The Evidence Anchoring Boundary: Where Visual Interventions Matter and When They Help in Multimodal LLMs

arXiv:2610.12981v1 Announce Type: cross Abstract: Visual interventions provide a training-free way to reduce hallucinations in multimodal large language models. However, an intervention may have little leverage at some layers and may damage a correct answer even when it changes the output. We investigate where visual interventions retain leverage and when they improve predictions. First, we apply the same visual-residual edit at successive decoder layers, matching its RMS strength across depth. Across six checkpoints from three model families, we observe a persistent drop in intervention effect and define its operational onset for this edit as the \textbf{Evidence Anchoring Boundary (EAB)}. With boundaries fixed on calibration queries, held-out effects on disjoint evaluation queries before EAB are $13.78\times$--$26.10\times$ larger than those after it. Comparison with intermediate readouts shows that the edit can lose leverage before answer preferences stabilize. Second, we compare counterfactual subtraction across presence, attribute, and relation questions. Subtraction can correct a misleading preference, leave a decision unchanged, or remove a preference supporting the correct answer. Based on these observations, we develop calibrated selective routing, which preserves the original prediction unless calibration supports an intervention for the question type. It improves full-set attribute accuracy by $1.56$--$1.81$ percentage points across three checkpoints, while universal subtraction is neutral or harmful.

Abstract:Visual interventions provide a training-free way to reduce hallucinations in multimodal large language models. However, an intervention may have little leverage at some layers and may damage a correct answer even when it changes the output. We investigate where visual interventions retain leverage and when they improve predictions. First, we apply the same visual-residual edit at successive decoder layers, matching its RMS strength across depth. Across six checkpoints from three model families, we observe a persistent drop in intervention effect and define its operational onset for this edit as the \textbf{Evidence Anchoring Boundary (EAB)}. With boundaries fixed on calibration queries, held-out effects on disjoint evaluation queries before EAB are $13.78\times$--$26.10\times$ larger than those after it. Comparison with intermediate readouts shows that the edit can lose leverage before answer preferences stabilize. Second, we compare counterfactual subtraction across presence, attribute, and relation questions. Subtraction can correct a misleading preference, leave a decision unchanged, or remove a preference supporting the correct answer. Based on these observations, we develop calibrated selective routing, which preserves the original prediction unless calibration supports an intervention for the question type. It improves full-set attribute accuracy by $1.56$--$1.81$ percentage points across three checkpoints, while universal subtraction is neutral or harmful.

Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)

arXiv:2610.12981 [cs.CV]

(or arXiv:2610.12981v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2610.12981 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

analiza aiqbench

The Evidence Anchoring Boundary: Where Visual Interventions Matter and When They Help in Multimodal LLMs

Subiecte conexe