arXiv：cs.LG· Zhangyong Liang, Ying Huang, Haibin Ling·· 2 天前AI 评分41
TDAction：通过训练动力学动作实现一步生成建模
One-Step Generative Modeling via Training Dynamics Action
AI 导读
研究者提出 TDAction（Training Dynamics Action），一种按共享参数实现代价选择传输目标的一步生成模型训练方法，在保留指定分布进展的同时将代价建模为软终端控制问题，并推导出闭式 Batch Tangent Action-to-Go 值。在 ImageNet 256×256 上，TDAction 无需蒸馏即达到低于 1.1 的 FID。
正文
Abstract:One-step generative models construct a static generator through iterative training-time transport. Existing transport objectives primarily assess distributional motion, although a neural generator needs to realize the requested sample displacements jointly through shared parameter updates. The training-time construction raises the question: \emph{once training becomes the iterative process that constructs the final one-step map, what to optimize: the next distributional move, or the route by which the finite generator learns the final map?} To address the question, we introduce \textbf{T}raining \textbf{D}ynamics \textbf{A}ction (\textbf{TDAction}), which selects transport targets according to local shared-parameter realization cost while retaining a prescribed level of distributional progress. We formulate the cost as a soft-terminal control problem and derive a closed-form Batch Tangent Action-to-Go value that accounts for parameter effort and terminal mismatch. The criterion captures cross-sample interactions omitted by independent pairwise costs; under isotropic mobility, the criterion agrees with quadratic Euclidean assignment for deterministic balanced couplings. Randomized tangent probes provide a low-rank implementation that constructs shared detached targets without adding an inference-time trajectory. Controlled studies examine the relationship between generator geometry, transport selection, and realized local action. On ImageNet $256\times256$, TDAction attains an FID below $1.1$ without distillation.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.00518 [cs.LG] |
| (or arXiv:2610.00518v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.00518 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zhangyong Liang [view email]
[v1]
Wed, 30 Sep 2026 18:10:48 UTC (8,417 KB)
来源：arXiv：cs.LG · arxiv.org