UniProbe: A Learnable Token-Level Hallucination Detector for Large VLMs using Multi-Structural Internal Representations

2026-08-11

AI Summary

This work addresses the prevalent issue of hallucinations in large vision-language models (LVLMs), where generated text often contradicts visual content, necessitating fine-grained intervention. The authors propose UniProbe, a lightweight, learnable token-level hallucination detector that unifies heterogeneous internal representations—image patches, query tokens, and generated tokens—into a directed relational graph for the first time. By integrating spatial, sequential, and relational information through an alternating structure-aware module, UniProbe enables real-time hallucination detection and resampling during streaming decoding with only a single forward pass. Evaluated across multiple LVLM backbones, UniProbe achieves state-of-the-art performance in both token-level and object-level hallucination detection, reducing object hallucinations by up to 55% during decoding while introducing merely 6% additional latency.

Abstract

Large Vision-Language Models (LVLMs) achieve impressive visual reasoning and dialogue capabilities, yet frequently hallucinate content unsupported by the visual input. Effective mitigation requires token-level localization, enabling targeted intervention without discarding the entire response. Existing detectors require expensive full-model fine-tuning, rely on external verifiers that ignore the model's generation process, or reduce internal signals to isolated features and hand-crafted statistics, discarding spatial, sequential, and relational structure. We introduce \textbf{UniProbe}, a lightweight, unified, learnable detector that models a frozen LVLM's heterogeneous computational trace from a single forward pass. UniProbe constructs a directed graph over image patches, query tokens, and generated tokens, with attention weights encoding their relations. It processes this trace with alternating structure-aware modules: a GNN for relational evidence, a ViT for 2-D visual geometry, and a GRU for response order. Interleaving them allows spatial, relational, and sequential evidence to interact throughout the detector. We further develop a streaming variant for hallucination-aware decoding, which detects and resamples hallucinated tokens during generation, and a self-adaptation strategy aligning the detector with the LVLM's own generations. Across diverse LVLM backbones, UniProbe achieves state-of-the-art token-level and object-hallucination detection. During decoding, it reduces object hallucinations by up to 55\% at $1.06\times$ the latency of standard generation.

Problem

• hallucination detection
• large vision-language models
• token-level localization
• visual reasoning
• model hallucination

Innovation

• token-level hallucination detection
• multi-structural representation
• structure-aware modules
• streaming decoding
• frozen LVLM probing