1# OH_NN_QuantParam 2<!--Kit: Neural Network Runtime Kit--> 3<!--Subsystem: AI--> 4<!--Owner: @GbuzhidaoR--> 5<!--Designer: @GbuzhidaoR--> 6<!--Tester: @GbuzhidaoR--> 7<!--Adviser: @ge-yafang--> 8 9## Overview 10 11Defines the quantization information. 12 13In quantization scenarios, the 32-bit floating-point data type is quantized into the fixed-point data type according to the following formula: 14 15 16 17where, **s** and **z** are quantization parameters, which are stored by **scale** and **zeroPoint** in OH_NN_QuanParam. **r** is a floating point number, **q** is the quantization result, **q_min** is the lower bound of the quantization result, and **q_max** is the upper bound of the quantization result. The calculation method is as follows: 18 19 20 21 22 23The **clamp** function is defined as follows: 24 25 26 27**Since**: 9 28 29**Deprecated from**: 11 30 31**Substitute**: [NN_QuantParam](capi-neuralnetworkruntime-nn-quantparam.md) 32 33**Related module**: [NeuralNetworkRuntime](capi-neuralnetworkruntime.md) 34 35**Header file**: [neural_network_runtime_type.h](capi-neural-network-runtime-type-h.md) 36 37## Summary 38 39### Member Variables 40 41| Name| Description| 42| -- | -- | 43| uint32_t quantCount | Length of the numBits, scale, and zeroPoint arrays. In the per-layer quantization scenario, **quantCount** is usually set to **1**. That is, all channels of a tensor share a set of quantization parameters. In the per-channel quantization scenario, **quantCount** is usually the same as the number of tensor channels, and each channel uses its own quantization parameters.| 44| const uint32_t *numBits | Number of quantization bits.| 45| const double *scale | Pointer to the scale data in the quantization formula.| 46| const int32_t *zeroPoint | Pointer to the zero point data in the quantization formula.| 47
