Li Yuan
Doctoral Supervisor, Peking University
Li Yuan is a PhD supervisor at Peking University, an assistant professor at Shenzhen Graduate School of Peking University, a recipient of the National Excellent Youth (Overseas) Fund, and the chair of the National Science and Technology Major Project and the National Natural Youth Foundation of China, etc. His research interests include multimodal deep learning and AI4S. His research interests include multimodal deep learning and AI4S. His representative academic work includes VOLO, T2T-ViT and other deep neural network frameworks and knowledge distillation related work, and his first paper has been cited more than a thousand times, and his representative application work includes ChatExcel, ChatLaw, and other vertical domain language models.
Topic
Open-Sora Plan Video Generation Open Source Program - Progress and Shortcomings
Sora, a Vincentian video model, can generate high-quality video content, but the technology and model are not open-source to the public; this report focuses on the Open Sora Plan initiated by the Peking University-Rabbit Expo AIGC Joint Laboratory, aiming to gather the power of the open-source community to jointly reproduce a version of the Vincentian video model with the Sora-like architecture, and to develop the framework for all to use. The Open Sora Plan has been widely supported by the open source community and also by domestic AI computing power, and the first version of the model and algorithm details have been open-sourced. This report will introduce in detail the original intention of the open source project , the technical route , the current progress and shortcomings , as well as future plans . Link to the open source program project: https://github.com/PKU-YuanGroup/Open-Sora-Plan .
