1# OH_NN_QuantParam 2 3 4## Overview 5 6Defines the quantization information. 7 8In quantization scenarios, the 32-bit floating-point data type is quantized into the fixed-point data type according to the following formula: 9 10 11 12where, **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: 13 14 15 16 17 18The **clamp** function is defined as follows: 19 20 21 22**Since**: 9 23 24**Deprecated**: This module is deprecated since API version 11. 25 26**Substitute**: You are advised to use [NN_QuantParam](_neural_network_runtime.md#nn_quantparam). 27 28**Related module**: [NeuralNetworkRuntime](_neural_network_runtime.md) 29 30 31## Summary 32 33 34### Member Variables 35 36| Name| Description| 37| -------- | -------- | 38| [quantCount](#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.| 39| [numBits](#numbits) | Number of quantization bits.| 40| [scale](#scale) | Pointer to the scale data in the quantization formula.| 41| [zeroPoint](#zeropoint) | Pointer to the zero point data in the quantization formula.| 42 43 44## Member Variable Description 45 46 47### numBits 48 49``` 50const uint32_t* OH_NN_QuantParam::numBits 51``` 52 53**Description** 54 55Number of quantization bits. 56 57 58### quantCount 59 60``` 61uint32_t OH_NN_QuantParam::quantCount 62``` 63 64**Description** 65 66Length 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. 67 68 69### scale 70 71``` 72const double* OH_NN_QuantParam::scale 73``` 74 75**Description** 76 77Pointer to the scale data in the quantization formula. 78 79 80### zeroPoint 81 82``` 83const int32_t* OH_NN_QuantParam::zeroPoint 84``` 85 86**Description** 87 88Pointer to the zero point data in the quantization formula. 89
