Saved in:
Bibliographic Details
Main Authors: Zhu, Baoheng, Bo, Deyu, Zhang, Delvin Ce, Wang, Xiao
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.10395
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912960306216960
author Zhu, Baoheng
Bo, Deyu
Zhang, Delvin Ce
Wang, Xiao
author_facet Zhu, Baoheng
Bo, Deyu
Zhang, Delvin Ce
Wang, Xiao
contents Graph generation is a fundamental task with broad applications, such as drug discovery. Recently, discrete flow matching-based graph generation, \aka, graph flow model (GFM), has emerged due to its superior performance and flexible sampling. However, effectively aligning GFMs with complex human preferences or task-specific objectives remains a significant challenge. In this paper, we propose Graph-GRPO, an online reinforcement learning (RL) framework for training GFMs under verifiable rewards. Our method makes two key contributions: (1) We derive an analytical expression for the transition probability of GFMs, replacing the Monte Carlo sampling and enabling fully differentiable rollouts for RL training; (2) We propose a refinement strategy that randomly perturbs specific nodes and edges in a graph, and regenerates them, allowing for localized exploration and self-improvement of generation quality. Extensive experiments on both synthetic and real datasets demonstrate the effectiveness of Graph-GRPO. With only 50 denoising steps, our method achieves 95.0\% and 97.5\% Valid-Unique-Novelty scores on the planar and tree datasets, respectively. Moreover, Graph-GRPO achieves state-of-the-art performance on the molecular optimization tasks, outperforming graph-based and fragment-based RL methods as well as classic genetic algorithms.
format Preprint
id arxiv_https___arxiv_org_abs_2603_10395
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Graph-GRPO: Training Graph Flow Models with Reinforcement Learning
Zhu, Baoheng
Bo, Deyu
Zhang, Delvin Ce
Wang, Xiao
Machine Learning
Graph generation is a fundamental task with broad applications, such as drug discovery. Recently, discrete flow matching-based graph generation, \aka, graph flow model (GFM), has emerged due to its superior performance and flexible sampling. However, effectively aligning GFMs with complex human preferences or task-specific objectives remains a significant challenge. In this paper, we propose Graph-GRPO, an online reinforcement learning (RL) framework for training GFMs under verifiable rewards. Our method makes two key contributions: (1) We derive an analytical expression for the transition probability of GFMs, replacing the Monte Carlo sampling and enabling fully differentiable rollouts for RL training; (2) We propose a refinement strategy that randomly perturbs specific nodes and edges in a graph, and regenerates them, allowing for localized exploration and self-improvement of generation quality. Extensive experiments on both synthetic and real datasets demonstrate the effectiveness of Graph-GRPO. With only 50 denoising steps, our method achieves 95.0\% and 97.5\% Valid-Unique-Novelty scores on the planar and tree datasets, respectively. Moreover, Graph-GRPO achieves state-of-the-art performance on the molecular optimization tasks, outperforming graph-based and fragment-based RL methods as well as classic genetic algorithms.
title Graph-GRPO: Training Graph Flow Models with Reinforcement Learning
topic Machine Learning
url https://arxiv.org/abs/2603.10395