Bespoke Solvers for Generative Flow Models
This addresses the sampling bottleneck for users of diffusion or flow-based generative models, offering a more efficient alternative to existing methods like distillation or dedicated solvers, though it is incremental as it builds on prior solver optimization approaches.
The paper tackles the high computational cost of sampling from generative flow models by introducing Bespoke solvers, which are custom ODE solvers tailored to pre-trained models, achieving sample quality close to ground truth with significantly fewer function evaluations, such as 2.73 FID on CIFAR10 with 10 NFE and 2.2 FID on ImageNet-64x64 with 10 NFE.
Diffusion or flow-based models are powerful generative paradigms that are notoriously hard to sample as samples are defined as solutions to high-dimensional Ordinary or Stochastic Differential Equations (ODEs/SDEs) which require a large Number of Function Evaluations (NFE) to approximate well. Existing methods to alleviate the costly sampling process include model distillation and designing dedicated ODE solvers. However, distillation is costly to train and sometimes can deteriorate quality, while dedicated solvers still require relatively large NFE to produce high quality samples. In this paper we introduce "Bespoke solvers", a novel framework for constructing custom ODE solvers tailored to the ODE of a given pre-trained flow model. Our approach optimizes an order consistent and parameter-efficient solver (e.g., with 80 learnable parameters), is trained for roughly 1% of the GPU time required for training the pre-trained model, and significantly improves approximation and generation quality compared to dedicated solvers. For example, a Bespoke solver for a CIFAR10 model produces samples with Fréchet Inception Distance (FID) of 2.73 with 10 NFE, and gets to 1% of the Ground Truth (GT) FID (2.59) for this model with only 20 NFE. On the more challenging ImageNet-64$\times$64, Bespoke samples at 2.2 FID with 10 NFE, and gets within 2% of GT FID (1.71) with 20 NFE.