过渡模型：重新思考生成学习目标
Transition Models: Rethinking the Generative Learning Objective
TiM 提出了一种精确的连续时间动力学方程，可解析定义任意有限时间间隔的状态转移，使生成模型能在单步跳跃与多步细化之间无缝切换，并随采样步数增加单调提升质量。这对视觉生成研究意味着可以设计更灵活的采样策略，在少步与多步生成间取得平衡，同时支持高达 4096x4096 的原生分辨率生成。
原文依据 TiM demonstrates monotonic quality improvement as the sampling budget increases.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论生成建模中一个根本性困境持续存在：迭代扩散模型实现了出色的保真度，但计算成本显著；而高效少步替代方案则受限于硬性质量上限。生成步数与输出质量之间的这种冲突源于限制性训练目标，这些目标仅关注无穷小动力学（PF-ODE）或直接端点预测。我们通过引入一个精确的连续时间动力学方程来解决这一挑战，该方程解析地定义了任意有限时间间隔（Δt）内的状态转移。这引出了一个新的生成范式——过渡模型（TiM），它适应任意步数的过渡，从单步跳跃到更多步的细粒度细化，无缝遍历生成轨迹。尽管仅有 8.65 亿参数，TiM 实现了最优性能，在所有评估的步数上超越了 SD3.5（80 亿参数）和 FLUX.1（120 亿参数）等领先模型。重要的是，与之前的少步生成器不同，TiM 展现出随着采样预算增加而单调提升的质量改进。此外，当采用我们的原生分辨率策略时，TiM 在高达 4096x4096 的分辨率下提供了卓越的保真度。
英文原文摘要 1283 字符
A fundamental dilemma in generative modeling persists: iterative diffusion models achieve outstanding fidelity, but at a significant computational cost, while efficient few-step alternatives are constrained by a hard quality ceiling. This conflict between generation steps and output quality arises from restrictive training objectives that focus exclusively on either infinitesimal dynamics (PF-ODEs) or direct endpoint prediction. We address this challenge by introducing an exact, continuous-time dynamics equation that analytically defines state transitions across any finite time interval (\Delta t). This leads to a novel generative paradigm, Transition Models (TiM), which adapt to arbitrary-step transitions, seamlessly traversing the generative trajectory from single leaps to fine-grained refinement with more steps.Despite having only 865M parameters, TiM achieves state-of-the-art performance, surpassing leading models such as SD3.5 (8B parameters) and FLUX.1 (12B parameters) across all evaluated step counts. Importantly, unlike previous few-step generators, TiM demonstrates monotonic quality improvement as the sampling budget increases. Additionally, when employing our native-resolution strategy, TiM delivers exceptional fidelity at resolutions up to (4096x4096).