org.nd4j.linalg.api.ops.impl.reduce.custom

Class BatchMmul

java.lang.Object

org.nd4j.autodiff.functions.DifferentialFunction

org.nd4j.linalg.api.ops.DynamicCustomOp

org.nd4j.linalg.api.ops.impl.reduce.custom.BatchMmul

All Implemented Interfaces:

CustomOp

public class BatchMmul
extends DynamicCustomOp

Batched matrix multiplication. Matrix multiply a batch of matrices. First and second batch of matrices have to be arrays of same length and each pair taken from these sets has to have dimensions (M, N) and (N, K), respectively. The result of this operation will be a batch of multiplied matrices. The result has the same length as both input batches and each output matrix is of shape (M, K).

Author:

Max Pumperla

Nested Class Summary

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

DynamicCustomOp.DynamicCustomOpsBuilder

Field Summary

Fields

Modifier and Type	Field and Description
protected int	batchSize
protected int	K
protected int	M
protected int	N
protected int	transposeA
protected int	transposeB

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

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

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

dimensions, extraArgs, inPlace, sameDiff, scalarValue

Constructor Summary

Constructors

Constructor and Description
BatchMmul()
BatchMmul(INDArray[] matricesA, INDArray[] matricesB, boolean transposeA, boolean transposeB)
BatchMmul(SameDiff sameDiff, SDVariable[] matrices, boolean transposeA, boolean transposeB)
BatchMmul(SameDiff sameDiff, SDVariable[] matricesA, SDVariable[] matricesB, boolean transposeA, boolean transposeB)

Method Summary

Modifier and Type	Method and Description
void	addArgs()
List<DataType>	calculateOutputDataTypes(List<DataType> dataTypes) Calculate the data types for the output arrays.
List<SDVariable>	doDiff(List<SDVariable> grads) The actual implementation for automatic differentiation.
int	getNumOutputs()
String	opName() This method returns op opName as string

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

addBArgument, addDArgument, addIArgument, addIArgument, addInputArgument, addOutputArgument, addTArgument, assertValidForExecution, bArgs, builder, calculateOutputShape, calculateOutputShape, clearArrays, dArgs, getBArgument, getDescriptor, getIArgument, getInputArgument, getOutputArgument, getTArgument, iArgs, initFromOnnx, initFromTensorFlow, inputArguments, numBArguments, numDArguments, numIArguments, numInputArguments, numOutputArguments, numTArguments, onnxName, opHash, opNum, opType, outputArguments, outputVariables, outputVariables, removeIArgument, removeInputArgument, removeOutputArgument, removeTArgument, setInputArgument, setInputArguments, setOutputArgument, tArgs, tensorflowName, toString, wrapFilterNull, wrapOrNull, wrapOrNull

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

arg, arg, argNames, args, attributeAdaptersForFunction, configFieldName, diff, dup, equals, getValue, hashCode, isConfigProperties, larg, mappingsForFunction, onnxNames, outputs, outputVariable, outputVariablesNames, propertiesForFunction, rarg, replaceArg, setInstanceId, setPropertiesForFunction, setValueFor, 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

transposeA

protected int transposeA

transposeB

protected int transposeB

batchSize

protected int batchSize

M

protected int M

N

protected int N

K

protected int K

Constructor Detail

BatchMmul

public BatchMmul(SameDiff sameDiff,
                 SDVariable[] matricesA,
                 SDVariable[] matricesB,
                 boolean transposeA,
                 boolean transposeB)

BatchMmul

public BatchMmul(SameDiff sameDiff,
                 SDVariable[] matrices,
                 boolean transposeA,
                 boolean transposeB)

BatchMmul

public BatchMmul(INDArray[] matricesA,
                 INDArray[] matricesB,
                 boolean transposeA,
                 boolean transposeB)

BatchMmul

public BatchMmul()

Method Detail

getNumOutputs

public int getNumOutputs()

Overrides:

getNumOutputs in class DifferentialFunction

addArgs

public void addArgs()

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:

doDiff

public List<SDVariable> doDiff(List<SDVariable> grads)

Description copied from class: DifferentialFunction

The actual implementation for automatic differentiation.

Overrides:

doDiff in class DynamicCustomOp

Returns:

calculateOutputDataTypes

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

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:

dataTypes - The data types of the inputs

Returns:

The data types of the outputs