ArmNN
 24.08
QASymmS8Decoder Class Reference

#include <BaseIterator.hpp>

Inheritance diagram for QASymmS8Decoder:
[legend]
Collaboration diagram for QASymmS8Decoder:
[legend]

Public Member Functions

 QASymmS8Decoder (const int8_t *data, const float scale, const int32_t offset)
 
 QASymmS8Decoder (const float scale, const int32_t offset)
 
float Get () const override
 
std::vector< float > DecodeTensor (const TensorShape &tensorShape, const bool) override
 
- Public Member Functions inherited from TypedIterator< const int8_t, Decoder< float > >
 TypedIterator (const int8_t *data=nullptr)
 
void Reset (void *data) override
 
TypedIteratoroperator++ () override
 
TypedIteratoroperator+= (const unsigned int increment) override
 
TypedIteratoroperator-= (const unsigned int increment) override
 
TypedIteratoroperator[] (const unsigned int index) override
 
- Public Member Functions inherited from Decoder< float >
 Decoder ()
 
virtual ~Decoder ()
 
virtual float Get () const=0
 
- Public Member Functions inherited from BaseIterator
 BaseIterator ()
 
virtual ~BaseIterator ()
 

Additional Inherited Members

- Protected Attributes inherited from TypedIterator< const int8_t, Decoder< float > >
const int8_t * m_Iterator
 
const int8_t * m_Start
 

Detailed Description

Definition at line 145 of file BaseIterator.hpp.

Constructor & Destructor Documentation

◆ QASymmS8Decoder() [1/2]

QASymmS8Decoder ( const int8_t *  data,
const float  scale,
const int32_t  offset 
)
inline

Definition at line 148 of file BaseIterator.hpp.

149  : TypedIterator(data), m_Scale(scale), m_Offset(offset) {}

◆ QASymmS8Decoder() [2/2]

QASymmS8Decoder ( const float  scale,
const int32_t  offset 
)
inline

Definition at line 151 of file BaseIterator.hpp.

152  : QASymmS8Decoder(nullptr, scale, offset) {}

Member Function Documentation

◆ DecodeTensor()

std::vector<float> DecodeTensor ( const TensorShape tensorShape,
const bool   
)
inlineoverridevirtual

Implements Decoder< float >.

Definition at line 158 of file BaseIterator.hpp.

159  {
160  const unsigned int size = tensorShape.GetNumElements();
161  std::vector<float> decodedTensor;
162  decodedTensor.reserve(size);
163 
164  for (uint32_t i = 0; i < size; ++i)
165  {
166  this->operator[](i);
167  decodedTensor.emplace_back(armnn::Dequantize(*m_Iterator, m_Scale, m_Offset));
168  }
169 
170  return decodedTensor;
171  }

References armnn::Dequantize(), TensorShape::GetNumElements(), TypedIterator< const int8_t, Decoder< float > >::m_Iterator, and TypedIterator< const int8_t, Decoder< float > >::operator[]().

◆ Get()

float Get ( ) const
inlineoverride

Definition at line 154 of file BaseIterator.hpp.

155  {
156  return armnn::Dequantize(*m_Iterator, m_Scale, m_Offset);
157  }

References armnn::Dequantize(), and TypedIterator< const int8_t, Decoder< float > >::m_Iterator.


The documentation for this class was generated from the following file:
armnn::TypedIterator< const int8_t, Decoder< float > >::m_Iterator
const int8_t * m_Iterator
Definition: BaseIterator.hpp:107
armnn::QASymmS8Decoder::QASymmS8Decoder
QASymmS8Decoder(const int8_t *data, const float scale, const int32_t offset)
Definition: BaseIterator.hpp:148
armnn::TypedIterator< const int8_t, Decoder< float > >::TypedIterator
TypedIterator(const int8_t *data=nullptr)
Definition: BaseIterator.hpp:68
armnn::Dequantize
float Dequantize(QuantizedType value, float scale, int32_t offset)
Dequantize an 8-bit data type into a floating point data type.
Definition: TypesUtils.cpp:48
armnn::TypedIterator< const int8_t, Decoder< float > >::operator[]
TypedIterator & operator[](const unsigned int index) override
Definition: BaseIterator.hpp:99