Grounding Spatial Relations in Text-Only Language Models

Open in new window