Peking University

Detecting Data Contamination from Reinforcement Learning Post-training for Large Language Models