SOURCE-LINKED INTELLIGENCE

Look Where It Counts: A Free, Label-Free Visual Evidence Signal for Fine-Grained Vision-Language Reasoning

arXiv · AI, language, vision and robotics · article · Sep 21, 2026 · UTC

Multimodal large language models (MLLMs) fail at fine-grained visual questions less because they cannot reason than because they never see the evidence: high-resolution images are downsampled before encoding, so the model answers from linguistic priors. The standard remedies are expensive: annotated answers (SFT), hand-engineered verifiers (RLVR), or a large external teacher (on-policy distillation). We ask whether the visual evidence itself can supply the signal for free. We formalize the contrastive evidence gap, the per-token log-likelihood ratio that a model assigns to its own output when

Read original source ↗ Open in workspace

recordType

paper

region

Global