GRPO is Secretly a Process Reward Model 文章

ArXiv CS.AI2026-05-29NEWSen作者: Michael Sullivan, Alexander Koller
查看原文 →
关系图谱

GRPO is Secretly a Process Reward Model · 相关技术

相关技术

Monte-Carlo-based PRMreinforcement learningGroup Relative Policy Optimizationoutcome reward modelsprocess reward model