Renmin (RUC)
academicRenmin University's Gaoling School of Artificial Intelligence (GSAI, executive dean Ji-Rong Wen) is China's most focused academic LLM school — and one of very few university groups anywhere that pretrains from scratch: the 38-author YuLan-Base-12B (~1.7T tokens) and the data-efficient YuLan-Mini (2.42B / 1.08T tokens, fully open recipe with 24 intermediate checkpoints).
Two engines drive the catalog. Wayne Xin Zhao's RUCAIBox wrote the earliest open o1/R1 reproduction reports (the STILL slow-thinking series), R1-Searcher's agentic-RAG RL, the OlymMATH olympiad benchmark, and maintains the field-defining LLMSurvey (12.2K★, basis of the LLMBook). Zhicheng Dou's RUC-NLPIR owns the 2025–26 deep-search-agent wave — Search-o1 → WebThinker → ARPO → DeepAgent/Arbor — plus the FlashRAG standard RAG toolkit (3.5K★).
The work ships under three GitHub orgs (RUCAIBox, RUC-GSAI for YuLan, RUC-NLPIR) with weights under HF orgs RUC-AIBOX and yulan-team. Industry entanglement is collaboration-shaped rather than spinout-shaped: ByteDance Seed (Pass@k Training), Kuaishou (ARPO), Baichuan/BAAI (Virgo, HierSearch), and BAAI/DataCanvas compute credits on STILL.
People
- Ji-Rong Wen Google Scholar — Executive Dean, GSAI; Dean, School of Information (formerly Senior Researcher, Microsoft Research Asia (14 years))
- Wayne Xin Zhao Google Scholar — Professor, GSAI; RUCAIBox lead (YuLan, STILL, LLMSurvey); joint work with ByteDance Seed, Baichuan, BAAI
- Zhicheng Dou Google Scholar — Professor & Vice Dean, GSAI; RUC-NLPIR lead (Search-o1, WebThinker, FlashRAG) (formerly Microsoft Research Asia)
- Rui Yan Google Scholar — Professor, GSAI; Aloha Lab (dialogue & LLM agents) (formerly Peking University; Baidu NLP)
- Xu Chen Google Scholar — Associate Professor, GSAI; agent-simulation line (YuLan-OneSim, RecAgent)