PAIR Public Forum for Research and Innovation: Prof. YANG Hongxia of PolyU delivers "DeepSeek and Beyond"

Abstract

DeepSeek’s recently released model has demonstrated that very strong AI capabilities can be achieved without the need of particularly large models. Their AI models perform in a manner comparable to top AI models in the United States, but require much fewer computing resources. More importantly, they have made this technology sharable through open sourcing. This breakthrough has changed the way how people think about AI development, and has also triggered concerns over privacy security, technological competition and other issues. Our Co-Gen AI project aims to build on this idea to help enhance Hong Kong’s AI competitiveness. Our platform has three innovative features. The first one is Domain-Adaptive Continual Pretraining (DACP), which allows AI models to become more specialised through learning data from specific domains (e.g., industry, scientific research) and demonstrates better training outcomes that open-source models and ChatGPT. The second one is Advanced Model Fusion Infrastructure, which comprises a “Model Fusion” over existing specialised models, significantly reducing the use of computing resources—training a 7B model requires only 64–128 GPU cards, and a 100B large model requires just 512–1,024 cards, thus bringing over 90% resource saving. The third one is Resource-Efficient Architecture, which allows us to leverage ordinary computing resources in Cyberport, Science Park, Pengcheng Lab and other places to develop AI, thus enabling the efficient training of large models by integrating small models, and avoiding the problems in traditional approaches that require large clusters of identical high-end GPUs for training. Through this innovative approach, AI development can be made more accessible and less dependent on massive centralised computational resources. This project expects to help Hong Kong occupy an important position in the global generative AI development landscape.

Summary

The PolyU Academy for Interdisciplinary Research (PAIR) of The Hong Kong Polytechnic University (PolyU) today hosted its inaugural Public Forum for Research and Innovation. Titled “DeepSeek and Beyond”, the keynote speech was delivered by Prof. YANG Hongxia, Associate Dean (Global Engagement) of the PolyU Faculty of Computer and Mathematical Sciences and Professor of the Department of Computing, who highlighted the latest developments in artificial intelligence (AI). The event attracted over a thousand participants, including faculty members, students, alumni, and leaders from the innovation and technology sector, as well as academics and the public. Additionally, over 390,000 viewers tuned in through the live streaming platforms.

The Forum began with a welcoming speech delivered by Prof. CHEN Qingyan, Director of PAIR and Chair Professor of Building Thermal Science of the PolyU Department of Building Environment and Energy Engineering. This was followed by Prof. ZHANG Chenqi, Chair Professor of Artificial Intelligence of the PolyU Department of Data Science and Artificial Intelligence, and Director of the PolyU Shenzhen Research Institute introducing the speaker.

Prof. Zhang said, “The development of large models is at the core of competition in the AI wave. DeepSeek has demonstrated that high-performance AI models can be achieved using fewer and less advanced graphics processing units (GPUs), demonstrating that cutting-edge AI technology can be realised through the optimisation of algorithms.”

The large AI model developed by the mainland Chinese startup DeepSeek has garnered wide acclaim around the world for its low-cost, high-performance, and open-source framework, disrupting the traditional “computing power-first” logic of AI model training. At the Forum, Prof. Yang highlighted the potential of generative AI (GenAI), adding that it presents abundant opportunities for various sectors, including healthcare, finance, manufacturing, retail, media and fashion, and for applications in medical imaging analysis, fraud detection, predictive maintenance, retail inventory management, content creation, and design and marketing.

Prof. Yang also recounted the evolution of AI and shared her professional milestones with the audience, notably the development of the M6 large model, which trained a 10-trillion-parameters model using just 512 GPUs. Prof. Yang further elaborated on how her GenAI project, Co-GenAI, improves the accessibility of AI technology while minimising dependence on large-scale centralised computing resources, thereby transforming the trajectory of AI progress. This ground-breaking effort has positioned Hong Kong and the Mainland at the forefront of global advancement in GenAI.

Moderated by Prof. Zhang Chenqi, a panel discussion was also held, featuring esteemed panellists Prof. Yang Hongxia and Prof. LI Qing, Head and Chair Professor of Data Science of the PolyU Department of Computing, and Co-Director of the Research Centre for Digital Transformation of Tourism. The scholars discussed the opportunities and challenges that advancements in AI present for higher education and research. They also engaged in fruitful discussion with participants during the question-and-answer session. The topics included the application of AI in industry, the regulation of information, its impact on the employment environment and economic development, and the integration of AI technologies.

PolyU is committed to advancing AI education and research. In January 2025, the University established the Faculty of Computer and Mathematical Sciences with a vision to lead global advancements in digital transformation and AI through distinguished education, research, and knowledge transfer.

Please click here for an online review.