System-based outcome evaluation and causality-based behavior assessment for model robustness and explainability
Knowledge-enhanced generation techniques and foundation model development
Multimodal large language models and their applications
Welcome to the new Data Intelligence Lab homepage.
Towards Fully-Automated Materials Discovery via Large-Scale Synthesis Dataset and Expert-Level LLM-as-a-Judge
Rethinking the Training Paradigm of Discrete Token-Based Multimodal LLMs: An Analysis of Text-Centric Bias
Overcoming Source Object Grounding for Semantic Image Editing