specified index in the specified list.
static SDVariable	itemAtIndex(SameDiff sd, SDVariable[] inputs, String outputVarName)	Returns the item at the specified index in the specified list.
String	opName()	This method returns op opName as string
Op.Type	opType()	The type of the op
SDVariable	read(int index)
SDVariable	read(SDVariable index)
SDVariable	read(SDVariable from, SDVariable index)
SDVariable	remove(SDVariable value)
SDVariable	remove(SDVariable value, int idx)
SDVariable	remove(SDVariable value, SDVariable idx)
static SDVariable	removeFromTensorArray(SameDiff sameDiff, SDVariable inputSequence)	Remove the last element from the relevant TensorArray
static SDVariable	removeFromTensorArray(SameDiff sameDiff, SDVariable inputSequence, SDVariable position)	Remove an element from the relevant TensorArray
static SDVariable	removeFromTensorArray(SameDiff sameDiff, SDVariable inputSequence, SDVariable position, String outputVarName)	Remove an element from the relevant TensorArray
long[]	requiredShape()	Returns the required shape for elements in this tensor array.
SDVariable	scatter(SDVariable flow, SDVariable value, int... indices)
SDVariable	scatter(SDVariable flow, SDVariable value, SDVariable indices)
SDVariable	size(SDVariable value)
static SDVariable	sizeOfTensorArray(SameDiff sd, SDVariable sequence)	Create an empty sequence with the specified data type.
static SDVariable	sizeOfTensorArray(SameDiff sd, SDVariable sequence, String outputVarName)	Create an empty sequence with the specified data type.
SDVariable	stack(SDVariable flow)
String	tensorflowName()	The opName of this function tensorflow
String	toString()
SDVariable	unstack(SDVariable flow, SDVariable value)
SDVariable	write(SDVariable flow, int index, SDVariable value)
SDVariable	write(SDVariable flow, SDVariable index, SDVariable value)

Methods inherited from class org.nd4j.linalg.api.ops.impl.shape.tensorops.BaseTensorOp

calculateOutputShape, computeArrays, doDiff, onnxName

Methods inherited from class org.nd4j.linalg.api.ops.DynamicCustomOp

addBArgument, addDArgument, addIArgument, addIArgument, addInputArgument, addOutputArgument, addOutputsToOp, addSArgument, addTArgument, assertValidForExecution, bArgs, builder, calculateOutputShape, clearArrays, dArgs, generateFake, generateFake, getBArgument, getDescriptor, getIArgument, getInputArgument, getOutputArgument, getSArgument, getTArgument, getValue, iArgs, initFromOnnx, inputArguments, mappingsForFunction, numBArguments, numDArguments, numIArguments, numInputArguments, numOutputArguments, numSArguments, numTArguments, opHash, opNum, outputArguments, outputVariables, outputVariables, propertiesForFunction, removeIArgument, removeInputArgument, removeOutputArgument, removeSArgument, removeTArgument, sArgs, setInputArgument, setInputArguments, setOutputArgument, setPropertiesForFunction, setValueFor, tArgs, wrapFilterNull, wrapOrNull, wrapOrNull

Methods inherited from class org.nd4j.autodiff.functions.DifferentialFunction

arg, arg, argNames, args, attributeAdaptersForFunction, configFieldName, configureWithSameDiff, diff, dup, equals, getBooleanFromProperty, getDoubleValueFromProperty, getIntValueFromProperty, getLongValueFromProperty, getStringFromProperty, hashCode, isConfigProperties, larg, onnxNames, outputs, outputVariable, outputVariablesNames, rarg, replaceArg, setInstanceId, tensorflowNames

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Methods inherited from interface org.nd4j.linalg.api.ops.CustomOp

isInplaceCall

Field Detail

tensorArrayDataType

protected DataType tensorArrayDataType

flow

protected SDVariable flow

clearOnRead

protected boolean clearOnRead

Constructor Detail

TensorArray

public TensorArray(String name,
                   SameDiff sameDiff,
                   DataType dataType)

TensorArray

public TensorArray(SameDiff sameDiff,
                   DataType dataType)

TensorArray

public TensorArray(TensorArray ta)

TensorArray

public TensorArray(TensorArray ta,
                   SDVariable[] inputs)

TensorArray

public TensorArray()

TensorArray

public TensorArray(DataType dataType)

Method Detail

tensorflowName

public String tensorflowName()

Description copied from class: DifferentialFunction

The opName of this function tensorflow

Overrides:

tensorflowName in class DynamicCustomOp

Returns:

configureFromArguments

public void configureFromArguments()

Description copied from interface: CustomOp

This allows a custom op to configure relevant fields from its arguments. This is needed when ops are created via reflection for things like model import.

Specified by:

configureFromArguments in interface CustomOp

Overrides:

configureFromArguments in class DynamicCustomOp

initFromTensorFlow

public void initFromTensorFlow(NodeDef nodeDef,
                               SameDiff initWith,
                               Map<String,AttrValue> attributesForNode,
                               GraphDef graph)

Description copied from class: DifferentialFunction

Initialize the function from the given NodeDef

Overrides:

initFromTensorFlow in class BaseTensorOp

toString

public String toString()

Overrides:

toString in class BaseTensorOp

opName

public String opName()

Description copied from class: DynamicCustomOp

This method returns op opName as string

Specified by:

opName in interface CustomOp

Overrides:

opName in class DynamicCustomOp

Returns:

opType

public Op.Type opType()

Description copied from class: DifferentialFunction

The type of the op

Overrides:

opType in class BaseTensorOp

Returns:

getVar

public SDVariable getVar()

getSameDiff

public SameDiff getSameDiff()

read

public SDVariable read(int index)

read

public SDVariable read(SDVariable from,
                       SDVariable index)

read

public SDVariable read(SDVariable index)

gather

public SDVariable gather(SDVariable flow,
                         int... indices)

gather

public SDVariable gather(SDVariable flow,
                         SDVariable indices)

stack

public SDVariable stack(SDVariable flow)

concat

public SDVariable concat(SDVariable flow)

write

public SDVariable write(SDVariable flow,
                        int index,
                        SDVariable value)

write

public SDVariable write(SDVariable flow,
                        SDVariable index,
                        SDVariable value)

scatter

public SDVariable scatter(SDVariable flow,
                          SDVariable value,
                          int... indices)

scatter

public SDVariable scatter(SDVariable flow,
                          SDVariable value,
                          SDVariable indices)

unstack

public SDVariable unstack(SDVariable flow,
                          SDVariable value)

size

public SDVariable size(SDVariable value)

remove

public SDVariable remove(SDVariable value,
                         SDVariable idx)

remove

public SDVariable remove(SDVariable value,
                         int idx)

remove

public SDVariable remove(SDVariable value)

calculateOutputDataTypes

public List<DataType> calculateOutputDataTypes(List<DataType> inputDataType)

Description copied from class: DifferentialFunction

Calculate the data types for the output arrays. Though datatypes can also be inferred from DifferentialFunction.calculateOutputShape(), this method differs in that it does not require the input arrays to be populated. This is important as it allows us to do greedy datatype inference for the entire net - even if arrays are not available.

Overrides:

calculateOutputDataTypes in class DifferentialFunction

Parameters:

inputDataType - The data types of the inputs

Returns:

The data types of the outputs

getNumOutputs

public int getNumOutputs()

Overrides:

getNumOutputs in class BaseTensorOp

itemAtIndex

public static SDVariable itemAtIndex(SameDiff sd,
                                     SDVariable[] inputs)

Returns the item at the specified index in the specified list.

Parameters:

sd - the same diff instance to use

inputs - the inputs including the relevant tensor array variable and position

Returns:

itemAtIndex

public static SDVariable itemAtIndex(SameDiff sd,
                                     SDVariable[] inputs,
                                     String outputVarName)

Returns the item at the specified index in the specified list. The output variable name to specify for the final output.

Parameters:

sd - the same diff instance to use

inputs - the inputs including the relevant tensor array variable and position

outputVarName - the name of the output variable for the read

Returns:

requiredShape

public long[] requiredShape()

Returns the required shape for elements in this tensor array. If a second input is not present an IllegalArgumentException is thrown.

Returns:

getTensorArray

public static TensorArray getTensorArray(SameDiff sd,
                                         SDVariable sequenceVar)

Get the associated TensorArray instance related to this op. Sometimes when a TensorArray op is returned it can be renamed or may not directly be the associated TensorArray instance. This helps discover the underlying TensorArray op for use to declare other operations to manipulate that instance such as read(int)

Parameters:

sd - the input instance

sequenceVar - the relevant variable to discover the TensorArray for

Returns:

removeFromTensorArray

public static SDVariable removeFromTensorArray(SameDiff sameDiff,
                                               SDVariable inputSequence)

Remove the last element from the relevant TensorArray

Parameters:

sameDiff - the samediff instance to use

inputSequence - the relevant variable for the associated TensorArray

Returns:

removeFromTensorArray

public static SDVariable removeFromTensorArray(SameDiff sameDiff,
                                               SDVariable inputSequence,
                                               SDVariable position)

Remove an element from the relevant TensorArray

Parameters:

sameDiff - the samediff sinstance to use

inputSequence - the relevant variable for the associated TensorArray

position - the position to remove

Returns:

removeFromTensorArray

public static SDVariable removeFromTensorArray(SameDiff sameDiff,
                                               SDVariable inputSequence,
                                               SDVariable position,
                                               String outputVarName)

Remove an element from the relevant TensorArray

Parameters:

sameDiff - the samediff instance to use

inputSequence - the relevant variable for the associated TensorArray

position - the position to remove

outputVarName - the name of the output variable

Returns:

sizeOfTensorArray

public static SDVariable sizeOfTensorArray(SameDiff sd,
                                           SDVariable sequence)

Create an empty sequence with the specified data type. An output variable name will be generated.

Parameters:

sd - the samediff instance to use

sequence - the output variable of the sequence to get the size of

Returns:

the output variable of the created sequence

sizeOfTensorArray

public static SDVariable sizeOfTensorArray(SameDiff sd,
                                           SDVariable sequence,
                                           String outputVarName)

Create an empty sequence with the specified data type. An output variable name will be generated.

Parameters:

sd - the samediff instance to use

sequence - the output variable of the sequence to get the size of

outputVarName - the output name of the size variable

Returns:

the output variable of the created sequence

createEmpty

public static SDVariable createEmpty(SameDiff sd,
                                     DataType dataType)

Create an empty sequence with the specified data type. An output variable name will be generated.

Parameters:

sd - the samediff instance to use

dataType - the data type of the sequence

Returns:

the output variable of the created sequence

createEmpty

public static SDVariable createEmpty(SameDiff sd,
                                     DataType dataType,
                                     String outputVarName)

Create an empty sequence with the specified data type.

Parameters:

sd - the samediff instance to use

dataType - the data type of the sequence

outputVarName - the output variable name of the sequence

Returns:

the output variable of the created sequence

createTensorArrayFrom

public static SDVariable createTensorArrayFrom(SameDiff sd,
                                               SDVariable[] inputs)

Create an TensorArray op from the given inputs, note this is the same as calling createTensorArrayFrom(SameDiff, SDVariable[],String) with null. The null value will avoid renaming the output

Parameters:

sd - the SameDiff instance to use

inputs - the input variables to create a TensorArray for

Returns:

the output variable for the tensor array

createTensorArrayFrom

public static SDVariable createTensorArrayFrom(SameDiff sd,
                                               SDVariable[] inputs,
                                               String outputVarName)

Create an TensorArray op from the given inputs

Parameters:

sd - the SameDiff instance to use

inputs - the input variables to create a TensorArray for

outputVarName - the name of the output variable to use for the final output of the loop

Returns:

the output variable for the tensor array