Progressive Seed Pruning—start many noise seeds, score early partially-denoised images, prune aggressively—improves prompt-aligned image generation at fixed denoising compute over best-of-N, resampling, and tree-search baselines on three diffusion/flow backbones.
Probability and its Applications
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Inference-Time Scaling of Diffusion Models via Progressive Seed Pruning
Progressive Seed Pruning—start many noise seeds, score early partially-denoised images, prune aggressively—improves prompt-aligned image generation at fixed denoising compute over best-of-N, resampling, and tree-search baselines on three diffusion/flow backbones.