Two-Token Features and Small-Large Ensembles for VLM Hallucination Detection

Paper recorded by Signals 4 on 2026-09-09 in cs.CL. Abstract reproduced from arXiv; link to the original below.

Published 2026-09-09 on arXiv · recorded by Signals 4 on 2026-09-10

Category: cs.CL · 自然语言处理 · first seen 2026-09-10

Abstract

We present our system for the SHROOM-Visions 2026 shared task on character-level VLM hallucination detection. A small ($4$B-parameter) VLM is fine-tuned as a per-token classifier reading a two-token feature from its own hidden states, and is ensembled with a $\sim$400B zero-shot VLM judge at prediction time. Both components see off-the-shelf OCR of any visible in-image text. We use synthetic hallu

Read on arXiv →