On September 1, the Institute of Automation, Chinese Academy of Sciences (CAS) announced to the media that the world's first three-modal large-scale model "Zidong Taichu", jointly developed by the institute and Huawei, opened in Shanghai on the same day in 2022. At the World Artificial Intelligence Conference, he won the "Superior AI Leader" (SAIL, Superior AI Leader) award, the highest award of the conference.
The "Zidong Taichu" project jointly created by the Institute of Automation of the Chinese Academy of Sciences, Huawei and the Wuhan Institute of Artificial Intelligence based on the full-stack localized software and hardware platform "Shengteng AI" breaks through the limitations of current artificial intelligence technology and effectively solves the "one The pain points of "specificity", lack of small sample learning ability, and cross-modal semantic gap, through cross-modal multi-task self-supervised learning to achieve "unified representation" and "mutual generation" between image, text, and voice three-modal data, Constructing large-scale entity concepts and mutual relationship maps, understanding and generating capabilities closer to human beings, providing an innovative foundation for building multimodal artificial intelligence industry applications, thus taking an important step towards general artificial intelligence.
(Reporter Sun Zifa produced Wu Rui's video from the Institute of Automation, Chinese Academy of Sciences)
Responsible editor: [Ji Xiang]