Tao Wei

On the Self-awareness of Large Reasoning Models' Capability Boundaries

TrojanTools: Adaptive Indirect Prompt Injection on LLM Agents via Malicious Tool-Calling

Unlearning Shapley: Data Valuation through Machine Unlearning

MPCache: MPC-Friendly KV Cache Eviction for Efficient Private LLM Inference

AegisGuard: RL-Guided Adapter Tuning for TEE-Based Efficient & Secure On-Device Inference

Kona: An Efficient Privacy-Preservation Framework for KNN Classification by Communication Optimization

NESTLE: An Efficient and Robust Data Valuation Framework for Large Language Models

Mitigating Catastrophic Forgetting in Large Language Models with Forgetting-aware Pruning