通过自适应机制修复生成模型中的错误优化路径，提升少步生成质量。
Optimizing Few-Step Generation with Adaptive Matching Distillation
- 引入奖励代理检测并逃离不稳定的优化区域
- 在SDXL上将HPSv2得分从30.64提升至31.25
- 适合关注生成模型稳定性与高质量少步生成的研究者
分布匹配蒸馏（DMD）是一种强大的加速范式，但在某些区域（即‘禁用区’）其稳定性易受破坏——真实教师提供不可靠指导，虚假教师则缺乏足够排斥力。本文提出统一优化框架，将先前方法重新理解为规避这些异常区域的隐式策略。基于此，我们设计自适应匹配蒸馏（AMD），一种利用奖励代理显式探测并逃离禁用区的自校正机制。AMD通过结构信号分解动态优先处理修正梯度，并引入排斥景观锐化以增强能量壁垒，防止失败模式坍塌。在图像与视频生成任务（如SDXL、Wan2.1）及严格基准测试（如VBench、GenEval）上的大量实验表明，AMD显著提升样本保真度与训练鲁棒性。例如，在SDXL上，HPSv2得分从30.64提升至31.25，优于现有最先进方法。结果验证了在禁用区内显式纠正优化轨迹对突破少步生成模型性能上限至关重要。
原文摘要 · Abstract (English)
Distribution Matching Distillation (DMD) is a powerful acceleration paradigm, yet its stability is often compromised in Forbidden Zone, regions where the real teacher provides unreliable guidance while the fake teacher exerts insufficient repulsive force. In this work, we propose a unified optimization framework that reinterprets prior art as implicit strategies to avoid these corrupted regions. Based on this insight, we introduce Adaptive Matching Distillation (AMD), a self-correcting mechanism that utilizes reward proxies to explicitly detect and escape Forbidden Zones. AMD dynamically prioritizes corrective gradients via structural signal decomposition and introduces Repulsive Landscape Sharpening to enforce steep energy barriers against failure mode collapse. Extensive experiments across image and video generation tasks (e.g., SDXL, Wan2.1) and rigorous benchmarks (e.g., VBench, GenEval) demonstrate that AMD significantly enhances sample fidelity and training robustness. For instance, AMD improves the HPSv2 score on SDXL from 30.64 to 31.25, outperforming state-of-the-art baselines. These findings validate that explicitly rectifying optimization trajectories within Forbidden Zones is essential for pushing the performance ceiling of few-step generative models.
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。