onert::exec::train::optimizer::Optimizer Class Reference

Base class for all optimizers. More...

#include <Optimizer.h>

Public Member Functions
virtual	~Optimizer ()=default
virtual std::string	name () const
	Get the name of optimizer.
virtual double	getLearningRate (uint32_t iteration) const =0
	Get the Learning Rate.
virtual uint32_t	getVarCount () const =0
	Get the number of optimizer variables s.
virtual void	applyGradient (const UpdateFactors &factors) const =0
	Apply gradient to a trainable tensor.

Detailed Description

Base class for all optimizers.

for optimizers

Definition at line 42 of file Optimizer.h.

Constructor & Destructor Documentation

~Optimizer()

virtual onert::exec::train::optimizer::Optimizer::~Optimizer ( )

virtualdefault

Member Function Documentation

applyGradient()

virtual void onert::exec::train::optimizer::Optimizer::applyGradient ( const UpdateFactors &  factors ) const

pure virtual

Apply gradient to a trainable tensor.

Parameters

factors

UpdateFactors to be used for applying gradient to a trainable tensor

Implemented in onert::backend::train::optimizer::Adam, and onert::backend::train::optimizer::SGD.

Referenced by onert::backend::train::ops::GradientApplier::applyGradient().

getLearningRate()

virtual double onert::exec::train::optimizer::Optimizer::getLearningRate ( uint32_t  iteration ) const

pure virtual

Get the Learning Rate.

Parameters

iteration

The number of training steps

Returns

Learning rate

Implemented in onert::backend::train::optimizer::SGD, and onert::backend::train::optimizer::Adam.

getVarCount()

virtual uint32_t onert::exec::train::optimizer::Optimizer::getVarCount ( ) const

pure virtual

Get the number of optimizer variables s.

Returns

The number of optimizer variables

Implemented in onert::backend::train::optimizer::Adam, and onert::backend::train::optimizer::SGD.

Referenced by onert::backend::train::TensorBuilder::registerBackwardTensorInfo().

name()

virtual std::string onert::exec::train::optimizer::Optimizer::name ( ) const

inlinevirtual

Get the name of optimizer.

Returns

The name of optimizer

Reimplemented in onert::backend::train::optimizer::Adam, and onert::backend::train::optimizer::SGD.

Definition at line 52 of file Optimizer.h.

{ return std::string{"Invalid"}; }

---

The documentation for this class was generated from the following file:

• runtime/onert/core/include/exec/train/optimizer/Optimizer.h