Abstract

Unified architecture for vision-language, audio, and cross-modal understanding in a single multimodal framework.