Hello! I am a third-year Ph.D. student at Carnegie Mellon University (CMU) in Electrical and Computer Engineering (ECE). I am fortunate to be advised by Andrea Zanette. Prior to that, I completed my undergraduate program from Yao Class at IIIS, Tsinghua University.

My research passion lies in expanding the boundary of Large Language Model (LLM) reasoning. My research experience spans across LLM post-training, diffusion language models and reinforcement learning.

I am actively looking for research internship opportunities in summer 2027.

Selected Research

Training Small Language Models for Transferable Tool Selection

May 2026 - Aug 2026

We train a small Qwen3 model via reinforcement learning for tool searching, a bottleneck in recent LLM agents tool calling. The trained model outperforms Llama-4-Maverick on ToolBench. We also demonstrate the transferability of the model across retrievers and to new tool corpora domains.

Control Variates Evaluation for head-to-head LLM comparison

Aug 2024 - Jan 2025

We apply control variates to evaluate the win rate in head-to-head LLM comparison, leading to unbiased evaluation with 12.2% saving of human annotations using an off-the-shelf synthetic evaluator and 24.8% saving using a finetuned variant. In addition, the saving of human annotations is predictable without actually running the evaluation procedure. [website]

Model-Based Return-Conditioned Supervised Learning

Research intern at UW (Feb. 2023 - Aug. 2023).

Supervisor: Simon S. Du.

We proposed model-based return-conditioned supervised learning (MBRCSL), a novel offline RL framework that is able to do trajectory stitching while retaining the strength of return-conditioned supervised learning (RCSL) to avoid Bellman completeness requirements. [website]

Skills

• Proficient in mathematical knowledge for ML research: calculus, linear algebra, abstract algebra, probability theory and optimization.
• Experienced in common programming language: C++, Python, Go, Verilog.
• Familiar with AI frameworks: Pytorch –>

News

• May 2026: I have started my internship at Meta! <!– - Sep 2024: I have joined CMU as a PhD student! I look forward to future collaborations with my advisor, Andrea Zanette.
• Jan 2024: One paper accepted by ICLR 2024!
• Nov 2023: One paper accepted by NeurIPS 2023 FMDM workshop (oral)! Thanks for efforts from all collaborators!
• Aug 2023: Participate in UAI 2023.
• May 2023: One paper accepted by UAI 2023! Thanks for generous support from Dr. Zaiwei Chen, Yiheng Lin and Prof. Adam Wierman.
• Feb 2023: Begin my spring visit at UW. Look forward to cooperation with Prof. Simon Shaolei Du. –>