onert::backend::train::ops::GradientApplier Class Reference

#include <GradientApplier.h>

Collaboration diagram for onert::backend::train::ops::GradientApplier:

Public Member Functions
	GradientApplier ()
	~GradientApplier ()=default
void	configure (const exec::train::optimizer::Optimizer *optimizer, const IPortableTensor *gradient, ITrainableTensor *trainable)
void	applyGradient (uint32_t training_step) override
	Apply gradients to a trainable tensor.
Public Member Functions inherited from onert::exec::train::IGradientApplier
virtual	~IGradientApplier ()=default

Detailed Description

Definition at line 27 of file GradientApplier.h.

Constructor & Destructor Documentation

GradientApplier()

onert::backend::train::ops::GradientApplier::GradientApplier

(

)

Definition at line 24 of file GradientApplier.cc.

                                 : _optimizer{nullptr}, _gradient_tensor{}, _trainable_tensor{}
{
  // DO NOTHING
}

~GradientApplier()

onert::backend::train::ops::GradientApplier::~GradientApplier ( )

default

Member Function Documentation

applyGradient()

void onert::backend::train::ops::GradientApplier::applyGradient ( uint32_t  training_step )

overridevirtual

Apply gradients to a trainable tensor.

Parameters

training_step

The number of iterations of the training process.

Implements onert::exec::train::IGradientApplier.

Definition at line 37 of file GradientApplier.cc.

{
  _optimizer->applyGradient(
    std::forward_as_tuple(*_gradient_tensor, *_trainable_tensor, training_step));
}

References onert::exec::train::optimizer::Optimizer::applyGradient().

configure()

void onert::backend::train::ops::GradientApplier::configure	(	const exec::train::optimizer::Optimizer *	optimizer,
		const IPortableTensor *	gradient,
		ITrainableTensor *	trainable
	)

Definition at line 29 of file GradientApplier.cc.

{
  _optimizer = optimizer;
  _gradient_tensor = gradient;
  _trainable_tensor = trainable;
}

---

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

• runtime/onert/backend/train/ops/GradientApplier.h
• runtime/onert/backend/train/ops/GradientApplier.cc