VQAv-2

VQAv-2 is a dataset used to evaluate visual question answering systems by providing images paired with questions and answers, assessing the models' ability to understand and reason about visual content.

Papers using VQAv-2 (3)

When Thinking Hurts: Epistemic Signals in the Reasoning Chains of Visual Language Models2026CP-MoE: Consistency-Preserving Mixture-of-Experts for Continual Learning2026ManagerTower: Aggregating the Insights of Uni-Modal Experts for Vision-Language Representation Learning2023