8 papers
Mitigating Action-Relation Hallucinations in LVLMs via Relation-aware Visual Enhancement
Zhenxin Qin, Qiang Li, Qingzhuo Wang +3
Large Vision-Language Models (LVLMs) have achieved remarkable performance on diverse vision-language tasks. However, LVLMs still suffer from hallucinations, generating text that co…
A Unified Approach to Interpreting Knowledge Distillation for Large Language Models via Interactions
Qingzhuo Wang, Ruiyang Qin, Zhenxin Qin +2
Despite the success of knowledge distillation (KD) in Large Language Models (LLMs), the underlying mechanism behind its efficacy remains unclear. In this paper, we propose a unifie…
Understanding and Defending VLM Jailbreaks via Jailbreak-Related Representation Shift
Zhihua Wei, Qiang Li, Jian Ruan +4
Large vision-language models (VLMs) often exhibit weakened safety alignment with the integration of the visual modality. Even when text prompts contain explicit harmful intent, add…
Harnessing Pairwise Ranking Prompting Through Sample-Efficient Ranking Distillation
Junru Wu, Le Yan, Zhen Qin +6
While Pairwise Ranking Prompting (PRP) with Large Language Models (LLMs) is one of the most effective zero-shot document ranking methods, it has a quadratic computational complexit…
Boosting Reward Model with Preference-Conditional Multi-Aspect Synthetic Data Generation
Jiaming Shen, Ran Xu, Yennie Jun +6
Reward models (RMs) are crucial for aligning large language models (LLMs) with human preferences. They are trained using preference datasets where each example consists of one inpu…
RRM: Robust Reward Model Training Mitigates Reward Hacking
Tianqi Liu, Wei Xiong, Jie Ren +15
Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human preferences. However, traditional RM training, which relies on response pairs tied to sp…