Xidong Feng

Computer Science, University of College London

xidong.jpg

I am a fourth-year Ph.D. student at Computer Science, University College London, advised by Prof. Jun Wang. I am also a student researcher in Google DeepMind working on generative models. Previously I did my undergraduate at the Department of Automation, Tsinghua University.

My research interests lie in Large Language Model, Single-agent/Multi-agent and Meta Reinforcement Learning.

news

Jan 30, 2024 Happy to announce that I will join Google DeepMind as a student Researcher in March!
Dec 1, 2023 Two of my cofirst-authored papers have been accecpted in JMLR, including TorchOpt and Heterogeneous-Agent Reinforcement Learning, check them out!
Oct 31, 2023 Our new work Alphazero-like Tree-Search can Guide Large Language Model Decoding and Training has been accepted by the NeurIPS 2023 FMDM workshop, check our codebase. In this paper, we systematically discuss how we can do AlphaZero-like tree search to enhance LLM inference and training.
Oct 1, 2023 ChessGPT - Bridging Policy Learning and Language Modeling have been accepted by NeurIPS 2023 Dataset and Benchmark Track! Also check our open-source code, model, and data.

selected publications

2023

  1. Neurips FMDM
    Alphazero-like Tree-Search can Guide Large Language Model Decoding and Training
    Xidong Feng*, Ziyu Wan*, Muning Wen, Ying Wen, Weinan Zhang, and Jun Wang
    arXiv preprint arXiv:2309.17179, 2023
  2. JMLR
    Heterogeneous-Agent Reinforcement Learning
    Yifan Zhong*, Jakub Grudzien*, Xidong Feng*, Siyi Ji, Jiaming Ji, and Yaodong Yang
    Journal of Machine Learning Research, 2023
  3. JMLR
    TorchOpt: An Efficient Library for Differentiable Optimization
    Jie Ren*, Xidong Feng*, Bo Liu*, Xuehai Pan*, Yao Fu, Luo Mai, and Yaodong Yang
    Journal of Machine Learning Research, 2023
  4. NeurIPS 2023
    ChessGPT: Bridging Policy Learning and Language Modeling
    Xidong Feng, Yicheng Luo, Ziyan Wang, Hongrui Tang, Mengyue Yang, Kun Shao, David Mguni, and 2 more authors
    Advances in Neural Information Processing Systems, 2023

2022

  1. NeurIPS
    A Theoretical Understanding of Gradient Bias in Meta-Reinforcement Learning
    Xidong Feng*, Bo Liu*, Jie Ren, Luo Mai, Rui Zhu, Haifeng Zhang, Jun Wang, and 1 more author
    In NeurIPS, 2022

2021

  1. NeurIPS
    Neural auto-curricula in two-player zero-sum games
    Xidong Feng*, Oliver Slumbers*, Ziyu Wan, Bo Liu, Stephen McAleer, Ying Wen, Jun Wang, and 1 more author
    Advances in Neural Information Processing Systems, 2021