A prompt-adaptive model selects the minimum denoising step count that reaches perceptual quality saturation, saving about 5 seconds per image with no detectable quality loss.
Speed is all you need: On-device acceleration of large diffu- sion models via gpu-aware optimizations
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
BudgetFusion: Perceptually-Guided Adaptive Diffusion Models
A prompt-adaptive model selects the minimum denoising step count that reaches perceptual quality saturation, saving about 5 seconds per image with no detectable quality loss.