NaturalBench: Evaluating Vision-Language Models on Natural Adversarial Samples Paper • 2410.14669 • Published 18 days ago • 35