aiqbench analysis
This research introduces Atomic Visual Entailment (AVE), a zero-shot framework that decomposes complex visual hypotheses into atomic facts to improve reasoning without fine-tuning large vision-language models. By learning to selectively trust predictions from either the full hypothesis or its decomposed parts, the method achieves 80.3% accuracy on SNLI-VE and successfully localizes visual evidence without region-level supervision. The approach offers a practical alternative for scenarios where acquiring labeled data or computational resources for direct model fine-tuning is prohibitive.
Atomic Visual Entailment: Enhancing Zero-Shot Vision-Language Reasoning through Atomic Fact Decomposition and Learned Selection
arXiv:2610.05630v1 Announce Type: new Abstract: Visual entailment (VE) asks whether an image supports, contradicts, or leaves undecided a textual hypothesis. Strong results come from fine-tuning large vision-language models on labelled data, while zero-shot and hybrid approaches remain far behind. A VE hypothesis often bundles several visual claims, yet existing zero-shot methods reason over it as a single unit. We propose Atomic Visual Entailment (AVE), which decomposes the hypothesis into atomic facts, produces candidate predictions from both the full hypothesis and its facts using frozen vision-language models, and predicts the final label with a lightweight classifier trained only on how those candidates behave. We find that decomposition helps only when the hypothesis context is preserved: judging facts in isolation is worse than not decomposing at all. Full-hypothesis and atomic prediction make complementary errors, and learning which to trust recovers far more of that complementarity than majority voting, reaching 0.803 test accuracy on SNLI-VE without fine-tuning any vision-language model. AVE also localises the visual evidence behind its prediction without region-level supervision. These results suggest that learning which candidate prediction to trust can close much of the gap to fine-tuned systems, offering a practical alternative where fine-tuning a vision-language model directly would need more labelled data or compute than is available.
Abstract:Visual entailment (VE) asks whether an image supports, contradicts, or leaves undecided a textual hypothesis. Strong results come from fine-tuning large vision-language models on labelled data, while zero-shot and hybrid approaches remain far behind. A VE hypothesis often bundles several visual claims, yet existing zero-shot methods reason over it as a single unit. We propose Atomic Visual Entailment (AVE), which decomposes the hypothesis into atomic facts, produces candidate predictions from both the full hypothesis and its facts using frozen vision-language models, and predicts the final label with a lightweight classifier trained only on how those candidates behave. We find that decomposition helps only when the hypothesis context is preserved: judging facts in isolation is worse than not decomposing at all. Full-hypothesis and atomic prediction make complementary errors, and learning which to trust recovers far more of that complementarity than majority voting, reaching 0.803 test accuracy on SNLI-VE without fine-tuning any vision-language model. AVE also localises the visual evidence behind its prediction without region-level supervision. These results suggest that learning which candidate prediction to trust can close much of the gap to fine-tuned systems, offering a practical alternative where fine-tuning a vision-language model directly would need more labelled data or compute than is available.
Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2610.05630 [cs.CL]
(or arXiv:2610.05630v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.05630 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.
Full text under the publisher's open licence · 304 words
Related coverage
Dual Adversarial Fine-tuning for Enhancing Robustness of Large Vision Language Model
arXiv · Jul 23, 2026
SentZero: An Enhanced Sentence-Centric Vision-Language Pretraining for Multi-Task Zero-Shot Chest X-Ray Analysis
arXiv · Sep 29, 2026
Subspace Alignment for Vision-Language Model Test-time Adaptation
arXiv · Aug 28, 2026
HiMemVLN: Enhancing Reliability of Open-Source Zero-Shot Vision-and-Language Navigation with Hierarchical Memory System
arXiv · Jul 21, 2026