不重训练也能提速生成，靠的是预测终点图像。
x-Prediction Is All You Need:Training-Free Accelerated Generation via Endpoint Decodability
- 采样时直接解码中间状态，得到初始图像的最优估计。
- 在多个模型上减少20%~70%计算量，质量几乎不变。
- 无需重训练或改架构，适合快速部署的生成应用。
扩散模型和流匹配模型虽能生成高质量样本，但其ODE采样器通常需要数十到上百次神经网络函数评估（NFE）。现有加速方法多依赖重训练、蒸馏或轨迹重构，成本高。本文提出基于x- prediction的新路径：标准仿射概率路径天然包含x₀信息——中间状态与路径速度可确定清洁样本的合理估计。我们将其形式化为“终点可解码性”，并证明解码器是ℓ₂目标下的最小均方误差估计器𝔼[x₀∣xₜ]。由此提出截断跳跃采样（TJS）：在早期退出时间t*停止ODE，直接返回解码后的x₀。TJS无需重训练、蒸馏或结构修改。在SDXL、SD3.5M、Z-Image-Turbo及三个类别条件基准上，均实现20%–70%的NFE降低，且质量近似原版。分析还表明，终点预测无需拉直轨迹，可在不重构轨迹的前提下实现推理加速。
原文摘要 · Abstract (English)
Diffusion and flow matching models generate high-quality samples, but their ODE samplers often need tens to hundreds of neural function evaluations (NFEs). This remains a practical challenge for released checkpoints, since many accelerators require additional design choices and training cost through retraining, distillation, or trajectory redesign. We investigate a different route based on $x$-prediction. During sampling, standard affine probability paths already expose $x_0$ information: an intermediate state and its path velocity determine a principled estimate of the clean sample. We formalize this property as \textbf{endpoint decodability} and show that the decoder is the minimum-MSE estimator $\mathbb{E}[x_0\mid x_t]$ under the usual $\ell_2$ objective. This yields \textbf{Truncated Jump Sampling} (TJS): stop the ODE at an early-exit time $t^*$ and return the decoded $x_0$. TJS requires no retraining, distillation, or architecture change. Across SDXL, SD3.5M, Z-Image-Turbo, and three class-conditional benchmarks, it reduces NFEs by 20--70\% with near-matched quality. The analysis also shows why endpoint prediction can work without straightening the trajectory, providing inference acceleration without trajectory redesign.
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。