Open-ended VQA benchmarking of Vision-Language models by exploiting Classification datasets and their semantic hierarchy

VLMRe-assign community