Yiming Yang<o:p>

Recent Research<o:p>

<![if !supportLists]>�        <![endif]>LLM-Based Agents for Problem Solving: Research spans instruction tuning, demonstration-guided control, retrieval-augmented reasoning, and reinforcement learning enabling adaptive agents for complex tasks such as math problem solving, combinatorial search, code synthesis and revision, and PDE solving.<o:p>

<![if !supportLists]>�        <![endif]>Combinatorial Optimization and AI for Science: Developed learning-based solvers for NP-hard problems using diffusion models, Langevin dynamics, and Fourier neural operators; contributed to neural PDE solving, scientific workload scheduling, and code-generation-driven scientific computing.<o:p>

<![if !supportLists]>�        <![endif]>Scalable Oversight and Learning Beyond Human Supervision: Advanced reinforcement learning based alignment frameworks that scale supervision using self-play, instructible reward models, and principle-driven fine-tuning enabling controllable and generalizable LLMs that go beyond human-written data and guidance.<o:p>

Publications (list) <o:p>

Teaching<o:p>

Last updated: June 7, 2025 <o:p>