过渡模型：重新思考生成式学习目标

摘要

生成模型领域存在一个根本性难题：迭代扩散模型虽能实现卓越的保真度，却需付出巨大的计算代价；而高效的少步生成方案则受限于难以突破的质量天花板。这一生成步数与输出质量之间的矛盾，源于训练目标过于局限，要么专注于无限小动态（PF-ODEs），要么仅着眼于直接预测终点。针对这一挑战，我们提出了一种精确的连续时间动态方程，它能解析地定义任意有限时间间隔内的状态转移。由此，我们开创了一种新的生成范式——过渡模型（Transition Models, TiM），它能够适应任意步数的转移，在生成轨迹上自如穿梭，从单步跨越到多步精细优化。尽管仅拥有8.65亿参数，TiM在所有评估步数下均实现了业界领先的性能，超越了如SD3.5（80亿参数）和FLUX.1（120亿参数）等顶尖模型。尤为重要的是，与以往的少步生成器不同，TiM在采样预算增加时展现出单调的质量提升。此外，采用我们的原生分辨率策略时，TiM在高达4096x4096的分辨率下仍能提供卓越的保真度。

English

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. 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.

过渡模型：重新思考生成式学习目标

Transition Models: Rethinking the Generative Learning Objective

摘要

Support