← Search

Jusang Oh

2 accepted papers

2026

VALUEFLOW: Toward Pluralistic and Steerable Value-based Alignment in Large Language Models

ICML 2026oral

Aligning Large Language Models (LLMs) with the diverse spectrum of human values remains a central challenge: preference-based methods often fail to capture deeper motivational principles. Value-based approaches offer a more principled path, yet three gaps persist– extraction often ignores hierarchic…

Cited by 0SourceScholar
2024

Towards Efficient Visual-Language Alignment of the Q-Former for Visual Reasoning Tasks

EMNLP 2024finding

Recent advancements in large language models have demonstrated enhanced capabilities in visual reasoning tasks by employing additional encoders for aligning different modalities. While the Q-Former has been widely used as a general encoder for aligning several modalities including image, video, audi…