Yifan Song 宋一帆

I am a final-year PhD candidate at Peking University, advised by Prof. Sujian Li. My research focuses on LLM-based agents — how to make language models plan, act, and learn from interaction with the real world.

Since December 2024, I have been a research intern at Xiaomi's LLM Core team, where I am a core contributor to the MiMo series of models, including MiMo-7B, MiMo-VL, MiMo-V2-Flash, and MiMo-V2-Pro.

Find me on Google Scholar, GitHub, and X/Twitter.

• RestGPT: Connecting Large Language Models with Real-World RESTful APIs Yifan Song, Weimin Xiong, Dawei Zhu, Wenhao Wu, Han Qian, Mingbo Song, Hailiang Huang, Cheng Li, Ke Wang, et al. arXiv 2023 · Cited by 201
• Trial and Error: Exploration-Based Trajectory Optimization for LLM Agents Yifan Song, Da Yin, Xiang Yue, Jie Huang, Sujian Li, Bill Yuchen Lin ACL 2024 · Cited by 190
• Calibrating Factual Knowledge in Pretrained Language Models Qingxiu Dong, Damai Dai, Yifan Song, Jingjing Xu, Zhifang Sui, Lei Li EMNLP 2022 Findings · Cited by 178
• R1-V: Reinforcing Super Generalization Ability in Vision-Language Models Liang Chen, Liangtao Li, Hao Zhao, Yifan Song, et al. 2025 · Cited by 143
• The Good, The Bad, and The Greedy: Evaluation of LLMs Should Not Ignore Non-Determinism Yifan Song, Guoyin Wang, Sujian Li, Bill Yuchen Lin NAACL 2025 · Cited by 142
• PoSE: Efficient Context Window Extension of LLMs via Positional Skip-wisE Training Dawei Zhu, Nan Yang, Liang Wang, Yifan Song, Wenhao Wu, Furu Wei, Sujian Li ICLR 2024 · Cited by 116
• MiMo: Unlocking the Reasoning Potential of Language Model — From Pretraining to Posttraining Xiaomi LLM, Yifan Song (core contributor), et al. arXiv 2025 · Cited by 57
• MiMo-V2-Flash Technical Report Xiaomi LLM, Yifan Song (core contributor), et al. arXiv 2026 · Cited by 18
• MiMo-VL Technical Report Zhiyang Yue, Zhaoyang Lin, Yifan Song, et al. arXiv 2025 · Cited by 15
• VisualWebBench: How Far Have Multimodal LLMs Evolved in Web Page Understanding and Grounding? Junpeng Liu, Yifan Song, Bill Yuchen Lin, Wai Lam, Graham Neubig, Yuanzhi Li, Xiang Yue COLM 2024 · Cited by 91
• Watch Every Step! LLM Agent Learning via Iterative Step-Level Process Refinement Weimin Xiong, Yifan Song, Xiutian Zhao, Wenhao Wu, Xun Wang, Ke Wang, Cheng Li, Wei Peng, Sujian Li EMNLP 2024 · Cited by 70
• AgentBank: Towards Generalized LLM Agents via Fine-Tuning on 50000+ Interaction Trajectories Yifan Song, Weimin Xiong, Xiutian Zhao, Dawei Zhu, Wenhao Wu, Ke Wang, Cheng Li, Wei Peng, Sujian Li EMNLP 2024 Findings · Cited by 40

Full list on Google Scholar.