The Hidden Life of Tokens: Reducing Hallucination of Large Vision-Language Models via Visual Information Steering

TL;DR

This paper analyzes how large vision-language models hallucinate ungrounded content and introduces VISTA, a training-free inference method that significantly reduces hallucinations by leveraging internal token dynamics and early layer activations.

Contribution

The paper reveals internal token dynamics in LVLMs and proposes VISTA, a novel inference-time intervention that reduces hallucination without external supervision.

Findings

VISTA reduces hallucination by about 40% on average.

VISTA outperforms existing methods across multiple benchmarks.

VISTA is applicable to various decoding strategies and architectures.

Abstract

Large Vision-Language Models (LVLMs) can reason effectively over both textual and visual inputs, but they tend to hallucinate syntactically coherent yet visually ungrounded contents. In this paper, we investigate the internal dynamics of hallucination by examining the tokens logits ranking throughout the generation process, revealing three key patterns in how LVLMs process information: (1) gradual visual information loss - visually grounded tokens gradually become less favored throughout generation, and (2) early excitation - semantically meaningful tokens achieve peak activation in the layers earlier than the final layer. (3) hidden genuine information - visually grounded tokens though not being eventually decoded still retain relatively high rankings at inference. Based on these insights, we propose VISTA (Visual Information Steering with Token-logit Augmentation), a training-free inference-time intervention framework that reduces hallucination while promoting genuine information. VISTA works by combining two complementary approaches: reinforcing visual information in activation space and leveraging early layer activations to promote semantically meaningful decoding. Compared to existing methods, VISTA requires no external supervision and is applicable to various decoding strategies. Extensive experiments show that VISTA on average reduces hallucination by about 40% on evaluated open-ended generation task, and it consistently outperforms existing methods on four benchmarks across four architectures under three decoding strategies. Code is available at https://github.com/LzVv123456/VISTA.