WebSight
收藏资源简介:
WebSight是由Hugging Face创建的一个综合合成数据集,包含200万对HTML代码及其对应的网页截图。该数据集通过使用大型语言模型生成多样化的网站主题和设计,确保了数据的高质量和多样性。创建过程中,首先使用小型语言模型生成网站概念,然后通过大型语言模型生成最终的HTML代码。WebSight的应用领域主要集中在自动化网页开发,旨在通过AI技术加速UI开发任务,推动无代码解决方案的发展。
WebSight is a comprehensive synthetic dataset created by Hugging Face, containing 2 million pairs of HTML code and their corresponding webpage screenshots. This dataset leverages large language models (LLMs) to generate diverse website themes and designs, ensuring high data quality and diversity. During its development, small language models are first employed to generate website concepts, followed by large language models to produce the final HTML code. The primary application scenarios of WebSight focus on automated web development, aiming to accelerate UI development tasks via AI technologies and promote the advancement of no-code solutions.

- 1Unlocking the conversion of Web Screenshots into HTML Code with the WebSight DatasetHugging Face · 2024年



