Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow

使用校正流生成并迁移数据.

Improving the Training of Rectified Flows

改进校正流的训练.

Improving and generalizing flow-based generative models with minibatch optimal transport

通过小批量最优传输改进和泛化基于流的生成模型.

Variational Flow Matching for Graph Generation

图生成的变分流匹配.

Building Normalizing Flows with Stochastic Interpolants

通过随机插值构造归一化流.

Flow Matching for Generative Modeling

生成式建模的流匹配.