Exploring Causes and Mitigation of Hallucinations in Large Vision Language Models
Journal:
arXiv
Published Date:
Feb 24, 2025
Abstract
Large Vision-Language Models (LVLMs) integrate image encoders with Large
Language Models (LLMs) to process multi-modal inputs and perform complex visual
tasks. However, they often generate hallucinations by describing non-existent
objects or attributes, compromising their reliability. This study analyzes
hallucination patterns in image captioning, showing that not all tokens in the
generation process are influenced by image input and that image dependency can
serve as a useful signal for hallucination detection. To address this, we
develop an automated pipeline to identify hallucinated objects and train a
token-level classifier using hidden representations from parallel inference
passes-with and without image input. Leveraging this classifier, we introduce a
decoding strategy that effectively controls hallucination rates in image
captioning at inference time.
