From Token Correction to Claim Admission: Training-Free Hallucination Mitigation for Large Vision-Language Models

arXiv:2610.17706v1 Announce Type: new Abstract: Large vision-language models often exhibit object hallucination, generating claims that are not supported by the input image. To mitigate this issue, we propose Evidence-Guided Visual Claim Admission (EGVCA), a training-free framework that admits a semantic claim only when it receives sufficient image-dependent support. EGVCA measures this support by comparing the claim likelihood under the observed image and a visual-null input. For captioning, it controls semantic expansion without altering token generation. For binary visual question answering, it combines atomic visual support with the image-induced change in the model's native Yes--No margin. Experiments on CHAIR, POPE, and MME across three LVLM backbones show that EGVCA substantially reduces object hallucination while preserving caption quality. It achieves the highest POPE F1 in eight of nine backbone--sampling settings and improves the native MME total score on all three backbones. These results demonstrate the effectiveness of claim-level admission as a simple alternative to token-level distribution correction.

Related

• GEASS: Training-Free Caption Steering for Hallucination Mitigation in Vision-Language Models
• Steer Where It Matters: Token-Level Visual-Sensitivity Steering for LVLMs Hallucination Mitigation
• ReWEIGH the Evidence: Calibrating Token-Level Ordinal Visual Evidence to Mitigate Hallucinations in Large Vision-Language Models
• From Token Importance to Conditional Removability: Rethinking Visual Token Pruning in Multimodal Large Language Models

Source: arXiv cs.CV | 2026-10-14