{
  "abstract": "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 a mixture path and a masked source distribution. While several recent works have explored reinforcement learning for 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 the denoising process 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 Flow-GRPO, while also demonstrating strong capabilities on understanding tasks.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2605.09291",
  "author_total": 4,
  "authors": [
    {
      "name": "Zhengyan Wan"
    },
    {
      "name": "Yidong Ouyang"
    },
    {
      "name": "Panwen Hu"
    },
    {
      "name": "Qiang Sun"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/00htg3twy00ea.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2605.09291",
  "fulltext_url": null,
  "id": "00htg3twy00ea",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2605.09291",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.AP"
    ],
    [
      "openalex",
      "1707|Generative Adversarial Networks and Image Synthesis"
    ]
  ],
  "openalex": "W7160905945",
  "pdf_url": "https://arxiv.org/pdf/2605.09291",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-01",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/00htg3twy00ea.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:26:14Z",
      "native_id": "2605.09291",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2605.09291"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.AP",
      "subfield": "statistics"
    }
  ],
  "title": "dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models",
  "title_text": "dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models",
  "type": "preprint",
  "updated_at": "2026-10-30T12:26:14Z",
  "url": "https://research.draconitic.com/paper/00htg3twy00ea",
  "venue": "arXiv",
  "versions": []
}