Multimodal Large Language Model

Multimodal large language models (MLLMs) are driving AI toward cross-modal understanding, generation, and interaction. Our research focuses on key challenges in modality fusion, cross-domain generalization, and interpretable reasoning. We develop integrated approaches to representation, architecture, and data for more capable and human-centered AI systems.