I am a first-year PhD student at Tsinghua University and a research intern at Tencent Hunyuan. I work on embodied agents driven by frontier models.
I don't keep count of my papers because they are an old currency. On this page they show up as evidence for a judgment I made at the time. What I care about is calling something early and putting real work behind the call, and when my call turns out wrong it stays here, struck through. I change topics when the technology moves, and I won't retell that as one long line of research, which is the polite fiction of our field. I believe that with the right people and the right taste, technology moves faster than any career plan can follow.
To go deep on any of this, whether to collaborate, consult or just chat, email me anytime.
“One finds limits by pushing them.”
Selected research topicseach with the bet I made going in
Embodied agents driven by frontier models
Bet, Oct 2026Embodied AI should be System 2-centric, with System 1 built around what System 2 needs, rather than the opposite.
One persistent GPT-6 conversation controls a dual-arm robot in RoboDojo. Controlled ablations show which parts of the harness actually move the score. I led the project.
Reinforcement learning for reasoning
Bet, Dec 2024RL is the key to reasoning. A clever method is a pleasure to build, but the results rest on verifiable rewards and a simple, reliable training stack.
Online RL for LLM reasoning with dense token-level rewards, taken from an implicit process reward model that needs outcome labels only. As a core contributor I built the training stack on verl and upstreamed the first parallel verification-reward code in the verl repo.
An 8B multimodal model with a unified 3D resampler for images and video, and a hybrid RL stage for short and long reasoning. I did the reasoning cold start and nearly all of the final RL experiments.
Data attribution for data curation
Bet, Aug 2024Data curation should run on optimization theory. In practice it runs on intuition and taste earned from empirical lessons, and for large-scale pretraining the best filter may be no filter at all.
Scores every training sample by its influence on the model's own rollouts, then continues training on the top 10%. I am the first author.
Early LLM agents
Bet, Jun 2023At that time agents were capped by the base model, so I expected them to take on the tedious work first, well before social simulation or full autonomy.
A tool-using LLM agent for root-cause analysis of cloud jobs, built on an internally deployed model and integrated into Alibaba Cloud's Flink platform. I am the first author.
Hobby
Dota 2A sick mental game that brings out the best and the worst in people. Favorite heroes: Axe and Necrophos. Favorite announcer: The Stanley Parable.
Strategy gamesGames that keep my mind fully engaged, such as Slay the Spire, Stellaris, Yu-Gi-Oh!, Go and no-guess Minesweeper. Most are on hold now, because they use the same part of my brain as research.