Decoupling Reasoning and Confidence: Resurrecting Calibration in Reinforcement Learning from Verifiable Rewards 文章

ArXiv CS.CL2026-05-28NEWSen作者: Zhengzhao Ma, Xueru Wen, Boxi Cao, Yaojie Lu, Hongyu Lin, Jinglin Yang, Min He, Xianpei Han, Le Sun
查看原文 →
关系图谱

Decoupling Reasoning and Confidence: Resurrecting Calibration in Reinforcement Learning from Verifiable Rewards · 相关技术

相关技术

Reinforcement learning from verifiable rewardsLLMCPOlarge language modelsGRPORLVR