dl::layers::Module< T > Class Template Reference

Base class for all neural network modules (PyTorch-like nn.Module) More...

#include <layers.hpp>

Inheritance diagram for dl::layers::Module< T >:

Public Member Functions
virtual	~Module ()=default
virtual Variable< T >	forward (const Variable< T > &input)=0
	Forward pass through the module.
virtual std::vector< Variable< T > * >	parameters ()=0
	Get all parameters of this module.
virtual void	zero_grad ()
	Zero gradients of all parameters.
virtual void	train (bool training=true)
	Set training mode.
virtual void	eval ()
	Set evaluation mode.
bool	is_training () const
	Check if module is in training mode.

Protected Attributes
bool	training_ = true

Detailed Description

template<typename T>
class dl::layers::Module< T >

Base class for all neural network modules (PyTorch-like nn.Module)

Definition at line 28 of file layers.hpp.

Constructor & Destructor Documentation

~Module()

template<typename T >

virtual dl::layers::Module< T >::~Module ( )

virtualdefault

Member Function Documentation

eval()

template<typename T >

virtual void dl::layers::Module< T >::eval ( )

inlinevirtual

Set evaluation mode.

Definition at line 65 of file layers.hpp.

forward()

template<typename T >

virtual Variable< T > dl::layers::Module< T >::forward ( const Variable< T > &  input )

pure virtual

Forward pass through the module.

Parameters

input

Input variable

Returns

Output variable

Implemented in dl::layers::Linear< T >, dl::layers::ReLU< T >, dl::layers::Sigmoid< T >, dl::layers::Tanh< T >, dl::layers::Dropout< T >, and dl::layers::Sequential< T >.

is_training()

template<typename T >

bool dl::layers::Module< T >::is_training ( ) const

inline

Check if module is in training mode.

Definition at line 72 of file layers.hpp.

parameters()

template<typename T >

virtual std::vector< Variable< T > * > dl::layers::Module< T >::parameters ( )

pure virtual

Get all parameters of this module.

Returns

Vector of parameter variables

Implemented in dl::layers::Linear< T >, dl::layers::ReLU< T >, dl::layers::Sigmoid< T >, dl::layers::Tanh< T >, dl::layers::Dropout< T >, and dl::layers::Sequential< T >.

train()

template<typename T >

virtual void dl::layers::Module< T >::train ( bool  training = true )

inlinevirtual

Set training mode.

Parameters

training

Whether in training mode

Reimplemented in dl::layers::Sequential< T >.

Definition at line 58 of file layers.hpp.

zero_grad()

template<typename T >

virtual void dl::layers::Module< T >::zero_grad ( )

inlinevirtual

Zero gradients of all parameters.

Reimplemented in dl::layers::Sequential< T >.

Definition at line 48 of file layers.hpp.

Member Data Documentation

training_

template<typename T >

bool dl::layers::Module< T >::training_ = true

protected

Definition at line 75 of file layers.hpp.

---

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

• include/neural_network/layers.hpp