通过离散化优化流程，大幅降低一步生成模型训练成本。
Discrete Meanflow Training Curriculum
- 提出离散均流训练课程，利用一致性特性提升训练稳定性
- 仅用2000轮即在CIFAR-10上达到3.36的一步FID
- 适合希望高效训练一步生成模型的研究者
基于流的图像生成模型在多步采样中表现稳定且生成质量高。单步生成模型虽能生成高质量图像，但优化困难，常伴随训练不稳定问题。均流模型在少步采样中表现优异，且具备诱人的单步采样潜力。值得注意的是，实现这一性能的均流模型通常需要极高的训练计算与数据开销。本文通过观察并利用均流目标的一种特定离散化形式所具有的一致性性质，提出了“离散均流”（DMF）训练课程。在预训练流模型初始化下，该方法仅需2000个训练周期即可在CIFAR-10上达到3.36的一步FID。我们预计，基于现有流模型微调的快速训练课程将推动未来单步生成模型的高效训练方法发展。
原文摘要 · Abstract (English)
Flow-based image generative models exhibit stable training and produce high quality samples when using multi-step sampling procedures. One-step generative models can produce high quality image samples but can be difficult to optimize as they often exhibit unstable training dynamics. Meanflow models exhibit excellent few-step sampling performance and tantalizing one-step sampling performance. Notably, MeanFlow models that achieve this have required extremely large training budgets. We significantly decrease the amount of computation and data budget it takes to train Meanflow models by noting and exploiting a particular discretization of the Meanflow objective that yields a consistency property which we formulate into a ``Discrete Meanflow'' (DMF) Training Curriculum. Initialized with a pretrained Flow Model, DMF curriculum reaches one-step FID 3.36 on CIFAR-10 in only 2000 epochs. We anticipate that faster training curriculums of Meanflow models, specifically those fine-tuned from existing Flow Models, drives efficient training methods of future one-step examples.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。