In-Context Collapse in Vision-Language Models and How to Mitigate it?

Title:In-Context Collapse in Vision-Language Models and How to Mitigate it?

Abstract:Many-shot in-context learning (ICL) lets vision-language models (VLMs) adapt from image--label demonstrations without weight updates, and is widely assumed to improve as more demonstrations are supplied. We show the opposite: as demonstrations accumulate, a subset of VLMs undergo an \emph{in-context collapse}, a sharp, sometimes catastrophic accuracy drop spanning synthetic classification, natural-image classification, and VQA benchmarks, in some models falling below chance while outputs remain well-formed. Across an open VLM panel ($0.5$B--$11$B) and a frontier model (Claude Sonnet 4.5), the collapse is graded. Two capabilities turn out to be dissociable: robustness to accumulating demonstrations and the ability to learn a novel rule in context, their combinations yield three reproducible regimes. A parameter-matched lesion-and-rescue causally localizes the collapse to the vision-language integration pathway: an adapter on the connector and early/mid layers restores genuine learning (remap accuracy $0.39!\rightarrow!0.91$ at 16 shots), while an equal-capacity adapter on the late readout does not. We propose \textsc{CircA}, whose core is a one-time integration vaccine: trained once on one synthetic task, it transfers collapse-resistance to unseen task families (chance$\rightarrow$$0.71$/$0.60$ on CIFAR/Fashion). The layers best for in-context integration are not the layers best for weight-based consolidation, the late readout achieves higher accuracy and less forgetting at fewer parameters. The collapse is an integration failure at the vision--language interface, correctable by a lightweight, transferable intervention.

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

arXiv:2608.02830 [cs.CV]

(or arXiv:2608.02830v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2608.02830 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.