Metraipolis › archive › dc_e98ef8d65aa8
LC-GRPO: Bridging Train-Inference Gap for Flow-Based GRPO with Langevin Correction — arXiv:2608.05600
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.05600
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T10:37:54.510Z
- SHA-256 of the archived text
4c042f5f87e214d4ba10d4d3499952a9e901d049cc0c7bd2575f38e9caedaf33- Stable URL of this page
https://metraipolis.com/documento/dc_e98ef8d65aa8
How to cite
LC-GRPO: Bridging Train-Inference Gap for Flow-Based GRPO with Langevin Correction — arXiv:2608.05600 — arXiv (en), https://arxiv.org/abs/2608.05600 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e98ef8d65aa8 (Metraipolis archive dc_e98ef8d65aa8, sha256 4c042f5f87e214d4ba10d4d3499952a9e901d049cc0c7bd2575f38e9caedaf33)
BibTeX
@misc{dce98ef8d65aa8,
title = {LC-GRPO: Bridging Train-Inference Gap for Flow-Based GRPO with Langevin Correction — arXiv:2608.05600},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.05600},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e98ef8d65aa8; sha256 4c042f5f87e214d4ba10d4d3499952a9e901d049cc0c7bd2575f38e9caedaf33},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.05600. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Flow-based generative models are typically sampled by solving a deterministic ordinary differential equation (ODE), whereas reinforcement learning (RL) requires stochastic rollouts for exploration and optimization. Existing online RL methods for flow models therefore replace the inference-time ODE with a stochastic differential equation (SDE) during training. Although the ODE and SDE share the same marginal distribution in continuous time, they can behave differently after finite-step discretization. In particular, SDE rollouts often become blurry as exploration noise increases, creating a mismatch between training rollouts and inference-time ODE samples. We introduce LC-GRPO, a flow-based GRPO framework with Langevin correction. Each rollout transition takes an inference-aligned ODE Euler step followed by a stochastic Langevin correction targeting the marginal at the resulting timestep. The required score is recovered directly from the flow velocity, requiring no additional score model, while the transition remains an isotropic Gaussian with tractable likelihood. We theoretically show that, under suitable conditions, a single Langevin correction step reduces the error of an imperfect ODE Euler step and, under matched randomness, yields a more accurate transition than standard Euler--Maruyama discretization of the reverse SDE. Additionally, experiments on SD3.5-Medium, FLUX.1-Dev, and HunyuanVideo show that LC-GRPO consistently improves reward optimization across text-to-image and text-to-video tasks, preserves generation quality, and substantially narrows the gap between stochastic training rollouts and deterministic ODE inference. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-07 in cs.CV). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2608.05600 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T10:23:31.081Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-08-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).