通过分块匹配学习更直的生成流，提升采样效率与样本质量
Block Flow: Learning Straight Flow on Data Blocks
- 利用标签将数据分块，对齐先验分布以学习低曲率流
- 调节先验方差可控制轨迹曲率上限，平衡多样性与数值误差
- 在相同参数量下性能媲美先进模型，适合生成任务研究者
流匹配模型为多种应用提供了强大框架，具备高效采样和灵活概率路径建模能力。其特点是生成轨迹曲率较低，从而降低每步采样时的截断误差。为进一步减小曲率，本文提出分块匹配（block matching）方法：利用标签信息将数据分布划分为多个块，并使用相同标签信息参数化的先验分布与之匹配，从而学习更直的流。我们证明了先验分布的方差可控制流匹配模型前向轨迹的曲率上界。通过设计灵活的正则化策略调节该方差，实现了最优生成性能，有效平衡了生成样本多样性与数值求解器误差之间的权衡。实验表明，本方法在相同参数规模下达到具有竞争力的性能。代码已开源：https://github.com/wpp13749/block_flow。
原文摘要 · Abstract (English)
Flow-matching models provide a powerful framework for various applications, offering efficient sampling and flexible probability path modeling. These models are characterized by flows with low curvature in learned generative trajectories, which results in reduced truncation error at each sampling step. To further reduce curvature, we propose block matching. This novel approach leverages label information to partition the data distribution into blocks and match them with a prior distribution parameterized using the same label information, thereby learning straighter flows. We demonstrate that the variance of the prior distribution can control the curvature upper bound of forward trajectories in flow-matching models. By designing flexible regularization strategies to adjust this variance, we achieve optimal generation performance, effectively balancing the trade-off between maintaining diversity in generated samples and minimizing numerical solver errors. Our results demonstrate competitive performance with models of the same parameter scale.Code is available at \url{https://github.com/wpp13749/block_flow}.
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。