Sunghwan Kim

Research: Self-Improving Agents

Hi! I am a second year M.S. student at Yonsei University advised by Jinyoung Yeo. Previously, I received my B.S. in Materials Science & Engineering from Yonsei University in August 2024.

I aim to build human-like intelligent systems that can autonomously learn, reason, and adapt to diverse environments. My recent research interests include reinforcement learning for long-horizon tasks and intelligent systems that learn through interaction with their environments.

News

2026.04 Our "On Training LLMS for Long-Horizon Tasks" got accepted to ICML 2026.
2026.01 Our "Memento" and "MEM1" got accepted to ICLR 2026.
2025.09 Our "Web-Shepherd" got accepted to NeurIPS 2025 as a Spotlight.
2025.05 Our "Reward Model Evaluation" and "LLM Meets Scene Graph" got accepted to ACL 2025.
2025.05 I will join Microsoft Research Asia (MSRA) as a research intern.
2025.01 Our work "World Model for Web Agent" got accepted to ICLR 2025.
2024.09 Our work "Think-and-Execute" got accepted to EMNLP 2024 and "Cactus" got accepted to EMNLP 2024 Findings.
2024.08 Our paper has been selected as an Outstanding Paper at ACL 2024.
2024.05 Our work "Can Large Language Models be Good Emotional Supporter? Mitigating Preference Bias on Emotional Support Conversation" got accepted to ACL 2024.

Selected Publications

ICML 2026

On Training Large Language Models for Long-Horizon Tasks: An Empirical Study of Horizon Length

Sunghwan Kim, Junhee Cho, Beong-woo Kwak, Taeyoon Kwon, Liang Wang, Nan Yang, Xingxing Zhang, Furu Wei, Jinyoung Yeo

Paper
NeurIPS 2025 Spotlight

Web-Shepherd: Advancing PRMs for Reinforcing Web Agents

Hyungjoo Chae*, Sunghwan Kim*, Junhee Cho*, Seungone Kim, Seungjun Moon, Gyeom Hwangbo, Dongha Lim, Minjin Kim, Yeonjun Hwang, Minju Gwak, Dongwook Choi, Minseok Kang, Gwanhoon Im, ByeongUng Cho, Hyojun Kim, Jun Hee Han, Taeyoon Kwon, Minju Kim, Beong-woo Kwak, Dongjin Kang, Jinyoung Yeo
* Equal contribution

Paper Website Code
ACL 2025 Oral

Rethinking Reward Model Evaluation Through the Lens of Reward Overoptimization

Sunghwan Kim*, Dongjin Kang*, Taeyoon Kwon, Hyungjoo Chae, Dongha Lee, Jinyoung Yeo
* Equal contribution

Paper Code
ACL 2024 Outstanding Paper Award

Can Large Language Models be Good Emotional Supporter? Mitigating Preference Bias on Emotional Support Conversation

Dongjin Kang*, Sunghwan Kim*, Taeyoon Kwon, Seungjun Moon, Hyunsouk Cho, Youngjae Yu, Dongha Lee, Jinyoung Yeo
* Equal contribution

Paper Code
→ View all publications

Vitae

2024.09 - Present

M.S. Student

Academy

Yonsei University

Advised by Jinyoung Yeo.

2025.07 - 2025.12

Research Intern

Industry

Microsoft Research Asia (MSRA)

RL for long-horizon LLM agents. (Mentored by Liang Wang, Nan Yang, and Xingxing Zhang.)

2018.03 - 2024.08

B.S. in Materials Science & Engineering

Academy

Yonsei University

Military Service (Aug. 2020 - Feb. 2022)