Associate Professor, University of Science and Technology of China
4 papers at NeurIPS 2025
The paper presents AnyMDP, a framework for procedurally generating diverse tasks to enhance In-Context Reinforcement Learning (ICRL) scalability, and explores the trade-off between generalization and adaptation efficiency.
Proposed PMQ-VE, a new quantization method for video enhancement that achieves high performance with low-bit models.
