A differentiable search over solver coefficients and sampling timesteps produces a fast diffusion sampler that outperforms DPM-Solver++ and UniPC at 5 to 10 steps.
Bespoke solvers for generative flow models
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
abstract
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 Fr\'echet 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.
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
KSDiff generates convolutional kernels in kernel space using low-rank core tensor and factor generators with multi-head attention for fast, high-quality pansharpening.
Enforcing the exact boundary values of the velocity field in Rectified Flow improves generation FID by about 8 percent on ImageNet and stabilizes stochastic sampling.
A reparameterization recipe that lets pre-trained Stable Diffusion checkpoints be finetuned as flow matching models, giving faster convergence and better performance under parameter-efficient constraints.
An online neural contextual bandit picks a per-prompt step budget for flow generators, yielding about 2.5x faster inference with largely unchanged quality.
A speech-to-speech dialogue model that predicts mel-spectrograms with flow matching, jointly trained with discrete text tokens, achieves lower WER than a discrete speech token baseline.
Flow Matching is a generative modeling framework with mathematical foundations, design choices, extensions, and open-source PyTorch code for applications like image and text generation.
citing papers explorer
-
Differentiable Solver Search for Fast Diffusion Sampling
A differentiable search over solver coefficients and sampling timesteps produces a fast diffusion sampler that outperforms DPM-Solver++ and UniPC at 5 to 10 steps.
-
Fast Kernel-Space Diffusion for Remote Sensing Pansharpening
KSDiff generates convolutional kernels in kernel space using low-rank core tensor and factor generators with multi-head attention for fast, high-quality pansharpening.
-
Improving Rectified Flow with Boundary Conditions
Enforcing the exact boundary values of the velocity field in Rectified Flow improves generation FID by about 8 percent on ImageNet and stabilizes stochastic sampling.
-
Diff2Flow: Training Flow Matching Models via Diffusion Model Alignment
A reparameterization recipe that lets pre-trained Stable Diffusion checkpoints be finetuned as flow matching models, giving faster convergence and better performance under parameter-efficient constraints.
-
Contextual Flow Matching: Adaptive Step Selection in Flow Models for Efficient Visual Generation
An online neural contextual bandit picks a per-prompt step budget for flow generators, yielding about 2.5x faster inference with largely unchanged quality.
-
Continuous Speech Tokens Makes LLMs Robust Multi-Modality Learners
A speech-to-speech dialogue model that predicts mel-spectrograms with flow matching, jointly trained with discrete text tokens, achieves lower WER than a discrete speech token baseline.
-
Flow Matching Guide and Code
Flow Matching is a generative modeling framework with mathematical foundations, design choices, extensions, and open-source PyTorch code for applications like image and text generation.