Minbeom Kim

Hi! I am a Ph.D student at Seoul National University. I am fortunate to be advised by Prof. Kyomin Jung. Previously, I received B.E. degree from KAIST. I have also interned at NAVER AI LAB and NAVER Labs Europe for controlled text generation.

My Ph.D. research primarily focuses on Controlled Generation from Large Language Models. More specifically, how can we control generations that are 1) guaranteed to strictly satisfy essential constraints (e.g. safety guarantee), 2) aligned with personal preferences, and 3) still preserving pre-aligned general preferences? As a practical and impactful solution for challenges, I strongly believe that harmony between training-time methods and inference-time methods is essential. Although it looks deeply challenging, I am enjoying this journey and looking forward to the new insights it will bring. 😎