Yining Li

Research Scientist, Shanghai AI Lab

prof_pic_1.jpg

I am a research scientist at Shanghai Artificial Intelligence Laboratory, and an adjunct Ph.D. supervisor at Shanghai Jiao Tong University. I received my Ph.D. from Multimedia Lab (MMLab) at The Chinese University of Hong Kong in 2019, advised by Prof. Chen Change Loy and Prof. Xiaoou Tang. Before that, I earned my B.S. Degree from Tsinghua University in 2014. Prior to joining Shanghai AI Lab, I was a Senior Research Scientist at SenseTime from 2019 to 2021.

I work on large language models and agentic systems. I am a member of the Intern Large Models team, where we develop foundation models and agentic systems to accelerate scientific discovery. With a background in computer vision, I was a core member of OpenMMLab and led the development of MMPose.

Hiring We have openings for research interns and full-time researchers in LLM training and infrastructure. I am also looking for Ph.D. students (in collaboration with SJTU). Feel free to email me if you are interested.

News

Aug 22, 2026 DataChef is accepted to EMNLP 2026 Findings.
Apr 15, 2026 TREX technical report is available. We introduce TREX, a tree-search enhanced agentic system that automates the challenging process of LLM fine-tuning, along with a new benchmark, FT-Bench, comprising 10 fine-tuning tasks to evaluate automatic research systems. Check out the project page for more information.
Apr 07, 2026 RouteMoA is accepted to ACL 2026.
Mar 30, 2026 Kernel-Smith technical report is online.
Dec 12, 2025 MG-LLaVA is accepted to TCSVT.
Feb 26, 2025 Auto Cherry-Picker is accepted to CVPR 2025.
Feb 16, 2025 MIG is accepted to ACL 2026 Findings.
Jan 23, 2025 RMP-SAM is accepted to ICLR 2025 as an oral presentation.
Jan 15, 2025 InternLM3-8B-Instruct is released, supporting both a normal response mode for general purpose and a deep thinking mode for solving complicated reasoning tasks via long CoT.
Sep 26, 2024 5 papers accepted to NeurIPS 2024, 3 in the main track (MotionBooth, ADC, XComposer2-4KHD) and 2 in the Datasets and Benchmarks track (GTA, MMBench-Video).
Jul 11, 2024 We released RTMW, the newest addition to RTMPose series, which specializes in predicting whole-body 2D and 3D keypoints simultaneously in real time.
Jul 01, 2024 Open-Vocabulary SAM is accepted to ECCV 2024.
May 26, 2024 InternLM2 technical report is online.
Feb 27, 2024 3 papers accepted to CVPR 2024: RTMO, OMG-Seg and ROVI.
Dec 08, 2023 We introduce AgentLego, a modular tool library to equip LLM agents with composable, multi-modal capabilities through standardized tool interfaces.

Selected Publications

  1. LLM/VLM
    GTA-2: Benchmarking General Tool Agents from Atomic Tool-Use to Open-Ended Workflows
    Jize Wang, Xuanxuan Liu, Yining Li, Songyang Zhang, Yijun Wang, and 5 more authors
    arXiv, 2026
    arXiv Code
  2. Agent
    RouteMoA: Dynamic Routing without Pre-Inference Boosts Efficient Mixture-of-Agents
    Jize Wang, Han Wu, Zhiyuan You, Yiming Song, Yijun Wang, and 7 more authors
    ACL, 2026
    arXiv HTML Code
  3. Agent
    TREX: Automating LLM Fine-tuning via Agent-Driven Tree-based Exploration
    Zerun Ma, Guoqiang Wang, Xinchen Xie, Yicheng Chen, He Du, and 5 more authors
    arXiv, 2026
    arXiv HTML Website
  4. LLM/VLM
    DataChef: Cooking Up Optimal Data Recipes for LLM Adaptation via Reinforcement Learning
    Yicheng Chen, Zerun Ma, Xinchen Xie, Yining Li, and Kai Chen
    In Findings of EMNLP, 2026
    arXiv HTML
  5. LLM/VLM
    MIG: Automatic Data Selection for Instruction Tuning by Maximizing Information Gain in Semantic Space
    Yicheng Chen, Yining Li, Kai Hu, Zerun Ma, Haochen Ye, and 1 more author
    In Findings of ACL, 2025
    DOI arXiv HTML Code Website
  6. Vision & Multimodality
    Auto Cherry-Picker: Learning from High-quality Generative Data Driven by Language
    Yicheng Chen, Xiangtai Li, Yining Li, Yanhong Zeng, Jianzong Wu, and 2 more authors
    In CVPR, 2025
    arXiv HTML Video Code Website
  7. Vision & Multimodality
    MotionBooth: Motion-Aware Customized Text-to-Video Generation
    Jianzong Wu, Xiangtai Li, Yanhong Zeng, Jiangning Zhang, Qianyu Zhou, and 3 more authors
    In NeurIPS Spotlight, 2024
    arXiv HTML Video Code Website
  8. LLM/VLM
    InternLM2 Technical Report
    Zhaowei Cai, Ming Cao, Hao Chen, Kai Chen, Kaibo Chen, and 4 more authors
    arXiv, 2024
    arXiv HTML Code Website
  9. LLM/VLM
    InternLM-XComposer2: Mastering Free-Form Text-Image Composition and Comprehension in Vision-Language Large Model
    Xiaoyi Dong, Pan Zhang, Yuhang Zang, Yan Cao, Boxiao Wang, and 4 more authors
    arXiv, 2024
    arXiv HTML Code Website
  10. Agent
    GTA: A Benchmark for General Tool Agents
    Jize Wang, Zerun Ma, Yining Li, Songyang Zhang, Cailian Chen, and 2 more authors
    In NeurIPS Datasets and Benchmarks Track, 2024
    arXiv HTML Code Website
  11. Vision & Multimodality
    RTMO: Towards High-Performance One-Stage Real-Time Multi-Person Pose Estimation
    Peng Lu, Tao Jiang, Yining Li, Xiangtai Li, Kai Chen, and 1 more author
    In CVPR, 2024
    arXiv HTML Code