用三体散射机制实现单步生成，突破传统扩散模型依赖多步迭代的限制。
Three-Body Scattering for Generative Modeling
- 将能量距离转化为粒子间局部相互作用，实现样本级直接监督。
- 在ImageNet-256上达成FID=1.63（潜空间）与FID=2.23（像素空间），仅需一次采样步骤。
- 适合追求高效率生成、无需复杂训练过程的研究者与应用开发者。
现代生成模型通常依赖对抗判别器、预设噪声到数据路径或自回归分解。本文提出一种新范式：通过合适的分布能量诱导样本级运动，并为单步生成器提供直接回归监督。三体散射建模（TBSM）将能量距离转化为固定规模的每粒子相互作用——每个粒子被一个真实源吸引，同时被一个独立生成的源排斥。在给定粒子及其条件时，其期望等于$\frac{1}{2}D_E^2(P_θ,Q)$的2-Wasserstein梯度流速度。一批$B$个冻结目标事件可产生$O(B)$个样本级损失，每个使用单一参考作为条件，而非类似漂移模型中的全局全对场。在线追踪此条件期望可降低场噪声。在冻结图像特征中使用散射，TBSM在ImageNet-256上训练单步生成器，像素空间的PixelDiT-XL达到FID=2.23，潜空间的DiT-XL达到FID=1.63，NFE=1。论文还提供了一个关联扩散监督、漂移动力学与GAN目标的设计图谱。这些结果确立了追踪散射作为高维单步生成的有效路径。代码见：https://github.com/sp12138/TBSM。
原文摘要 · Abstract (English)
Modern generative models typically rely on an adversarial critic, a prescribed noise-to-data path, or an autoregressive factorization. Instead, we show that a proper distributional energy can induce sample-level motion and provide direct regression supervision for a one-step generator. Three-Body Scattering Modeling (TBSM) for generation turns the energy distance into a constant-size per-projectile interaction: each projectile is attracted toward one real source and repelled from one independently generated source. Conditioned on the projectile and its condition, its expectation equals the $2$-Wasserstein gradient-flow velocity of $\frac12D_E^2(P_θ,Q)$. A batch of $B$ frozen-target events yields $O(B)$ sample-level losses, each using one reference for its condition instead of the minibatch-wide all-pairs field used by methods such as Drifting Models. Tracking this conditional expectation online can reduce field noise. Using scattering in frozen image features, TBSM trains one-step generators on ImageNet-256, achieving FID${}=2.23$ with pixel-space PixelDiT-XL and FID${}=1.63$ with latent-space DiT-XL at NFE${}=1$. We provide a design map relating diffusion-related supervision, Drift-like dynamics, and GAN-like objectives. These results establish tracked scattering as a route to high-dimensional one-step generation. Code: https://github.com/sp12138/TBSM.
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。