RLPR: EXTRAPOLATING RLVR TO GENERAL DOMAINS WITHOUT VERIFIERS

RLPR: Extrapolating RLVR to General Domains without Verifiershttps://arxiv.org/pdf/2506.18254v1

1.概述

        大规模的可验证奖励强化学习（RLVR）已成为提升大型语言模型（LLMs）推理能力的有前途的范式（Jaech et al., 2024; DeepSeek-AI et al.,