I study training models at OpenAI. I received my Ph.D. from Princeton, advised by Chi Jin. My PhD research focused on fundamental theories in reinforcement learning (RL), including partially observable RL (POMDP, PSR, Survey), multi-agent RL (Stochastic Game), and RL with large state spaces (Function Approximation). After that, I spent a wonderful year as a postdoctoral researcher at Microsoft Research NYC, where I explored language model research. During summer 2022, I interned at DeepMind, working with Csaba Szepesvári and Gellért Weisz. Previously, I received a B.E. degree in Electrical Engineering and a B.S. degree in Mathematics from Tsinghua University. [Google Scholar]

Models

I've co-trained GPT-5–series models (e.g. GPT-5, GPT-5.1) for multi-personality / personalization. Highlighted in

• Introducing multi-personality in GPT 5

• Making ChatGPT uniquely yours in GPT 5.1

• Moving beyond one-size-fits-all by Fidji

Selected Papers   (Show All)

(α-β order) denotes alphabetical authorship ordering

• Optimistic MLE – A Generic Model-based Algorithm for Partially Observable Sequential Decision Making
  Qinghua Liu, Praneeth Netrapalli, Csaba Szepesvári, Chi Jin
  Symposium on Theory of Computing (STOC), 2023

• V-Learning – A Simple, Efficient, Decentralized Algorithm for Multiagent RL
  (α-β order) Chi Jin, Qinghua Liu, Yuanhao Wang, Tiancheng Yu
  Mathematics of Operations Research (MOR), 2023
  Best Paper in ICLR 2022 ‘‘Gamification and Multiagent Solutions’’ Workshop
  

• When Is Partially Observable Reinforcement Learning Not Scary?
  Qinghua Liu, Alan Chung, Csaba Szepesvári, Chi Jin
  Conference on Learning Theory (COLT), 2022

• Bellman Eluder Dimension: New Rich Classes of RL Problems, and Sample-Efficient Algorithms
  (α-β order) Chi Jin, Qinghua Liu, Sobhan Miryoosefi
  Neural Information Processing Systems (NeurIPS), 2021 (Spotlight)