A task where you search for images using text descriptions (or vice versa) by comparing their embeddings in the shared semantic space.
Quality of vision, audio, and image understanding (distinct from modality support)