Fetching the paper…

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