大型语言模型及其在建筑行业应用研究综述

Large Language Model and its Applications in the field of Architecture:A Survey

近年来,以ChatGPT和GPT-4为代表的大型语言模型,在技术上出现了快速的进步和迭代,已经成为人工智能领域具有变革性的技术。大型语言模型在数据信息容量、模型参数量、底层模型结构、模型训练方法上都较之前的语言模型取得了关键突破,在自然语言处理、机器视觉等任务上乃至通用任务领域的表现都在持续提升,包括大型语言模型显示出的涌现能力。概述了大型语言模型的技术演进、技术架构、关键技术、主要特点,介绍了大模型的基础架构及核心原理,分享了大模型在建筑领域的应用,讨论了其局限性以及未来发展方向,旨在推动以大语言模型为代表的人工智能技术在建筑领域的应用与发展。

In recent years,large language models represented by ChatGPT and GPT-4 have made rapid technological progress and iteration,becoming the most revolutionary technology in the field of Artificial Intelligence.Large language models have made key breakthroughs in data information capacity,model parameter quantity,underlying model structure,and model training methods compared to previous language models.Their performance in tasks such as natural language processing,machine vision,and even general tasks continues to improve,including the emergence ability demonstrated by large language models.An overview of the technological evolution,architecture,key technologies,and main characteristics of large language models are provided in the paper.The basic architecture and core principles of large-scale models are introduced,their applications in the field of architecture are shared,their limitations,and future development directions are discussed.The aim is to promote the application and development of Artificial Intelligence technology represented by large language models in the field of architecture and civil engineering.