Activation Steering Decoding: Mitigating Hallucination in Large Vision-Language Models through Bidirectional Hidden State Intervention
Open Access
- 1 January 2025
- conference paper
- Published by Association for Computational Linguistics (ACL)
- p. 12964-12974
Abstract
Jingran Su, Jingfan Chen, Hongxin Li, Yuntao Chen, Li Qing, Zhaoxiang Zhang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.Keywords
This publication has 12 references indexed in Scilit:
- Beyond Multimodal Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimization2025
- IBD: Alleviating Hallucinations in Large Vision-Language Models via Image-Biased Decoding2025
- Mitigating Object Hallucinations in Large Vision-Language Models with Assembly of Global and Local Attention2025
- Hallucination‐Free? Assessing the Reliability of Leading AI Legal Research ToolsJournal of Empirical Legal Studies, 2025
- Woodpecker: hallucination correction for multimodal large language modelsScience China Information Sciences, 2024
- Less is More: Mitigating Multimodal Hallucination from an EOS Decision Perspective2024
- Steering Llama 2 via Contrastive Activation Addition2024
- Aligning Large Multimodal Models with Factually Augmented RLHF2024
- Volcano: Mitigating Multimodal Hallucination through Self-Feedback Guided Revision2024
- Extracting Latent Steering Vectors from Pretrained Language Models2022