摘要
随着人工智能技术的发展,文生图AI模型的出现为设计领域带来了新的设计方式。对于建筑设计领域而言,主流的文生图AI模型由于算法、训练数据、开发程度等方面的不同,对于建筑设计任务的完成程度存在显著差异,这种差异产生的一个主要原因体现在相关模型对于提示词的解释能力。文中在应用最新的预训练模型的前提下,通过主观评价方法比较了三款主流的文生图工具对建筑设计领域提示词的解释能力,并提出了一个利用训练LORA模型来提升文生图工具对建筑设计领域提示词理解能力的方法。
With the development of artificial intelligence technology,the emergence of Vincennes AI models has brought a new design approach to the design field.For the architectural design field,the mainstream Venn Diagram AI models should have significant differences in accomplishing the architectural design tasks due to the differences in algorithms,training data,and degree of development,etc..One of the main reasons for such differences is reflected in the interpretation ability of the relevant models for the cue words.In this paper,under the premise of using the latest pre-trained models,we compare the interpretation ability of three mainstream Venn diagram tools for cue words in the field of architectural design through subjective evaluation methods,and propose a method to improve the understanding ability of Venn diagram tools for cue words in the field of architectural design by utilizing the training of LORA models.
作者
赵文瀚
李涛
Zhao Wenhan;Li Tao(School of Architecture,Xi’an University of Architecture and Technology,Xi’an Shaanxi 710000,China)
出处
《山西建筑》
2026年第8期38-41,共4页
Shanxi Architecture
