TY - RPRT TI - What Vision-Language Models `See' when they See Scenes AU - Michele Cafagna AU - Kees van Deemter AU - Albert Gatt PY - 2021 UR - https://arxiv.org/abs/2109.07301 ID - 2109.07301 ER -