Research

Preprints

PRISM: Robust VLM Alignment with Principled Reasoning for Integrated Safety in Multimodality

Published in arXiv, 2025

This paper introduces PRISM, a framework for robust alignment of Vision-Language Models (VLMs) with principled reasoning to ensure integrated safety in multimodal settings. The key challenge addressed is overdefense, which harms utility, and the balance between safety and benign performance. Paper Code Website

Publications

The Labyrinth of Links: Navigating the Associative Maze of Multi-modal LLMs

Published in ICLR, 2025

In this paper, we first devise a standard association benchmark based on adjective and verb association semantic concepts. Instead of costly data annotation and organization, we propose a convenient annotation-free reconstruction method transforming the general dataset for our association tasks. Furthermore, we comprehensively investigate the MLLMs’s ability and potential for association ability. Project Code Paper Video

Nanxi Li

Research

Preprints

PRISM: Robust VLM Alignment with Principled Reasoning for Integrated Safety in Multimodality

Publications

The Labyrinth of Links: Navigating the Associative Maze of Multi-modal LLMs