PeRFlow: Piecewise Rectified Flow as Universal Plug-and-Play Accelerator
This work addresses the bottleneck of slow inference in diffusion models for generative AI applications, offering a plug-and-play accelerator that is compatible with various pre-trained models.
The authors tackled the problem of slow sampling in diffusion models by introducing PeRFlow, a flow-based method that accelerates generation by straightening trajectories in time windows, achieving superior performance in few-step generation with fast training convergence and transfer ability.
We present Piecewise Rectified Flow (PeRFlow), a flow-based method for accelerating diffusion models. PeRFlow divides the sampling process of generative flows into several time windows and straightens the trajectories in each interval via the reflow operation, thereby approaching piecewise linear flows. PeRFlow achieves superior performance in a few-step generation. Moreover, through dedicated parameterizations, the PeRFlow models inherit knowledge from the pretrained diffusion models. Thus, the training converges fast and the obtained models show advantageous transfer ability, serving as universal plug-and-play accelerators that are compatible with various workflows based on the pre-trained diffusion models. Codes for training and inference are publicly released. https://github.com/magic-research/piecewise-rectified-flow