Diffusion denoising increments can be treated as exchangeable, letting a single model speculate about many future steps and verify them in parallel, with a provable K^{1/3} speedup and exact sample quality.
Sampling from the sherrington-kirkpatrick gibbs measure via algorithmic stochastic localization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Diffusion Models are Secretly Exchangeable: Parallelizing DDPMs via Autospeculation
Diffusion denoising increments can be treated as exchangeable, letting a single model speculate about many future steps and verify them in parallel, with a provable K^{1/3} speedup and exact sample quality.