Can Vision-Language Models Count? A Synthetic Benchmark and Analysis of Attention-Based Interventions

CoGeVLMRe-assign community