Saved in:
Bibliographic Details
Main Authors: Chen, Junhua, Zhang, Zixi, Zhong, Hantao, Antonova, Rika
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.03679
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916990078156800
author Chen, Junhua
Zhang, Zixi
Zhong, Hantao
Antonova, Rika
author_facet Chen, Junhua
Zhang, Zixi
Zhong, Hantao
Antonova, Rika
contents We introduce Group Policy Gradient (GPG), a family of critic-free policy-gradient estimators for general MDPs. Inspired by the success of GRPO's approach in Reinforcement Learning from Human Feedback (RLHF), GPG replaces a learned value function with a group-based Monte Carlo advantage estimator, removing the memory, compute, and hyperparameter costs of training a critic while preserving PPO's clipped-objective structure. We prove the consistency of the GPG estimator, analyze the bias-variance tradeoffs, and demonstrate empirically that GPG matches or outperforms PPO on standard benchmarks. GPG makes better use of parallel simulations, which, together with its critic-free design, results in more efficient use of computational resources than PPO.
format Preprint
id arxiv_https___arxiv_org_abs_2510_03679
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Group Policy Gradient
Chen, Junhua
Zhang, Zixi
Zhong, Hantao
Antonova, Rika
Machine Learning
We introduce Group Policy Gradient (GPG), a family of critic-free policy-gradient estimators for general MDPs. Inspired by the success of GRPO's approach in Reinforcement Learning from Human Feedback (RLHF), GPG replaces a learned value function with a group-based Monte Carlo advantage estimator, removing the memory, compute, and hyperparameter costs of training a critic while preserving PPO's clipped-objective structure. We prove the consistency of the GPG estimator, analyze the bias-variance tradeoffs, and demonstrate empirically that GPG matches or outperforms PPO on standard benchmarks. GPG makes better use of parallel simulations, which, together with its critic-free design, results in more efficient use of computational resources than PPO.
title Group Policy Gradient
topic Machine Learning
url https://arxiv.org/abs/2510.03679