How to build your own large model


The Faculty of Data Science of City University of Macau successfully held an academic exchange workshop on the theme of "How to Build Your Own Large Model" on April 12, 2024, and invited two doctors, Wenhan Chang and Hanjin Tong from China University of Geosciences provide a seminar on learning about the development of large-scale language model fine-tuning technology, as well as the current status analysis and future prospects of large-scale language models. 

 

Dr. Wenhan Chang focused on the underlying technology architecture, intermediate development and future prospects of large model fine-tuning. The paper explains the widespread application of large language models in current society and emphasizes that due to limitations in computing resources and time costs, it has become increasingly important to fine-tune these large models for downstream tasks. Through Dr. Chang's explanation, students became more familiar with the technology of fine-tuning large models and gained a deeper understanding of how to apply these technologies in scientific tasks or industrial projects. 

 

Dr. Hanjin Tong focused on the graph large language model, which is a new model that combines graph learning technology and large language models. He introduced the basic concepts of Graph Neural Networks (GNNS) in detail, and analyzed how this technology can effectively display complex relationships between entities, as well as its important application value in the era of big data. In addition, he also discussed the ability of graph language models in capturing topological information in graph structures, and how this model can be used to improve the performance of downstream tasks. His report provided students with an in-depth understanding of this emerging technology and provided useful references and demonstrations for its future research and development directions. 

 

The successful holding of this academic exchange workshop "How to Build Your Own Large Model" provided students with a platform to gain an in-depth understanding of large model fine-tuning technology and inspired thinking about the future development of the field of artificial intelligence. The Faculty of Data Science of City University of Macau will continue to be committed to promoting technological innovation and academic exchanges and building bridges between academia and industry.