FSF-DMD replaces the fake-score network in distribution matching distillation with a generator-induced pseudo-velocity surrogate for flow-map generators, showing improved FID on ImageNet-1K 256x256.
Flow straight and fast: Learning to generate and transfer data with rectified flow
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Mixing unconditional Gaussian noise with a κ-conditioned source during training of rectified flows reduces path curvature, yielding 12% better FID scores and faster sampling than standard rectified flows.
citing papers explorer
-
Distribution Matching Distillation without Fake Score Network
FSF-DMD replaces the fake-score network in distribution matching distillation with a generator-induced pseudo-velocity surrogate for flow-map generators, showing improved FID on ImageNet-1K 256x256.
-
MixFlow: Mixed Source Distributions Improve Rectified Flows
Mixing unconditional Gaussian noise with a κ-conditioned source during training of rectified flows reduces path curvature, yielding 12% better FID scores and faster sampling than standard rectified flows.