dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wan, Zhengyan, Ouyang, Yidong, Hu, Panwen, Sun, Qiang
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918492946563072
author Wan, Zhengyan
Ouyang, Yidong
Hu, Panwen
Sun, Qiang
author_facet Wan, Zhengyan
Ouyang, Yidong
Hu, Panwen
Sun, Qiang
contents Discrete flow models (DFMs) are a class of flexible generative models for generating discrete data, and diffusion large language models (dLLMs) can be viewed as a special case with a specific choice of mixture path and a masked source distribution. While several recent works have explored reinforcement learning into dLLMs, its application to more general discrete flow models remains underexplored. In this work, we present discrete Flow-GRPO (dFlowGRPO), a unified reinforcement learning framework for discrete flow models that supports a broad family of probability paths and non-masked source distributions. We derive the full trajectory probability for DFMs and formulate denoising as a Markov decision process, enabling dFlowGRPO to incorporate information from both the associated conditional transition rates and the posterior model during reinforcement learning. We apply dFlowGRPO to FUDOKI, a recent multimodal discrete flow model, and evaluate it on both image generation and multimodal understanding tasks. Empirical results show that dFlowGRPO outperforms existing GRPO-type methods for dLLMs on text-to-image generation tasks and achieves performance competitive with continuous flow-based models trained using FlowGRPO, while also demonstrating strong capabilities on understanding tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2605_09291
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models
Wan, Zhengyan
Ouyang, Yidong
Hu, Panwen
Sun, Qiang
Machine Learning
Applications
Discrete flow models (DFMs) are a class of flexible generative models for generating discrete data, and diffusion large language models (dLLMs) can be viewed as a special case with a specific choice of mixture path and a masked source distribution. While several recent works have explored reinforcement learning into dLLMs, its application to more general discrete flow models remains underexplored. In this work, we present discrete Flow-GRPO (dFlowGRPO), a unified reinforcement learning framework for discrete flow models that supports a broad family of probability paths and non-masked source distributions. We derive the full trajectory probability for DFMs and formulate denoising as a Markov decision process, enabling dFlowGRPO to incorporate information from both the associated conditional transition rates and the posterior model during reinforcement learning. We apply dFlowGRPO to FUDOKI, a recent multimodal discrete flow model, and evaluate it on both image generation and multimodal understanding tasks. Empirical results show that dFlowGRPO outperforms existing GRPO-type methods for dLLMs on text-to-image generation tasks and achieves performance competitive with continuous flow-based models trained using FlowGRPO, while also demonstrating strong capabilities on understanding tasks.
title dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models
topic Machine Learning
Applications
url https://arxiv.org/abs/2605.09291