Changdae Oh

PhD student@Department of Computer Science, University of Wisconsin - Madison·美国·OpenReview

研究方向

uncertainty quantification and calibration · vision-language models · multimodal large language models · representation learning · robustness · fairness · out-of-distribution generalization

7.2

How Do Transformers Learn to Associate Tokens: Gradient Leading Terms Bring Mechanistic Interpretability

ICLR 2026Oral

二作

5.0

Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach

ICML 2025Poster

一作

4.2

Group rank for encrypted data

合作者 (20)

Alexander G Hauptmann

1 篇

Changdae Oh

How Do Transformers Learn to Associate Tokens: Gradient Leading Terms Bring Mechanistic Interpretability

Understanding Language Prior of LVLMs by Contrasting Chain-of-Embedding

General Exploratory Bonus for Optimistic Exploration in RLHF

Visual Instruction Bottleneck Tuning

DaWin: Training-free Dynamic Weight Interpolation for Robust Adaptation

Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach

Group rank for encrypted data