On the afternoon of December 9, South China University of Technology appointed Tian Qi, the chief scientist of HUAWEI CLOUD's artificial intelligence field, as a visiting professor and he later gave an academic lecture at A4-204, University City Campus. More than 200 people including Li Weiqing, vice president of the school, relevant principals of the Graduate School, the School of Journalism and Communication, the School of Computer Science and Engineering, and the School of Electronics and Information, as well as representatives of teachers and students attended the event. The event was hosted by Su Hongyuan, dean of the School of Journalism and Communication.

At the beginning of the event, Vice President Li Weiqing presented Tian Qi with a certificate of appointment as a visiting professor, and extended a warm welcome to Professor Tian Qi's arrival. She pointed out that Professor Tian Qi joining South China University of Technology, which is known for its engineering, will surely promote the rapid development of related disciplines in the university.

After the appointment ceremony, Professor Tian Qi gave an academic lecture entitled Pangu Pre-training Large Model. He cited the current status of artificial intelligence development and pointed out that workshop-style AI development cannot meet the needs of efficient and high-quality digital transformation in the industry. However, the Pangu pre-training model (hereinafter referred to as the Pangu model) released by Huawei Cloud in the first half of this year is an AI-driven model. Here comes a new model of industrial development. The Pangu model is an attempt to move towards AI industrial development from the customization and workshop-style opening of the past.

Tian Qi introduced that the Pangu model includes four major types: NLP model, CV model, multimodal model and scientific computing model. He gave a simple introduction to each type of model combined with practical application cases.

Later, he proposed that there are three core design principles guiding the Pangu model. One is a large network scale. The second is a large model with a strong network architecture. The third is excellent generalization ability and high scene coverage. Not only the individual 100m champion, but also the 10 all-around champion, said Professor Tian Qi.

Professor Tian Qi said that the Pangu model has not been standardized yet, considering the problems of energy consumption and ethical and regulatory risks in the Pangu model. He hopes that at the current stage when thousands of people are competing and thousands of thoughts are proposed, students can actively participate in it and jointly make a large AI model belonging to China.

During the question and answer session, the students asked Professor Tian Qi about the content of the lecture, such as interdisciplinary integration and multi-language model design. Professor Tian Qi answered the questions in detail one by one. He said that the Pangu model has different content at different training stages, and interdisciplinary talents are more useful. It is still in the process of continuous development, and issues such as language and culture have been paid attention to and will be improved in the future.

Professor Tian Qi is an expert in computer vision and multimedia information retrieval, an IEEE Fellow, an academician of the International Eurasian Academy of Sciences, an overseas outstanding young man of the National Natural Science Foundation of China, an overseas review expert of the Chinese Academy of Sciences, and serves as a chair professor or a visiting professor in many well-known domestic universities and research institutes.