ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models

Open in new window