Selected publications
-
Adaptive Stopping of Streamed Reasoning for Language Model Agents
JunHyeok Oh, Byung-Jun Lee
arXivPreprint, 2026
-
HorizonFlow: Variable-Length Planning for Offline Goal-Conditioned RL
JunHyeok Oh, Zian Jang, Byung-Jun Lee
arXivarXiv preprint, 2026
-
TABX: A High-Throughput Sandbox Battle Simulator for Multi-Agent Reinforcement Learning
Hayeong Lee*, JunHyeok Oh*, Byung-Jun Lee
ICMLInternational Conference on Machine Learning, 2026 · * equal contribution
-
Iterative Prompt Refinement for Safer Text-to-Image Generation
Jinwoo Jeon*, JunHyeok Oh*, Hayeong Lee, Byung-Jun Lee
EMNLPConference on Empirical Methods in Natural Language Processing, 2025 · * equal contribution
All publications
- 2026Adaptive Stopping of Streamed Reasoning for Language Model Agents. JunHyeok Oh, Byung-Jun Lee. arXiv preprint. arXiv · Project page · Dataset
- 2026HorizonFlow: Variable-Length Planning for Offline Goal-Conditioned RL. JunHyeok Oh, Zian Jang, Byung-Jun Lee. arXiv preprint. Project page · arXiv
- 2026TABX: A High-Throughput Sandbox Battle Simulator for Multi-Agent Reinforcement Learning. Hayeong Lee*, JunHyeok Oh*, Byung-Jun Lee. ICML 2026. Project page · arXiv
- 2025Prior-Guided Diffusion Planning for Offline Reinforcement Learning. Donghyeon Ki, JunHyeok Oh, Seong-Woong Shim, Byung-Jun Lee. NeurIPS 2025. arXiv · Code
- 2025Iterative Prompt Refinement for Safer Text-to-Image Generation. Jinwoo Jeon*, JunHyeok Oh*, Hayeong Lee, Byung-Jun Lee. EMNLP 2025. arXiv · Code
- 2025Rethinking DPO: The Role of Rejected Responses in Preference Misalignment. Jae Hyeon Cho, JunHyeok Oh, Myunsoo Kim, Byung-Jun Lee. Findings of EMNLP 2025. arXiv
- 2025Offline Reinforcement Learning with Penalized Action Noise Injection. JunHyeok Oh, Byung-Jun Lee. arXiv preprint. arXiv
* equal contribution
Education
- 2025 – presentKorea University, Seoul — Integrated M.S./Ph.D., Artificial IntelligenceAdvised by Prof. Byung-Jun Lee
- 2020.03 – 2024.02Korea National University of Education, Cheongju — B.S., Mathematics Education
Awards
- 2022.103rd Place (Encouragement Award), Han River Water Level Prediction AI Competition for Paldang Dam Flood Safety OperationKorea Hydro & Nuclear Power (KHNP)