Exploring Causes and Mitigation of Hallucinations in Large Vision Language Models
This addresses reliability issues in multi-modal AI systems, but it is incremental as it builds on existing methods for hallucination detection.
The study tackled the problem of hallucinations in Large Vision-Language Models by analyzing patterns in image captioning and developing a token-level classifier to detect and control hallucinations, resulting in effective mitigation of hallucination rates during inference.
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.