An InstanceNormalizationDescriptor for InstanceNormalizationLayer. More...

#include <Descriptors.hpp>

Inheritance diagram for InstanceNormalizationDescriptor:

Collaboration diagram for InstanceNormalizationDescriptor:

Public Member Functions
	InstanceNormalizationDescriptor ()
bool	operator== (const InstanceNormalizationDescriptor &rhs) const
Public Member Functions inherited from BaseDescriptor
virtual bool	IsNull () const
virtual	~BaseDescriptor ()=default

Public Attributes
float	m_Gamma
	Gamma, the scale scalar value applied for the normalized tensor. Defaults to 1.0.
float	m_Beta
	Beta, the offset scalar value applied for the normalized tensor. Defaults to 1.0.
float	m_Eps
	Epsilon, small scalar value added to variance to avoid dividing by zero. Defaults to 1e-12f.
DataLayout	m_DataLayout
	The data layout to be used (NCHW, NHWC).

Detailed Description

An InstanceNormalizationDescriptor for InstanceNormalizationLayer.

Definition at line 847 of file Descriptors.hpp.

Constructor & Destructor Documentation

◆ InstanceNormalizationDescriptor()

InstanceNormalizationDescriptor ( )

inline

Definition at line 849 of file Descriptors.hpp.

        : m_Gamma(1.0f)
        , m_Beta(0.0f)
        , m_Eps(1e-12f)
        , m_DataLayout(DataLayout::NCHW)
    {}

References m_Beta, m_DataLayout, m_Eps, m_Gamma, and armnn::NCHW.

Referenced by operator==().

Member Function Documentation

◆ operator==()

bool operator== ( const InstanceNormalizationDescriptor & rhs ) const

inline

Definition at line 856 of file Descriptors.hpp.

    {
        return m_Gamma      == rhs.m_Gamma &&
               m_Beta       == rhs.m_Beta &&
               m_Eps        == rhs.m_Eps &&
               m_DataLayout == rhs.m_DataLayout;
    }