profile photo of Jiyun Bae

Jiyun Bae

Contact: jiyun.bae [at] postech.ac.kr

I am a Ph.D. student in the Graduate School of Artificial Intelligence at POSTECH, advised by Jaeho Lee in the Efficient Learning Lab (EffL). My long-term goal is to build reasoning models that distinguish relevant evidence from irrelevant information. My current work focuses on understanding the failure modes of reasoning vision-language models under irrelevant information. More broadly, I am interested in reasoning, multimodal learning, and robustness.

Publications

Understanding the Effects of Distractors on Reasoning Vision-Language Models
Jiyun Bae, Hyunjong Ok, Sangwoo Mo, Jaeho Lee
EMNLP 2026
paper | code | project page

Education

Ph.D. in Artificial Intelligence, POSTECH (Advisor: Jaeho Lee), Mar. 2026 – Present
M.S. in Artificial Intelligence, POSTECH (Advisor: Jaeho Lee), Feb. 2024 – Feb. 2026
B.E. in Information Security, Seoul Women's University, Mar. 2020 – Feb. 2024

Experience

Undergraduate Research Fellow, Efficient Learning Lab, POSTECH, Jun. 2023 – Feb. 2024
Founder (1st Lead), Google Developer Groups on Campus, Google Korea, Aug. 2022 – Jul. 2023
NLP Researcher Intern, S2W, Jul. 2022 – Sep. 2022
Software Developer Intern, Ericsson-LG, Jan. 2022 – Feb. 2022

Teaching

Teaching Assistant, CSED554 Deep Learning for Natural Language Processing, POSTECH, Spring 2025