Publications

Bonmu Ku - arXiv Preprint, 2026

Documents a language model achieving a perfect score on the LSAT. Shows that removing internal reasoning reduces accuracy by up to 8 points, and introduces a QLoRA-trained process reward model that improves selection on logical reasoning.

HB Ku, J Shin, HJ Lee, S Na, I Jeon - Proceedings of the 12th Argument Mining Workshop, 2025

A multi-agent framework that casts implicit premise recovery as a dialogic reasoning task between two LLM agents, achieving state-of-the-art accuracy on premise selection.

HB Ku - Proceedings of the NLPSI Workshop at the 19th International AAAI Conference, 2025

Demonstrates that pragmatic reasoning in LLMs arises not from additional tuning, but from contextualized interaction enabled by a multi-agent framework.

HB Ku - 2025 4th International Conference on Artificial Intelligence, Internet of Things and Cloud Computing Technology (AIoTC), 2025

Applies Laplacian-regularized Low-Rank Adaptation to enhance pragmatic inference capabilities in tiny language models for edge deployment.