通过渐进式剪枝，用固定算力提升扩散模型生成质量。
Inference-Time Scaling of Diffusion Models via Progressive Seed Pruning
- 早期评估多组噪声种子，逐步剪枝保留优质路径。
- 在相同计算量下，生成图像的提示对齐度更高。
- 适合追求高质量生成且算力受限的场景。
扩散模型与流匹配模型主导条件图像生成，但其推理阶段的扩展性远不及自回归语言模型。由于最终生成质量高度依赖初始噪声种子，现有方法通常通过黑箱奖励进行种子搜索或重采样，但保持恒定内存占用。本文提出，放宽此限制可开辟一条未被充分探索的推理阶段扩展路径：通过前置探索，在早期评估大量种子并激进剪枝，使固定算力更高效利用。提出的「渐进式种子剪枝」（Progressive Seed Pruning, PSP）对中间去噪结果评分，逐步缩小候选集，仅对有潜力的轨迹完成全去噪过程，同时保持总模型调用次数不变。在多种扩散与流匹配骨干网络上，PSP 均优于 Best-of-N、重要性采样与树搜索基线，实现更高的 GenEval 自动评分与人类评估的提示对齐度，且计算开销相当。
原文摘要 · Abstract (English)
Diffusion and flow-matching models dominate conditional image generation, yet inference-time scaling for these models is far less developed than for autoregressive language models. Because final quality is highly sensitive to the initial noise seed, many approaches spend extra compute on seed search or resampling under a black-box reward, but typically maintaining a constant memory footprint throughout inference. We show that relaxing this constraint enables an underexplored inference-time scaling axis: by front-loading exploration, evaluating many seeds early, and pruning aggressively, we can use a fixed compute budget more effectively. \emph{Progressive Seed Pruning} (\PSP) scores intermediate denoised estimates and progressively narrows the candidate set so that only promising trajectories are fully denoised, while keeping the total number of model evaluations fixed. Across diffusion and flow-matching backbones, \PSP \ consistently improves reward-guided selection and achieves higher GenEval scores (automated) and better human evaluation on prompt-alignment than best-of-$N$, importance-sampling, and tree-search baselines at matched compute. Project page: https://www.vision.caltech.edu/psp. Code: https://github.com/rogerioagjr/psp.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。