iganet::IgANetGeneratorImpl< real_t > Class Template Reference

IgANetGeneratorImpl. More...

#include </home/runner/work/iganet/iganet/include/iganet/net/generator.hpp>

Inheritance diagram for iganet::IgANetGeneratorImpl< real_t >:

Public Member Functions
	IgANetGeneratorImpl ()=default
	Default constructor.
	IgANetGeneratorImpl (const std::vector< int64_t > &layers, const std::vector< std::vector< std::any > > &activations, Options< real_t > options=Options< real_t >{})
	Constructor.
torch::Tensor	forward (torch::Tensor x)
	Forward evaluation.
void	pretty_print (std::ostream &os) const noexcept override
	Provides the pretty_print operation.
torch::serialize::InputArchive &	read (torch::serialize::InputArchive &archive, const std::string &key="iganet")
	Reads the IgANet from a torch::serialize::InputArchive object.
torch::serialize::OutputArchive &	write (torch::serialize::OutputArchive &archive, const std::string &key="iganet") const
	Writes the IgANet into a torch::serialize::OutputArchive object.

Private Attributes
std::vector< std::unique_ptr< iganet::ActivationFunction > >	activations_
	Vector of activation functions.
std::vector< torch::nn::Linear >	layers_
	Vector of linear layers.

Detailed Description

template<typename real_t>
class iganet::IgANetGeneratorImpl< real_t >

IgANetGeneratorImpl.

Note

Following the discussion of module overship here.

https://pytorch.org/tutorials/advanced/cpp_frontend.html#module-ownership.

we implement a generator implementation class following.

https://pytorch.org/tutorials/advanced/cpp_frontend.html#the-generator-module.

Constructor & Destructor Documentation

IgANetGeneratorImpl() [1/2]

template<typename real_t >

iganet::IgANetGeneratorImpl< real_t >::IgANetGeneratorImpl ( )

default

Default constructor.

IgANetGeneratorImpl() [2/2]

template<typename real_t >

iganet::IgANetGeneratorImpl< real_t >::IgANetGeneratorImpl ( const std::vector< int64_t > &  layers, const std::vector< std::vector< std::any > > &  activations, Options< real_t >  options = Options<real_t>{}  )

inlineexplicit

Constructor.

Parameters

layers	Value of layers.
activations	Value of activations.
options	Configuration options.

Member Function Documentation

forward()

template<typename real_t >

torch::Tensor iganet::IgANetGeneratorImpl< real_t >::forward ( torch::Tensor  x )

inline

Forward evaluation.

Parameters

x	Value of x.

Returns

Result of the operation.

pretty_print()

template<typename real_t >

void iganet::IgANetGeneratorImpl< real_t >::pretty_print ( std::ostream &  os ) const

inlineoverridenoexcept

Provides the pretty_print operation.

Parameters

os	Output stream.

read()

template<typename real_t >

torch::serialize::InputArchive & iganet::IgANetGeneratorImpl< real_t >::read ( torch::serialize::InputArchive &  archive, const std::string &  key = "iganet"  )

inline

Reads the IgANet from a torch::serialize::InputArchive object.

Parameters

archive	Serialization archive.
key	Serialization key.

Returns

Result of the operation.

write()

template<typename real_t >

torch::serialize::OutputArchive & iganet::IgANetGeneratorImpl< real_t >::write ( torch::serialize::OutputArchive &  archive, const std::string &  key = "iganet"  ) const

inline

Writes the IgANet into a torch::serialize::OutputArchive object.

Parameters

archive	Serialization archive.
key	Serialization key.

Returns

Result of the operation.

Member Data Documentation

activations_

template<typename real_t >

std::vector<std::unique_ptr<iganet::ActivationFunction> > iganet::IgANetGeneratorImpl< real_t >::activations_

private

Vector of activation functions.

layers_

template<typename real_t >

std::vector<torch::nn::Linear> iganet::IgANetGeneratorImpl< real_t >::layers_

private

Vector of linear layers.

---

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

• include/iganet/net/generator.hpp