无需外部验证器，通过局部重生成提升扩散模型的全局一致性。
Inference-Time Scaling in Diffusion Models through Iterative Partial Refinement
- 对已有样本局部区域重加噪并重新生成，利用上下文修正初始决策。
- 在MNIST Sudoku任务中有效解率从55.8%提升至75.0%。
- 适用于需要全局约束满足的序列化扩散模型场景，适合研究生成质量优化者。
推理时扩展已成为提升推理能力的重要方法，并被越来越多地应用于扩散模型。然而，现有的扩散模型推理时扩展方法通常依赖外部验证器或奖励模型来排序和选择样本，限制了其在无可靠评估器环境中的可扩展性。此外，尽管近期扩散模型采用区域级混合噪声条件进行序列推理，但针对该设置的推理时扩展研究仍相对不足。本文提出迭代局部精炼（IPR），一种无需外部验证器的序列扩散模型推理时扩展方法。从已生成样本出发，IPR对部分区域重新加噪并基于剩余区域重新生成，使模型在更丰富的上下文中修正早期决策。该迭代精炼过程提升了样本的全局一致性，且无需外部验证。在需满足全局约束的推理任务中，IPR表现持续提升：在MNIST Sudoku任务中，有效解率由55.8%提升至75.0%。结果表明，仅通过迭代局部精炼即可成为序列化、混合噪声设置下扩散模型有效的推理时扩展策略。代码已开源：https://github.com/ahn-ml/IPR。
原文摘要 · Abstract (English)
Inference-time scaling has emerged as a major approach for improving reasoning capabilities, and has been increasingly applied to diffusion models. However, existing inference-time scaling methods for diffusion models typically rely on external verifiers or reward models to rank and select samples, limiting their scalability to settings where such evaluators are available and reliable. Moreover, while recent diffusion models perform sequential inference with region-wise, mixed-noise conditioning, inference-time scaling tailored to this setting remains relatively underexplored. We propose Iterative Partial Refinement (IPR), an inference-time scaling method for sequential diffusion that requires no external verifier. Starting from an already-generated sample, IPR re-noises a subset of regions and regenerates them conditioned on the remaining regions, enabling the model to revise earlier decisions under a richer context than was available during the initial generation. This iterative partial refinement produces more globally consistent samples without external verification. On reasoning tasks requiring global constraint satisfaction, IPR consistently improves performance: on MNIST Sudoku, the valid solution rate increases from 55.8% to 75.0%. These results show that iterative partial refinement alone can serve as an effective inference-time scaling strategy for diffusion models in sequential, mixed-noise settings. Code is available at: https://github.com/ahn-ml/IPR
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。