arXiv：cs.AI· Wei-Jin Huang, Yuan-Ming Li, Kun-Yu Lin, Wang Luo, Yinlin Zhu, Yue Yu, Shenghao Ye, Junbin Yuan, Fa-Ting Hong, Qing Zhang, Wei-Shi Zheng·· 5 小时前AI 评分39
TACD：通过终端放大控制蒸馏高效文本到动作模型
TACD: Distilling Efficient Text-to-Motion Models via Terminal Amplification Control
AI 导读
TACD 是一种无需真实动作训练数据的 on-policy 蒸馏方法，可从文本提示和预训练教师模型训练高效动作生成器。它通过将最新教师查询与学生步长绑定，限制干净动作空间的损失权重，在 HumanML3D 和 KIT-ML 上实现少步生成改进，八步 HY-Motion 学生 FID 较无此约束的蒸馏降低 58%。
正文
Authors:Wei-Jin Huang, Yuan-Ming Li, Kun-Yu Lin, Wang Luo, Yinlin Zhu, Yue Yu, Shenghao Ye, Junbin Yuan, Fa-Ting Hong, Qing Zhang, Wei-Shi Zheng
Abstract:Recent text-to-motion models have improved motion quality and instruction following, yet many-step denoising and large model components make deployment slow and memory-intensive. We present Terminal-Amplification-Controlled Distillation (TACD), an on-policy approach for training efficient motion generators from text prompts and pretrained teachers, without real-motion training data. Building on segmented on-policy flow distillation, we supervise clean-motion predictions along student-generated trajectories. We identify a failure mode in which velocity matching on a fixed supervision grid repeatedly overweights errors near the denoising endpoint, degrading few-step generation. TACD ties the latest teacher query to the student's step size, bounding the effective loss weights in clean-motion space without changing inference. Experiments on HumanML3D and KIT-ML demonstrate improved few-step generation, including a 58% reduction in eight-step HY-Motion student FID relative to distillation without this bound. For diffusion teachers, the endpoint-matching form of TACD yields four-step students with lower FID and matched or improved text-motion retrieval relative to their 50-step teachers on HumanML3D. On HY-Motion and Kimodo, eight-step students with compact components achieve 7.7-11.9x end-to-end speedups and reduce peak GPU memory by 3.8-6.7x relative to their teachers. Project page: this https URL
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.02867 [cs.AI] |
| (or arXiv:2610.02867v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02867 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Wei-Jin Huang [view email]
[v1]
Fri, 2 Oct 2026 06:06:36 UTC (4,162 KB)
来源：arXiv：cs.AI · arxiv.org