5 papers
Optimal Aggregation of LLM and PRM Signals for Efficient Test-Time Scaling
Peng Kuang, Yanli Wang, Xiaoyu Han +3
Process reward models (PRMs) are a cornerstone of test-time scaling (TTS), designed to verify and select the best responses from large language models (LLMs). However, this promise…
Break Me If You Can: Self-Jailbreaking of Aligned LLMs via Lexical Insertion Prompting
Devang Kulshreshtha, Hang Su, Haibo Jin +2
We introduce \emph{self-jailbreaking}, a threat model in which an aligned LLM guides its own compromise. Unlike most jailbreak techniques, which often rely on handcrafted prompts o…
Improving Noise Efficiency in Privacy-preserving Dataset Distillation
Runkai Zheng, Vishnu Asutosh Dasu, Yinong Oliver Wang +2
Modern machine learning models heavily rely on large datasets that often include sensitive and private information, raising serious privacy concerns. Differentially private (DP) da…
Tracing LLM Reasoning Processes with Strategic Games: A Framework for Planning, Revision, and Resource-Constrained Decision Making
Xiaopeng Yuan, Xingjian Zhang, Ke Xu +5
Large language models (LLMs) are increasingly used for tasks that require complex reasoning. Most benchmarks focus on final outcomes but overlook the intermediate reasoning steps -…
Examining Alignment of Large Language Models through Representative Heuristics: The Case of Political Stereotypes
Sullam Jeoung, Yubin Ge, Haohan Wang +1
Examining the alignment of large language models (LLMs) has become increasingly important, e.g., when LLMs fail to operate as intended. This study examines the alignment of LLMs wi…