| 问答ChatGPT之后:超大预训练模型的机遇和挑战 |
| |
| 引用本文： | 卢经纬,郭超,戴星原,缪青海,王兴霞,杨静,王飞跃.问答ChatGPT之后:超大预训练模型的机遇和挑战[J].自动化学报,2023,49(4):705-717. |
| |
| 作者姓名： | 卢经纬 郭超 戴星原 缪青海 王兴霞 杨静 王飞跃 |
| |
| 作者单位： | 1.中国科学院自动化研究所复杂系统管理与控制国家重点实验室 北京 100190 |
| |
| 基金项目： | 国家自然科学基金 (U1811463), 行动元联合研究项目: 伺服驱动系统的基础建模和平行驱控研究资助 |
| |
| 摘 要： | 超大预训练模型(Pre-trained model, PTM)是人工智能领域近年来迅速崛起的研究方向,在自然语言处理(Natural language processing, NLP)和计算机视觉等多种任务中达到了有史以来的最佳性能,促进了人工智能生成内容(Artificial intelligence-generated content, AIGC)的发展和落地. ChatGPT作为当下最火热的PTM,更是以优异的表现获得各界的广泛关注.本文围绕ChatGPT展开.首先概括PTM的基本思想并对其发展历程进行梳理;接着,详细探讨ChatGPT的技术细节,并以平行智能的视角阐述ChatGPT;最后,从技术、范式以及应用等多个方面对PTM的发展趋势进行展望.
|
| 关 键 词： | 预训练模型 ChatGPT Transformer 人工智能生成内容 平行智能 社会化大闭环 |
| 收稿时间： | 2023-03-05 |
The ChatGPT After: Opportunities and Challenges of Very Large Scale Pre-trained Models |
| |
| Affiliation: | 1.The State Key Laboratory for Management and Control of Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing 1001902.Qingdao Academy of Intelligent Industries, Qingdao 2661143.School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049 |
| |
| Abstract: | In recent years, very large scale pre-trained models (PTMs) have become a rapidly rising research direction in artificial intelligence, achieving state of the art in most tasks, especially natural language processing (NLP) and computer vision, and speeding up the development and implementation of artificial intelligence-generated content (AIGC). ChatGPT, as the hottest PTM, has been brought to the fore on account of its excellent performance. This paper is organized around ChatGPT. First, we outline the basic idea of PTM and review its development history. Then, the technical details of ChatGPT are explored, and ChatGPT is revisited from the perspective of parallel intelligence. Finally, the development trends of PTMs are presented in terms of technologies, paradigms, and applications. |
| |
| Keywords: | |
|
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
|
