Path:
Home >
GB/T >
Page225 > GB/T 46069.2-2025
| Std ID | Version | USD | Buy | Deliver [PDF] in | Title (Description) |
| GB/T 46069.2-2025 | English | RFQ |
ASK
|
3 days [Need to translate]
|
Artificial intelligence - Operator interface - Part 2: Neural network classes
|
Click to Preview a similar PDF
Basic data
| Standard ID | GB/T 46069.2-2025 (GB/T46069.2-2025) |
| Description (Translated English) | Artificial intelligence - Operator interface - Part 2: Neural network classes |
| Sector / Industry | National Standard (Recommended) |
| Classification of Chinese Standard | L70 |
| Classification of International Standard | 35.020 |
| Word Count Estimation | 214,248 |
| Date of Issue | 2025-08-29 |
| Date of Implementation | 2026-03-01 |
| Issuing agency(ies) | State Administration for Market Regulation; Standardization Administration of China |
GB/T 46069.2-2025: Artificial intelligence - Operator interface - Part 2: Neural network classes
---This is an excerpt. Full copy of true-PDF in English version (including equations, symbols, images, flow-chart, tables, and figures etc.), auto-downloaded/delivered in 9 seconds, can be purchased online: https://www.ChineseStandard.net/PDF.aspx/GBT46069.2-2025
ICS 35.020
CCSL70
National Standards of the People's Republic of China
Artificial Intelligence Operator Interface Part 2.Neural Network Class
Published on 2025-08-29
Implemented on 2026-03-01
State Administration for Market Regulation
The State Administration for Standardization issued a statement.
Table of Contents
Preface III
Introduction IV
1.Scope 1
2 Normative References 1
3.Terms and Definitions 1
4.Abbreviations 1
5 General Rules 2
6 Data Structures 2
7 Neural Network Operator Interface 2
7.1 Interface List 2
7.2 Interface Operations and Parameters 3
7.3 Operator Interface Minimal Set 114
Appendix A (Informative) C Language Reference Definition Example 116 for Neural Network Operator Interfaces
A.1 Data Structures 116
A.2 Neural Network Operator Interface 116
References 207
Foreword
This document complies with the provisions of GB/T 1.1-2020 "Standardization Work Guidelines Part 1.Structure and Drafting Rules of Standardization Documents".
Drafting.
This document is Part 2 of GB/T 46069 "Artificial Intelligence Operator Interfaces". GB/T 46069 has already published the following parts.
---Part 1.Basic Mathematics;
---Part 2.Neural Networks
Please note that some content in this document may involve patents. The issuing organization of this document assumes no responsibility for identifying patents.
This document was proposed and is under the jurisdiction of the National Information Technology Standardization Technical Committee (SAC/TC28).
This document was drafted by. Peking University, Peking University Changsha Institute of Computing and Digital Economy, and China Electronics Technology Standardization Institute.
Changsha 1011 Technology Co., Ltd., Pengcheng Laboratory, Institute of Software, Chinese Academy of Sciences, Beijing Baidu Netcom Technology Co., Ltd., Huawei Technologies Co., Ltd.
Cambricon Technologies Corporation Limited, SenseTime Technology Co., Ltd., Shanghai SenseTime Intelligent Technology Co., Ltd., Zhongguancun Audiovisual Industry Technology Innovation Alliance
Alliance, Inspur Electronic Information Industry Co., Ltd., Shanghai Artificial Intelligence Innovation Center, Beijing University of Aeronautics and Astronautics, Harbin Institute of Technology,
Shanghai Suiyuan Technology Co., Ltd., Shanghai Biren Technology Co., Ltd., Henan Kunlun Technology Co., Ltd., and Super Fusion Digital Technology Co., Ltd.
Limited Liability Company, Beijing Jiaotong University.
The main drafters of this document are. Yang Chao, Chen Jun, Hu Xiaoguang, Gou Haipeng, Ma Yanjun, Fan Ruibo, Li Kesen, Duan Lian, Bao Wei, and Yang Yuze.
Jia Mengzhu, Li Min, Li Ziyi, Yu Dianhai, Zhang Chengxing, Yu Tian, Guan He, Hu Shuai, Zhao Haiying, Li Jianxin, Liu Xianglong, Yang Muyun, Gao Xiang, Ma Shanshan
Gao Tiezhu, Wu Geng, Ding Ruiquan, Jiang Hui, Xing Feng, Wang Li, Liu Aishan, Sun Peiyuan, Zheng Ruolin, Tang Yinan, Li Hui, Mei Jingqing, Wang Sishan, Zhong Pu,
Song Wenlin, Luan Xiaoxu, Li Xiaoru, Chen Kai, Liu Wenfeng, Chen Deliang, Wang Xinmin, Zhou Hongli, Lu Shun, Xiao Yisong, Shen Zhiyue, Wang Hao, Liu Jinnan
Gao Ge, Liu Wei, Nie Jiandi, Yang Zheng, Feng Haijun, and Wang Qunbo.
introduction
In recent years, my country's artificial intelligence industry has shown a prosperous development trend, with related software and hardware developing in a diversified manner, including cloud servers and edge computing.
With the proliferation of different types of processors for devices and terminals, various computing libraries, intermediate representation tools, and programming frameworks have also emerged.
A flourishing landscape. The abundance of AI software and hardware has greatly facilitated the efficient deployment of applications, but it has also brought about a diverse range of challenges.
The challenges of increasing complexity and fragmentation. On the one hand, AI software practitioners need to consider the interaction between their software and various mainstream AI processors.
Adaptation involves investing significant effort in improving software portability. On the other hand, the development of every AI hardware device necessitates fundamentally adapting to the needs of everyday users.
Artificial intelligence software provides support; otherwise, it's difficult to integrate into the existing software ecosystem. This M×N level software-hardware mapping relationship has gradually become...
This has become a major obstacle hindering the development of artificial intelligence applications.
Artificial intelligence operators are the fundamental computations for building artificial intelligence applications. They encapsulate related hardware operations, and artificial intelligence software calls these operators.
Sub-interfaces are used to utilize hardware resources to complete computations; operator interfaces serve as a bridge between artificial intelligence software and hardware. GB/T 46069, "Artificial Intelligence," is a standard standard for this field.
The "Artificial Intelligence Operator Interface" standardizes and normalizes the core data structures, functions, and interface parameters of artificial intelligence operators, aiming to reduce the cost of artificial intelligence...
This involves fundamental work that addresses the challenges of software and hardware compatibility and promotes ecological development.
GB/T 46069 "Artificial Intelligence Operator Interface" is proposed to consist of five parts.
---Part 1.Fundamental Mathematics. The aim is to establish the general principles and core data structures applicable to artificial intelligence operator interfaces, and...
Standardize the basic functions and parameter requirements of the interface for fundamental mathematical operators.
---Part 2.Neural Networks. The purpose is to standardize the basic functions and parameter requirements of neural network operators.
---Part 3.Machine Learning Classes. The purpose is to standardize the basic functionalities and parameter requirements of machine learning operators.
---Part 4.Large Model Operators. The purpose is to standardize the basic functions and parameter requirements of large model operators.
---Part 5.Automated Testing Framework. The purpose is to provide automated testing methods and reference implementations for operator interfaces, verifying the operators.
Development standards compliance.
Artificial Intelligence Operator Interface Part 2.Neural Network Class
1.Scope
This document specifies the basic functions and parameter requirements of neural network operator interfaces for the field of artificial intelligence.
This document applies to the design, development, and application of artificial intelligence neural network operator libraries, as well as the development of related software, hardware, and systems.
2 Normative references
The contents of the following documents, through normative references within the text, constitute essential provisions of this document. Dated citations are not included.
For references to documents, only the version corresponding to that date applies to this document; for undated references, the latest version (including all amendments) applies.
This document.
GB/T 41867-2022 Terminology for Information Technology and Artificial Intelligence
GB/T 46069.1-2025 Artificial Intelligence Operator Interfaces Part 1.Basic Mathematical Categories
3.Terms and Definitions
The terms and definitions defined in GB/T 41867-2022 and GB/T 46069.1-2025, as well as the following terms and definitions, apply to this document.
3.1
neural network
A network consisting of one or more layers of neurons receives input data and produces output through weighted connections with adjustable weights.
[Source. GB/T 41867-2022, 3.2.26]
3.2
neural network model
An abstract model that is simulated by software or implemented as a neural computer.
3.3
An artificial neural network with a tree-like hierarchical structure, in which network nodes recursively process input information according to their connection order.
3.4
The process of using a trained deep neural network or probabilistic statistical model to process data and obtain prediction results.
3.5
overfitting
The created model is unable to generalize to new data because it has learned parts of the training data that are irrelevant to the task.
4.Abbreviations
The following abbreviations apply to this document.
...
