MMStar

Are We on the Right Way for Evaluating Large Vision-Language Models?