paper-with-me

Papers

Geometric-Mean Policy Optimization

2025-07-28 · Yuzhong Zhao, Yue Liu, Junpeng Liu, Jingye Chen, Xun Wu, Yaru Hao, Tengchao Lv, Shaohan Huang, Lei Cui, Qixiang Ye, Fang Wan, Furu Wei arxiv

Group Relative Policy Optimization (GRPO) has significantly enhanced the reasoning capability of large language models by optimizing the arithmetic mean of token-level rewards. Unfortunately, GRPO is observed to suffer from unstable policy updates when facing tokens with outlier importance-weighted rewards, which manifest as extreme importance sampling ratios during training. In this study, we propose Geometric-Mean Policy Optimization (GMPO), with the aim to improve the stability of GRPO through suppressing token reward outliers. Instead of optimizing the arithmetic mean, GMPO maximizes the geometric mean of token-level rewards, which is inherently less sensitive to outliers and maintains a more stable range of importance sampling ratio. GMPO is plug-and-play-simply replacing GRPO's arithmetic mean with the geometric mean of token-level rewards, as the latter is inherently less sensitive to outliers. GMPO is theoretically plausible-analysis reveals that both GMPO and GRPO are weighted forms of the policy gradient while the former enjoys more stable weights, which consequently benefits policy optimization and performance. Experiments on multiple mathematical reasoning benchmarks show that GMPO-7B improves the average Pass@1 of GRPO by up to 4.1%, outperforming many state-of-the-art approaches. Code is available at https://github.com/callsys/GMPO.

📄 PDF Abstract BibTeX arXiv:2507.20673

Code (0)

등록된 구현이 없습니다.

Tasks

Mathematical Reasoning

Similar Papers 제목 키워드 기반

Adapt to Thrive! Adaptive Power-Mean Policy Optimization for Improved LLM Reasoning

2026-04-11 · Yiming Huang, Zhenbo Shi, Shuzheng Gao, Cuiyun Gao 외 arxiv

Reinforcement Learning with Verifiable Rewards (RLVR) is an essential paradigm that enhances the reasoning capabilities of Large Language Models (LLMs). However, existing methods typically rely on static policy optimizat…

Reinforcement LearningMathematical Reasoning

The Geometry of Memoryless Stochastic Policy Optimization in Infinite-Horizon POMDPs

2021-10-14 · ICLR 2022 4 · Johannes Müller, Guido Montúfar

We consider the problem of finding the best memoryless stochastic policy for an infinite-horizon partially observable Markov decision process (POMDP) with finite state and action spaces with respect to either the discoun…

Beyond expected value: geometric mean optimization for long-term policy performance in reinforcement learning

2025-08-29 · Xinyi Sheng, Dominik Baumann arxiv

Reinforcement learning (RL) algorithms typically optimize the expected cumulative reward, i.e., the expected value of the sum of scalar rewards an agent receives over the course of a trajectory. The expected value averag…

Reinforcement Learning

Semantic Geometry for policy-constrained interpretation

2025-12-10 · Nikit Phadke arxiv

We present a geometric framework for policy-constrained semantic interpretation that provably prevents hallucinated commitments in high-stakes domains. Semantic meaning is represented as direction on a unit sphere, evide…

Bayesian Inference

R-GTD: A Geometric Analysis of Gradient Temporal-Difference Learning in Singular Regimes

2026-01-28 · Hyunjun Na, Donghwan Lee arxiv

Gradient temporal-difference (GTD) learning algorithms are widely used for off-policy policy evaluation with function approximation. However, existing convergence analyses rely on the restrictive assumption that the so-c…