The process of training a model to understand and connect different types of data (like audio and text) by mapping them into a shared space where related concepts are close together.
Quality of vision, audio, and image understanding (distinct from modality support)