https://steven-yl.github.io/mywebsite/tags/flow-matching/Flow Matching - 标签 - mywebsiteHugo -- gohugo.iozh-CNsteven@gmail.com (Steven)steven@gmail.com (Steven)This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.Wed, 25 Mar 2026 00:00:00 +0800https://steven-yl.github.io/mywebsite/kl_div/Wed, 25 Mar 2026 00:00:00 +0800Stevenhttps://github.com/steven-ylhttps://steven-yl.github.io/mywebsite/kl_div/本文把 KL 散度相关的几个核心概念串起来，给出离散流匹配中广义 KL 损失的直观解释与 PyTorch 实现示例。https://steven-yl.github.io/mywebsite/flow-matching-guide-and-code-5-non-euclidean-flow-matching/Thu, 05 Mar 2026 00:00:00 +0000Stevenhttps://github.com/steven-ylhttps://steven-yl.github.io/mywebsite/flow-matching-guide-and-code-5-non-euclidean-flow-matching/第5章 Non-Euclidean Flow Matching 解读：从动机与黎曼流形设定出发，说明流形上的流、概率路径与连续性方程，边际化技巧（定理10）、RCFM 损失（定理11），以及测地线条件流与基于预度量的条件流；并对照欧氏 FM 与代码8（球面测地线 RCFM）做小结。https://steven-yl.github.io/mywebsite/flow-matching-guide-and-code-5-flattorus-riemannian-flow-matching/Mon, 10 Jun 2024 00:00:00 +0800Stevenhttps://github.com/steven-ylhttps://steven-yl.github.io/mywebsite/flow-matching-guide-and-code-5-flattorus-riemannian-flow-matching/平坦环面 M=[0,2π)² 上 Riemannian Flow Matching 的训练目标、概率路径、损失形式及实现细节的技术文档。