Compute Library
 20.08
UniformQuantizationInfo Struct Reference

Quantization info when assuming per layer quantization. More...

#include <QuantizationInfo.h>

Public Member Functions

 UniformQuantizationInfo ()
 Default constructor. More...
 
 UniformQuantizationInfo (float scale, int32_t offset)
 Constructor. More...
 
bool empty () const
 Checks if the scale and offset are both zero. More...
 

Data Fields

float scale
 
int32_t offset
 

Detailed Description

Quantization info when assuming per layer quantization.

Definition at line 43 of file QuantizationInfo.h.

Constructor & Destructor Documentation

◆ UniformQuantizationInfo() [1/2]

Default constructor.

Definition at line 46 of file QuantizationInfo.h.

◆ UniformQuantizationInfo() [2/2]

UniformQuantizationInfo ( float  scale,
int32_t  offset 
)
inline

Constructor.

Parameters
[in]scaleQuantization scale
[in]offsetQuantization offset

Definition at line 55 of file QuantizationInfo.h.

Member Function Documentation

◆ empty()

bool empty ( ) const
inline

Checks if the scale and offset are both zero.

Definition at line 60 of file QuantizationInfo.h.

61  {
62  return (scale == 0) && (offset == 0);
63  }

References UniformQuantizationInfo::offset, and UniformQuantizationInfo::scale.

Field Documentation

◆ offset

int32_t offset

Definition at line 66 of file QuantizationInfo.h.

Referenced by arm_compute::compute_requantization_scale_offset(), CLActivationLayerKernel::configure(), CLWidthConcatenate2TensorsKernel::configure(), CLHeightConcatenateLayerKernel::configure(), CLWidthConcatenateLayerKernel::configure(), CLWidthConcatenate4TensorsKernel::configure(), CLGEMMMatrixVectorMultiplyKernel::configure(), CLDepthConcatenateLayerKernel::configure(), CLBatchConcatenateLayerKernel::configure(), CLComparisonKernel::configure(), CLQuantizationLayerKernel::configure(), CLPoolingLayerKernel::configure(), CLBoundingBoxTransformKernel::configure(), CLDepthwiseConvolutionLayer3x3NCHWKernel::configure(), CLDepthwiseConvolutionLayer3x3NHWCKernel::configure(), CLROIAlignLayerKernel::configure(), CLDepthwiseConvolutionLayerNativeKernel::configure(), CLDirectConvolutionLayerKernel::configure(), CLGEMMLowpMatrixMultiplyCore::configure(), NEGEMMLowpMatrixMultiplyCore::configure(), NELSTMLayerQuantized::configure(), CLPixelWiseMultiplicationKernel::configure(), CLGEMMDeconvolutionLayer::configure(), NEQLSTMLayer::configure(), CLLSTMLayerQuantized::configure(), CLQLSTMLayer::configure(), UniformQuantizationInfo::empty(), arm_compute::get_quantized_activation_min_max(), main(), arm_compute::operator<<(), arm_compute::operator==(), arm_compute::roi_align_1x1_qasymm8(), CPPUpsampleKernel::run(), QuantizationInfo::uniform(), NEGenerateProposalsLayer::validate(), CLGEMMLowpMatrixMultiplyCore::validate(), NEGEMMLowpMatrixMultiplyCore::validate(), CLGenerateProposalsLayer::validate(), NELSTMLayerQuantized::validate(), CLLSTMLayerQuantized::validate(), NEQLSTMLayer::validate(), arm_compute::validate(), CLQLSTMLayer::validate(), arm_compute::vdequantize(), arm_compute::vquantize(), arm_compute::vquantize_qasymm16(), and arm_compute::vquantize_signed().

◆ scale

float scale

Definition at line 65 of file QuantizationInfo.h.

Referenced by arm_compute::quantization::compute_quantized_multipliers_and_shifts(), arm_compute::compute_requantization_scale_offset(), CLActivationLayerKernel::configure(), CLWidthConcatenate2TensorsKernel::configure(), CLHeightConcatenateLayerKernel::configure(), CLWidthConcatenateLayerKernel::configure(), CLWidthConcatenate4TensorsKernel::configure(), CLDepthConcatenateLayerKernel::configure(), NEQLSTMLayerNormalizationKernel::configure(), CLBatchConcatenateLayerKernel::configure(), CLQLSTMLayerNormalizationKernel::configure(), CLComparisonKernel::configure(), CLQuantizationLayerKernel::configure(), CLPoolingLayerKernel::configure(), NEDepthwiseConvolutionLayerNativeKernel::configure(), CLBoundingBoxTransformKernel::configure(), CLDepthwiseConvolutionLayer3x3NCHWKernel::configure(), CLDepthwiseConvolutionLayer3x3NHWCKernel::configure(), CLROIAlignLayerKernel::configure(), CLDepthwiseConvolutionLayerNativeKernel::configure(), CLDirectConvolutionLayerKernel::configure(), NEGEMMLowpMatrixMultiplyCore::configure(), NELSTMLayerQuantized::configure(), CLPixelWiseMultiplicationKernel::configure(), CLGEMMDeconvolutionLayer::configure(), NEQLSTMLayer::configure(), CLLSTMLayerQuantized::configure(), CLQLSTMLayer::configure(), UniformQuantizationInfo::empty(), main(), arm_compute::operator<<(), arm_compute::operator==(), QuantizationInfo::uniform(), NEDepthwiseConvolutionAssemblyDispatch::validate(), NEGenerateProposalsLayer::validate(), NEGEMMLowpMatrixMultiplyCore::validate(), CLGenerateProposalsLayer::validate(), NELSTMLayerQuantized::validate(), CLLSTMLayerQuantized::validate(), NEQLSTMLayer::validate(), arm_compute::validate(), CLQLSTMLayer::validate(), arm_compute::vdequantize(), arm_compute::vquantize(), arm_compute::vquantize_qasymm16(), arm_compute::vquantize_qsymm16(), and arm_compute::vquantize_signed().


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