{
  "abstract": "Preference alignment for flow and diffusion models now spans online reinforcement learning and offline preference optimization, but the relation between these methods remains unclear. In particular, existing forward-process alignment methods require fresh samples from the current model, while offline methods based on fixed preference pairs rely primarily on positive-only fine-tuning or DPO-style likelihood-ratio surrogates. We organize these approaches through a divergence-based framework and introduce FlowCPO, an offline forward-KL objective that uses both preferred and dispreferred samples without online rollouts. For linear interpolation, we show under explicit regularity conditions that the forward-KL objective is bounded by a contrastive flow matching loss, yielding a tractable surrogate on fixed data. We further show that this loss is nonnegative, whereas the signed regression loss of simplified FlowDPO can be unbounded below. In the in-domain setting, FlowCPO achieves higher mean GenEval and OCR scores than the evaluated baselines, reaching 0.84 and 0.87 versus 0.81 and 0.74 for FlowDPO at CFG 3.0. In the out-of-domain setting, the results are mixed, with the best GenEval result but lower reward scores than RFT on several metrics.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.09905",
  "author_total": 7,
  "authors": [
    {
      "name": "Yansen Han"
    },
    {
      "name": "Shengyi Liao"
    },
    {
      "name": "Peng Sun"
    },
    {
      "name": "Deyuan Liu"
    },
    {
      "name": "Yuanxing Zhang"
    },
    {
      "name": "Pengfei Wan"
    },
    {
      "name": "Tao Lin"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/27c3236xcv7nz.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.09905",
  "fulltext_url": null,
  "id": "27c3236xcv7nz",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.09905",
  "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.AI"
    ],
    [
      "arxiv",
      "cs.CV"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1707|Generative Adversarial Networks and Image Synthesis"
    ]
  ],
  "openalex": "W7212149694",
  "pdf_url": "https://arxiv.org/pdf/2609.09905",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-09",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/27c3236xcv7nz.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:20:22Z",
      "native_id": "2609.09905",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.09905"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models",
  "title_text": "FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models",
  "type": "preprint",
  "updated_at": "2026-10-30T12:20:22Z",
  "url": "https://research.draconitic.com/paper/27c3236xcv7nz",
  "venue": "arXiv",
  "versions": []
}