dl::optimization::AutogradOptimizer< T > Class Template Reference

Base class for autograd-compatible optimizers. More...

#include <optimizers.hpp>

Inheritance diagram for dl::optimization::AutogradOptimizer< T >:

Collaboration diagram for dl::optimization::AutogradOptimizer< T >:

Public Member Functions
	AutogradOptimizer (std::vector< Variable< T > * > parameters)
	Constructor.
virtual	~AutogradOptimizer ()=default
virtual void	step ()=0
	Perform one optimization step.
virtual void	zero_grad ()
	Zero gradients of all parameters.
virtual T	get_lr () const =0
	Get learning rate.
virtual void	set_lr (T lr)=0
	Set learning rate.

Protected Attributes
std::vector< Variable< T > * >	parameters_

Detailed Description

template<typename T>
class dl::optimization::AutogradOptimizer< T >

Base class for autograd-compatible optimizers.

Definition at line 28 of file optimizers.hpp.

Constructor & Destructor Documentation

AutogradOptimizer()

template<typename T >

dl::optimization::AutogradOptimizer< T >::AutogradOptimizer ( std::vector< Variable< T > * >  parameters )

inlineexplicit

Constructor.

Parameters

parameters

Vector of parameter variables to optimize

Definition at line 34 of file optimizers.hpp.

~AutogradOptimizer()

template<typename T >

virtual dl::optimization::AutogradOptimizer< T >::~AutogradOptimizer ( )

virtualdefault

Member Function Documentation

get_lr()

template<typename T >

virtual T dl::optimization::AutogradOptimizer< T >::get_lr ( ) const

pure virtual

Get learning rate.

Implemented in dl::optimization::SGD< T >, dl::optimization::Adam< T >, dl::optimization::AdamW< T >, and dl::optimization::RMSprop< T >.

set_lr()

template<typename T >

virtual void dl::optimization::AutogradOptimizer< T >::set_lr ( T  lr )

pure virtual

Set learning rate.

Implemented in dl::optimization::SGD< T >, dl::optimization::Adam< T >, dl::optimization::AdamW< T >, and dl::optimization::RMSprop< T >.

step()

template<typename T >

virtual void dl::optimization::AutogradOptimizer< T >::step ( )

pure virtual

Perform one optimization step.

Implemented in dl::optimization::SGD< T >, dl::optimization::Adam< T >, dl::optimization::AdamW< T >, and dl::optimization::RMSprop< T >.

zero_grad()

template<typename T >

virtual void dl::optimization::AutogradOptimizer< T >::zero_grad ( )

inlinevirtual

Zero gradients of all parameters.

Definition at line 47 of file optimizers.hpp.

Member Data Documentation

parameters_

template<typename T >

std::vector<Variable<T>*> dl::optimization::AutogradOptimizer< T >::parameters_

protected

Definition at line 64 of file optimizers.hpp.

---

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

• include/optimization/optimizers.hpp