arXiv cs.AI· Yun Zhu, Da Wang·· 13 小时前AI 评分38
TAD：仅用教师轨迹蒸馏视觉-语言-动作策略
TAD: Distilling Vision-Language-Action Policies from Teacher Trajectories Alone
AI 导读
TAD（Trajectory-only Adversarial Distillation）提出仅凭固定教师轨迹特征库蒸馏 VLA 策略，学生无需在线查询教师或环境任务奖励。
正文 · 原文
Abstract:Online distillation of vision-language-action (VLA) policies commonly relies on teacher action scores that an inference-only interface may not expose. Trajectory-only Adversarial Distillation (\gad{}) instead turns a fixed bank of teacher joint-position features into adaptive feedback: a discriminator compares feature marginals from teacher and student rollouts and supplies rewards for on-policy learning. The student interacts with the environment without online teacher queries or environment task rewards. On ManiSkill \texttt{PutOnPlateInScene25Main}, \gad{} improves OpenVLA-OFT (7B) from $26.2\%$ to $76.6\%$ \successonce{} ($+50.4$ percentage points) and from $20.3\%$ to $58.6\%$ \successend{}. It trails white-box on-policy distillation (OPD) at $81.6\%$ / $59.4\%$ by $5.0$ / $0.8$ percentage points ($n{=}256$ evaluation trajectories per checkpoint). The tested four-epoch continuation achieves the highest \gad{} result without additional teacher data. This best run uses $5\times$ OPD's student interaction budget and $12.5\times$ its complete-rollout batch-passes. The comparison demonstrates online VLA improvement without teacher scoring while quantifying unequal interaction and optimization budgets.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.33650 [cs.AI] |
| (or arXiv:2610.33650v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.33650 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Da Wang [view email]
[v1]
Sat, 24 Oct 2026 13:05:18 UTC (279 KB)
来源：arXiv cs.AI · arxiv.org