Yuzhen Mao

IceCache: Memory-Efficient KV-cache Management for Long-Sequence LLMs

MEM-$\alpha$: LEARNING MEMORY CONSTRUCTION VIA REINFORCEMENT LEARNING

SelectLLM – Calibrating LLMs for Selective Prediction: Balancing Coverage and Risk

ADiff4TPP: Asynchronous Diffusion Models for Temporal Point Processes

Performative Risk Control: Calibrating Models for Reliable Deployment under Performativity