CAR-Flow：条件感知重参数化对齐源域与目标域，优化流匹配

摘要

条件生成建模旨在从包含数据-条件对的样本中学习条件数据分布。为此，基于扩散和流的方法已取得了引人注目的成果。这些方法利用一个学习到的（流）模型，将初始的标准高斯噪声（忽略条件）传输到条件数据分布。因此，模型需要同时学习质量传输和条件注入。为了减轻模型的学习负担，我们提出了条件感知重参数化流匹配（CAR-Flow）——一种轻量级的学习偏移，用于对源分布、目标分布或两者进行条件化处理。通过重新定位这些分布，CAR-Flow缩短了模型必须学习的概率路径，从而在实践中加速了训练。在低维合成数据上，我们可视化和量化了CAR的效果。在高维自然图像数据（ImageNet-256）上，为SiT-XL/2配备CAR-Flow将FID从2.07降至1.68，同时仅引入不到0.6%的额外参数。

English

Conditional generative modeling aims to learn a conditional data distribution from samples containing data-condition pairs. For this, diffusion and flow-based methods have attained compelling results. These methods use a learned (flow) model to transport an initial standard Gaussian noise that ignores the condition to the conditional data distribution. The model is hence required to learn both mass transport and conditional injection. To ease the demand on the model, we propose Condition-Aware Reparameterization for Flow Matching (CAR-Flow) -- a lightweight, learned shift that conditions the source, the target, or both distributions. By relocating these distributions, CAR-Flow shortens the probability path the model must learn, leading to faster training in practice. On low-dimensional synthetic data, we visualize and quantify the effects of CAR. On higher-dimensional natural image data (ImageNet-256), equipping SiT-XL/2 with CAR-Flow reduces FID from 2.07 to 1.68, while introducing less than 0.6% additional parameters.

CAR-Flow：条件感知重参数化对齐源域与目标域，优化流匹配

CAR-Flow: Condition-Aware Reparameterization Aligns Source and Target for Better Flow Matching

摘要

Support