CSRP: Chain-of-Thought Reasoning for Chinese Text Correction via Reinforcement Learning with Efficiency-Aware Rewards 文章

ArXiv CS.CL2026-06-02NEWSen作者: Wei Tian, Yuhao Zhou, Man Lan
查看原文 →
关系图谱

CSRP: Chain-of-Thought Reasoning for Chinese Text Correction via Reinforcement Learning with Efficiency-Aware Rewards · 相关技术

相关技术

LLMCPTSFTCGECEfficiency-Aware RewardGroup Relative Policy Optimizationchain-of-thoughtcontinual pre-trainingMaximum Likelihood EstimationSupervised Fine-TuningChinese Grammatical Error Correctionlarge language models