Jaewan Park

Jaewan Park

Hi! I'm an undergraduate student at Seoul National University studying computer science and math.

My research focuses on machine learning, particularly large language models and reinforcement learning. My interests span both theoretical questions and practical applications, including how these systems can be better understood, trained, and deployed. I'm currently a research intern at KAIST OSI Lab advised by Professor Se-Young Yun.

Publications

Cross-Agent Learning Signals Enable Coordinated Role-Decomposed LLM Training

Jaewan Park, Solbee Cho, Jay-Yoon Lee

Preprint

Stop-RAG: Value-Based Retrieval Control for Iterative RAG

Jaewan Park*, Solbee Cho*, Jay-Yoon Lee

MTI-LLM @ NeurIPS 2025

* Equal contribution

Recent Posts

Dec 27, 2025

RL for LLMs II: Stabilization Tricks for Modern LLMs

Dec 21, 2025

RL for LLMs I: The Token-Level MDP and Off-Policy Policy Gradients

Jul 24, 2025

From Learning-to-Rank to Direct Alignment

Sep 15, 2024

An Overview of Supervised Learning