publications

2026

1. Preprint

   

   LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL

   Yujin Kim^*, Namgyu Ho^*, Sangmin Hwang^*, and 7 more authors

   2026

   Preprint under review. * Equal contribution

   Abs arXiv

   Proposes a policy-aware prompt adaptation framework for instruction-following RL that leverages an LLM to detect non-discriminative prompts from policy rollouts and scale prompt difficulty.

2025

1. KSDIM

   An Analysis of How Social Hierarchy Impacts Decision-Making in LLM Agent Collaboration

   Jihwan Oh, Sangmin Hwang, Sehyeok Kang, and 1 more author

   Journal of the Korea Society of Digital Industry and Information Management, 2025

   Abs PDF

   Investigates how assigning hierarchical roles to LLM agents influences their decision-making processes in multi-LLM debate.