Scholar

Yequan Wang 王业全

Google Scholar ID: 7Gqp6FsAAAAJ

Beijing Academy of Artificial Intelligence

Large ModelEmboddied AGINLP

Citations & Impact

All-time

Citations

3,445

H-index

i10-index

Publications

Co-authors

list available

Contact

Publications

20 items

Browse publications on Google Scholar (top-right) ↗

Resume (English only)

Academic Achievements

Honorable Mention, AI 2000 Most Influential Scholars in NLP (2022)
Google Scholar citations: 4,000+
Key publications include:
- 'Not All Layers of LLMs Are Necessary During Inference': Proposed AdaInfer, reducing inference computation by up to 43% with <1% performance loss
- 'Few-Shot Learner Generalizes Across AI-Generated Image Detection': Introduced FSD, achieving +11.6% accuracy without retraining
- '52B to 1T: Lessons Learned via Tele-FLM Series': Insights on scaling large language models
PI of NSFC project 'Implicit Sentiment Analysis on Complicated Web Text' (Grant No. 62106249)

Research Experience

Currently a Researcher and Team Leader at BAAI and PKU
Principal Investigator (PI) of the National Key R&D Program 'Next-Generation Artificial Intelligence' (Nov 2022 – Nov 2025)
Leads development of the FLM family of large models (FLM-2, FLM-101B, FreeLM), featuring growth techniques, loss prediction, and the FreeLM framework
Developed and open-sourced the Arabic Language Model (ALM 1.0)
Involved in the national project 'Artificial Intelligence Fundamental Model Support Platform and Evaluation Technology' to build an open-source ecosystem for foundational models

Co-authors

22 total