Gregory Kang Ruey Lau

DUET: Optimizing LLM Training Data Mixtures via Noisy Feedback from Unseen, Downstream Evaluation Tasks

WaterDrum: Watermark-based Data-centric Unlearning Metric

The Chicken and Egg Dilemma: Co-optimizing Data and Model Configurations for LLMs

Uncertainty Quantification for Multimodal Large Language Models with Coherence-adjusted Semantic Volume

PIED: Physics-Informed Experimental Design for Inverse Problems