Sangkyu Lee

M.S. student, Yonsei University

prof_pic.jpg

I am an M.S. student in the Multimodal Intelligence Research (MIR) Lab, advised by Youngjae Yu. Concurrently, I am a research scientist intern at the Language Lab of LG AI Research. Before LG AI Research, I was a research scientist intern at the Foundation Research Team in Naver Cloud and a technical lead for World Model for Robotics and Vehicle Control Team at maumAI. Before the M.S. degree program, I received a B.A. in Applied Statistics and a B.S. in Computer Science & Engineering from Yonsei University.

My research goal is to develop an embodied agent that autonomously discovers and acquires knowledge through experiences. To this end, I aim to find scalable approaches for how embodied agents can learn through the physical world without human supervision, not limited to physical constraints. Currently, I am researching related research topics to achieve efficient long-context multimodal generative models and continual learning through self-supervision for AI alignment.

🙌 Looking for Ph.D. programs! 🙌

News

Feb 18, 2025 Our new preprint, KL Penalty Control via Perturbation for Direct Preference Optimization, is released!
Jan 17, 2025 Check out our new preprint, SEAL: Entangled White-box Watermarks on Low-Rank Adaptation.
Dec 10, 2024 Congratulation! Our paper is accepted at AAAI 2025.
Sep 30, 2024 I am starting a research scientist internship at the Language Lab in LG AI Research.
May 16, 2024 Our paper is accepted as a main conference paper at ACL 2024!

Publications

  1. e-dpo.png
    KL Penalty Control via Perturbation for Direct Preference Optimization
    Sangkyu Lee*, Janghoon Han, Hosung Song, and 3 more authors
    Under Review, 2025
  2. seal.png
    SEAL: Entangled White-box Watermarks on Low-Rank Adaptation
    Giyeong Oh, Saejin Kim, Woohyun Cho, and 4 more authors
    Under Review, 2025
  3. AAAI 2025
    mass.png
    MASS: Overcoming Language Bias in Image-Text Matching
    Jiwan Chung, Seungwon Lim, Sangkyu Lee, and 1 more author
    Accepted, AAAI 2025
  4. hyperclova_x.jpeg
    HyperCLOVA X Technical Report
    HyperCLOVA X Team
    arXiv preprint arXiv:2404.01954, 2024
  5. ACL 2024 Main
    self_judge.png
    Aligning Large Language Models by On-Policy Self-Judgment
    Sangkyu Lee*, Sungdong Kim, Ashkan Yousefpour, and 3 more authors
    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Bangkok, Thailand, August 2024. Association for Computational Linguistics.