Concept
Rectified flow
Rectified flow is a way to train generative models by learning to move samples along straight paths between a noise distribution and the data distribution. Like diffusion models, it maps Gaussian noise to data, except instead of reversing a curved stochastic process, it fits…
The rest of “Rectified flow” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→