template<typename T>
class TMVA::Experimental::SOFIE::ROperator_RNN< T >

Recurrent Neural Network operator.

Inference code generation for one-layer vanilla RNN. Supports forward, reverse and bidirectional RNNs. See the ONNX documentation for details about the supported RNN architectures.

Definition at line 22 of file ROperator_RNN.hxx.

Public Member Functions
	ROperator_RNN ()
	Default constructor of ROperator_RNN.

	ROperator_RNN (std::vector< float > activation_alpha, std::vector< float > activation_beta, std::vector< std::string > activations, float clip, std::string direction, size_t hidden_size, size_t layout, std::string nameX, std::string nameW, std::string nameR, std::string nameB, std::string nameSequence_lens, std::string nameInitial_h, std::string nameY, std::string nameY_h)
	Constructor of ROperator_RNN from the attributes.

std::string	Generate (std::string OpName)
	Generates the inference code.

std::string	GenerateSessionMembersCode (std::string opName)

void	Initialize (RModel &model)
	Initialize the model.

std::vector< std::vector< size_t > >	ShapeInference (std::vector< std::vector< size_t > > input)
	Infers the shape of the output tensors.

std::vector< ETensorType >	TypeInference (std::vector< ETensorType > input)
	Infers the type of the output tensors.

Public Member Functions inherited from TMVA::Experimental::SOFIE::ROperator
virtual	~ROperator ()

virtual std::string	GenerateInitCode ()

virtual std::string	Header ()

Private Attributes
std::vector< float >	fAttrActivationAlpha
	Scaling values used by some activation functions.

std::vector< float >	fAttrActivationBeta
	Scaling values used by some activation functions.

std::vector< std::string >	fAttrActivations
	Activation functions.

float	fAttrClip
	Clip threshold.

std::string	fAttrDirection
	Direction of processing.

size_t	fAttrHiddenSize
	Number of the hidden layers.

size_t	fAttrLayout
	Data layout.

std::string	fNB
	Name of the bias.

std::string	fNInitial_h
	Name of the initial value of the hidden states.

std::string	fNR
	Name of the recurrence.

std::string	fNSequence_lens
	Name of the length of the sequences.

std::string	fNW
	Name of the weights.

std::string	fNX
	Name of the input.

std::string	fNY
	Name of the output.

std::string	fNY_h
	Name of the last sequence of the output.

std::vector< size_t >	fShapeB
	Shape of the bias.

std::vector< size_t >	fShapeInitial_h
	Shape of the initial value of the hidden states.

std::vector< size_t >	fShapeR
	Shape of the recurrence.

std::vector< size_t >	fShapeSequence_lens
	Shape of the length of the sequences.

std::vector< size_t >	fShapeW
	Shape of the weights.

std::vector< size_t >	fShapeX
	Shape of the input.

std::vector< size_t >	fShapeY
	Shape of the output.

std::vector< size_t >	fShapeY_h
	Shape of the last sequence of the output.

std::string	fType
	Type of the tensors.

Additional Inherited Members
Protected Attributes inherited from TMVA::Experimental::SOFIE::ROperator
bool	fUseSession = false
	flag to identify if using the session class

const std::string	SP = " "
	space used to correctly indent the generated C++ code

#include <TMVA/ROperator_RNN.hxx>

Inheritance diagram for TMVA::Experimental::SOFIE::ROperator_RNN< T >:

[legend]

Constructor & Destructor Documentation

◆ ROperator_RNN() [1/2]

template<typename T >

TMVA::Experimental::SOFIE::ROperator_RNN< T >::ROperator_RNN ( )

inline

Default constructor of ROperator_RNN.

Definition at line 54 of file ROperator_RNN.hxx.

◆ ROperator_RNN() [2/2]

template<typename T >

TMVA::Experimental::SOFIE::ROperator_RNN< T >::ROperator_RNN	(	std::vector< float >	activation_alpha,
		std::vector< float >	activation_beta,
		std::vector< std::string >	activations,
		float	clip,
		std::string	direction,
		size_t	hidden_size,
		size_t	layout,
		std::string	nameX,
		std::string	nameW,
		std::string	nameR,
		std::string	nameB,
		std::string	nameSequence_lens,
		std::string	nameInitial_h,
		std::string	nameY,
		std::string	nameY_h
	)

inline

Constructor of ROperator_RNN from the attributes.

Parameters

activation_alpha	scaling values used by some activation functions
activation_beta	scaling values used by some activation functions
activations	activation functions
clip	clip threshold
direction	direction of processing of the sequneces
hidden_size	number of hidden layers
layout	data layout
nameX	name of the input tensor
nameW	name of the weight tensor
nameR	name of the recurrence tensor
nameB	name of the bias tensor
nameSequence_lens	name of the length of the sequences
nameInitial_h	name of the initial value of the hidden states
nameY	name of the output
nameY_h	name of the last sequence of the output

Definition at line 74 of file ROperator_RNN.hxx.

Member Function Documentation

◆ Generate()

template<typename T >

auto TMVA::Experimental::SOFIE::ROperator_RNN< T >::Generate ( std::string OpName )

virtual

Generates the inference code.

Parameters

OpName name of the operator

Implements TMVA::Experimental::SOFIE::ROperator.

Definition at line 221 of file ROperator_RNN.icc.

◆ GenerateSessionMembersCode()

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::GenerateSessionMembersCode ( std::string opName )

virtual

Reimplemented from TMVA::Experimental::SOFIE::ROperator.

Definition at line 190 of file ROperator_RNN.icc.

◆ Initialize()

template<typename T >

auto TMVA::Experimental::SOFIE::ROperator_RNN< T >::Initialize ( RModel & model )

virtual

Initialize the model.

Parameters

model Model

Implements TMVA::Experimental::SOFIE::ROperator.

Definition at line 38 of file ROperator_RNN.icc.

◆ ShapeInference()

template<typename T >

auto TMVA::Experimental::SOFIE::ROperator_RNN< T >::ShapeInference ( std::vector< std::vector< size_t > > input )

virtual

Infers the shape of the output tensors.

Parameters

input shape of the input tensors

Implements TMVA::Experimental::SOFIE::ROperator.

Definition at line 16 of file ROperator_RNN.icc.

◆ TypeInference()

template<typename T >

auto TMVA::Experimental::SOFIE::ROperator_RNN< T >::TypeInference ( std::vector< ETensorType > input )

virtual

Infers the type of the output tensors.

Parameters

input type of the input tensors

Implements TMVA::Experimental::SOFIE::ROperator.

Definition at line 9 of file ROperator_RNN.icc.

Member Data Documentation

◆ fAttrActivationAlpha

template<typename T >

std::vector<float> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrActivationAlpha

private

Scaling values used by some activation functions.

Definition at line 24 of file ROperator_RNN.hxx.

◆ fAttrActivationBeta

template<typename T >

std::vector<float> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrActivationBeta

private

Scaling values used by some activation functions.

Definition at line 25 of file ROperator_RNN.hxx.

◆ fAttrActivations

template<typename T >

std::vector<std::string> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrActivations

private

Activation functions.

Definition at line 26 of file ROperator_RNN.hxx.

◆ fAttrClip

template<typename T >

float TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrClip

private

Clip threshold.

Definition at line 27 of file ROperator_RNN.hxx.

◆ fAttrDirection

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrDirection

private

Direction of processing.

Definition at line 28 of file ROperator_RNN.hxx.

◆ fAttrHiddenSize

template<typename T >

size_t TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrHiddenSize

private

Number of the hidden layers.

Definition at line 29 of file ROperator_RNN.hxx.

◆ fAttrLayout

template<typename T >

size_t TMVA::Experimental::SOFIE::ROperator_RNN< T >::fAttrLayout

private

Data layout.

Definition at line 30 of file ROperator_RNN.hxx.

◆ fNB

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNB

private

Name of the bias.

Definition at line 35 of file ROperator_RNN.hxx.

◆ fNInitial_h

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNInitial_h

private

Name of the initial value of the hidden states.

Definition at line 37 of file ROperator_RNN.hxx.

◆ fNR

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNR

private

Name of the recurrence.

Definition at line 34 of file ROperator_RNN.hxx.

◆ fNSequence_lens

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNSequence_lens

private

Name of the length of the sequences.

Definition at line 36 of file ROperator_RNN.hxx.

◆ fNW

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNW

private

Name of the weights.

Definition at line 33 of file ROperator_RNN.hxx.

◆ fNX

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNX

private

Name of the input.

Definition at line 32 of file ROperator_RNN.hxx.

◆ fNY

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNY

private

Name of the output.

Definition at line 38 of file ROperator_RNN.hxx.

◆ fNY_h

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fNY_h

private

Name of the last sequence of the output.

Definition at line 39 of file ROperator_RNN.hxx.

◆ fShapeB

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeB

private

Shape of the bias.

Definition at line 44 of file ROperator_RNN.hxx.

◆ fShapeInitial_h

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeInitial_h

private

Shape of the initial value of the hidden states.

Definition at line 46 of file ROperator_RNN.hxx.

◆ fShapeR

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeR

private

Shape of the recurrence.

Definition at line 43 of file ROperator_RNN.hxx.

◆ fShapeSequence_lens

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeSequence_lens

private

Shape of the length of the sequences.

Definition at line 45 of file ROperator_RNN.hxx.

◆ fShapeW

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeW

private

Shape of the weights.

Definition at line 42 of file ROperator_RNN.hxx.

◆ fShapeX

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeX

private

Shape of the input.

Definition at line 41 of file ROperator_RNN.hxx.

◆ fShapeY

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeY

private

Shape of the output.

Definition at line 47 of file ROperator_RNN.hxx.

◆ fShapeY_h

template<typename T >

std::vector<size_t> TMVA::Experimental::SOFIE::ROperator_RNN< T >::fShapeY_h

private

Shape of the last sequence of the output.

Definition at line 48 of file ROperator_RNN.hxx.

◆ fType

template<typename T >

std::string TMVA::Experimental::SOFIE::ROperator_RNN< T >::fType

private

Type of the tensors.

Definition at line 50 of file ROperator_RNN.hxx.

tmva/sofie/inc/TMVA/ROperator_RNN.hxx

tmva/sofie/inc/TMVA/ROperator_RNN.icc

Public Member Functions

Private Attributes

Additional Inherited Members

Constructor & Destructor Documentation

◆ ROperator_RNN() [1/2]

◆ ROperator_RNN() [2/2]

Member Function Documentation

◆ Generate()

◆ GenerateSessionMembersCode()

◆ Initialize()

◆ ShapeInference()

◆ TypeInference()

Member Data Documentation

◆ fAttrActivationAlpha

◆ fAttrActivationBeta

◆ fAttrActivations

◆ fAttrClip

◆ fAttrDirection

◆ fAttrHiddenSize

◆ fAttrLayout

◆ fNB

◆ fNInitial_h

◆ fNR

◆ fNSequence_lens

◆ fNW

◆ fNX

◆ fNY

◆ fNY_h

◆ fShapeB

◆ fShapeInitial_h

◆ fShapeR

◆ fShapeSequence_lens

◆ fShapeW

◆ fShapeX

◆ fShapeY

◆ fShapeY_h

◆ fType