Research Center

The Center for AI Large Foundation Models

The Center for AI Large Foundation Models focuses on advancing fundamental and applied research, breaking through key theories and core technologies of large models, and driving the deep application of generative AI in fields such as healthcare, law, and intelligent devices. The center places high importance on the study of foundational theories and algorithms in AI, delving into areas such as the mathematical theory of large models, optimization algorithm design and analysis, and machine learning-assisted algorithms, aiming to provide solid theoretical support for ongoing advancements in AI and its downstream applications. The center’s self-developed HuatuoGPT has successfully achieved intelligent and precise medical services, making it the first Chinese medical foundation model in the country to integrate both academic and industrial applications. Additionally, it has developed an intelligent assessment system in the judicial field, using high-dimensional data fusion and intelligent algorithms to achieve real-time data integration and accurate assessments across departments, enhancing the level of intelligence in building a law-based society. Furthermore, the center has created a globally leading Arabic generative pre-training model and automatic speech recognition technology, supporting international collaboration under the “Belt and Road” framework. The center will continue to explore AI technology, actively promoting its widespread application and industrialization, contributing to the intelligent and sustainable development of society and supporting the economic and social growth of the Guangdong-Hong Kong-Macao Greater Bay Area.

Our Team


The Center for AI Large Foundation Models has attracted a number of internationally renowned talents with extensive research and engineering development experience, including academicians and scientists ranked among the top 2% globally. During the previous construction phase, the team successfully launched models such as Hua Tuo GPT and Arabic GPT. Additionally, it collaborated with organizations such as the National Health Commission, Shenzhen Municipal Justice Bureau, Huawei, Tencent, and Alibaba to tackle key scientific and technological challenges. Looking ahead, the team will focus on foundational theories and algorithms of large models, as well as their applications. Key areas of focus will include medical large models and their applications, legal large models and their applications, and Arabic large models and their applications.

Key Research

Large models have achieved human-level or even superior discriminative and generative capabilities in various fields such as computer vision and natural language processing. However, under the current scale of models, the continuous growth of data and computational demands, along with the maintenance and updating of models, has become a significant bottleneck constraining the sustained development of large models. To address this, the center has conducted a series of research initiatives focused on large model reduction theories and optimization theories to enhance the sustainability and scalability of large models, while maintaining model performance and knowledge accumulation, thereby truly empowering various industries.

Large models are not only advancing the field of artificial intelligence but also demonstrating significant application potential across various vertical industries. In this context, the center has undertaken a series of research initiatives focused on smart healthcare, intelligent legal systems, and multilingual model development. Notable projects include the establishment of a multimodal, trustworthy medical model platform—Hua Tuo GPT—centered on domestically developed technologies, the development of a model for generating administrative law enforcement documents and a model for supervising administrative enforcement, as well as the creation of AceGPT, the first localized Arabic large model, all aimed at promoting the widespread application and development of large models across diverse fields and contributing to global technological advancement.

Project & Service

AceGPT
In 2023, Shenzhen Research Institute of Big Data, in collaboration with The Chinese University of Hong Kong (Shenzhen) and King Abdullah University of Science and Technology in Saudi Arabia, released AceGPT—an open-source large language model specifically designed for the characteristics of the Arabic language, featuring leading performance.
HuatuoGPT
HuatuoGPT project focuses on using artificial intelligence technology to address challenges in traditional medical processes, such as patients choosing the wrong appointment, long wait times, and suboptimal healthcare experiences.

Project Cooperation


Contact

  • Email:xushuizhen@sribd.cn
  • Tel:(+86)0755-23517610

Address

  • 2001 Longxiang Avenue, Longgang District, Shenzhen