arXiv cs.CL· Jihao Huang, Guangjia Chai, Zhirui Chen, Lei Wang, Yao Liu, Liu Qi, Pengjie Ding, Xinyang Liu, Chengxun Jia, Ling Shao, Hongyu Chen, Junchen Wan·· 12 小时前AI 评分27
MGE-PO：面向数学推理的熵感知信用分配方法
MGE-PO: Entropy-Aware Credit Assignment for Mathematical Reasoning
AI 导读
MGE-PO 提出四象限（4Q）优势塑形，将验证正确性与 rollout 生成熵结合：正确性决定额外修正的符号，熵控制幅度，低熵成功获得更大奖励、高熵失败受到更大惩罚。
正文 · 原文
Authors:Jihao Huang, Guangjia Chai, Zhirui Chen, Lei Wang, Yao Liu, Liu Qi, Pengjie Ding, Xinyang Liu, Chengxun Jia, Ling Shao, Hongyu Chen, Junchen Wan
Abstract:GRPO assigns the same group-relative advantage to equally rewarded responses within a question, leaving their generation uncertainty unused. We introduce MGE-PO, centered on four-quadrant (4Q) advantage shaping that combines verified correctness with rollout generation entropy. Correctness determines the sign of an additional correction, while entropy controls its magnitude: lower-entropy successes receive larger bonuses and higher-entropy failures receive larger penalties. An online entropy-based question curriculum supports this credit rule using cached rollout measurements. At the final checkpoint after eight training epochs on Qwen3-8B-Base, adding 4Q across three policy optimization settings yields absolute gains of 3.23-3.69% in six-benchmark macro avg@32 and 4.24-11.06% in macro pass@32. On Qwen2.5-7B-Base, the complete configuration raises four-benchmark macro avg@32 from 41.5% to 45.4%.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| ACM classes: | I.2.7; I.2.6 |
| Cite as: | arXiv:2610.23245 [cs.CL] |
| (or arXiv:2610.23245v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.23245 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jihao Huang [view email]
[v1]
Fri, 16 Oct 2026 10:15:02 UTC (1,506 KB)
来源：arXiv cs.CL · arxiv.org