8 papers
Shadow in the Cache: Unveiling and Mitigating Privacy Risks of KV-cache in LLM Inference
Zhifan Luo, Shuo Shao, Su Zhang +5
The Key-Value (KV) cache, which stores intermediate attention computations (Key and Value pairs) to avoid redundant calculations, is a fundamental mechanism for accelerating Large…
Smooth Operator: Smooth Verifiable Reward Activates Spatial Reasoning Ability of Vision-Language Model
Siwen Jiao, Tianxiong Lv, Kangan Qian +7
Vision-Language Models (VLMs) face a critical bottleneck in achieving precise numerical prediction for 3D scene understanding. Traditional reinforcement learning (RL) approaches, p…
Incentivizing Cardiologist-Like Reasoning in MLLMs for Interpretable Echocardiographic Diagnosis
Yi Qin, Lehan Wang, Chenxu Zhao +2
Echocardiographic diagnosis is vital for cardiac screening yet remains challenging. Existing echocardiography foundation models do not effectively capture the relationships between…
HATS: High-Accuracy Triple-Set Watermarking for Large Language Models
Zhiqing Hu, Chenxu Zhao, Jiazhong Lu +1
Misuse of LLM-generated text can be curbed by watermarking techniques that embed implicit signals into the output. We propose a watermark that partitions the vocabulary at each dec…
Towards Benchmarking Privacy Vulnerabilities in Selective Forgetting with Large Language Models
Wei Qian, Chenxu Zhao, Yangyi Li +1
The rapid advancements in artificial intelligence (AI) have primarily focused on the process of learning from data to acquire knowledgeable learning systems. As these systems are i…
Beyond Over-Refusal: Scenario-Based Diagnostics and Post-Hoc Mitigation for Exaggerated Refusals in LLMs
Shuzhou Yuan, Ercong Nie, Yinuo Sun +3
Large language models (LLMs) frequently produce false refusals, declining benign requests that contain terms resembling unsafe queries. We address this challenge by introducing two…