验证了视觉-语言-动作模型中任务向量相减的局部性失效，发现其在机器人控制中不可靠。
Suppression Sticks, Locality Is Fragile: A Closed-Loop Target-and-Control Audit of Task-Vector Negation in VLA Policies
- 通过目标与控制双维度审计，检验任务向量相减对多技能机器人行为的影响
- 10个任务中5个可抑制目标、3个具抗性、2个导致全局崩溃，控制性能平均保留52%
- 结果表明该方法快速但脆弱，适合需快速编辑的场景，但需闭环测试验证
任务向量运算提供了一种模型修改的闭式方法，但在闭环机器人控制中的行为局部性仍不明确。本文对多任务视觉-语言-动作（VLA）策略中每项技能的任务向量减法进行了目标与控制双重审计。在全部10个LIBERO-Goal任务中，减法产生三种定性不同的行为模式：5个任务呈现目标控制分离，3个表现出抵抗，2个导致全局崩溃。在未见初始状态上，5个可抑制目标成功率降至0%；然而，基准归一化的控制保留率仅为52%，且每个目标抑制操作均对至少一项名义无关的控制造成实质性损害。额外的实验显示，在连续回归、离散令牌和流匹配动作头策略中存在分离现象，而在空间、物体及长时程面板上则观察到控制崩溃。任务向量余弦均值无法解释此差异。匹配范数控制揭示了单一目标锚点附近的局部符号不对称性，多向量结果随锚点和尺度变化。保留感知梯度基线提供数据依赖性比较器，但需编辑时的数据与优化；减法仅在编辑时刻无需数据与梯度，前提是专家增量已预计算。最后，单任务重学习探测一致支持行为掩蔽而非确证遗忘。这些结果将任务向量减法刻画为一种快速但脆弱的干预手段，并强调在具身模型编辑评估中必须进行闭环目标与控制评估。
原文摘要 · Abstract (English)
Task-vector arithmetic offers a closed-form way to modify a model, yet its behavioral locality remains unclear in closed-loop robot control. We present a target-and-control audit of per-skill task-vector subtraction from multitask vision-language-action (VLA) policies. Across all ten LIBERO-Goal skills, subtraction produces three qualitatively different regimes: target-control separation for five skills, resistance for three, and global collapse for two. On held-out initial states, the five suppressible targets remain at 0% success; however, mean baseline-normalized control retention is only 52%, and each target-suppressing edit materially harms at least one nominally unrelated control. Additional Goal panels show separation across tested policies with continuous-regression, discrete-token, and flow-matching action heads, whereas we observe no clean separation on Spatial and control collapse on the tested Object and Long-horizon panels. Mean task-vector cosine does not account for this variation. A matched-norm control identifies a local sign asymmetry around one Goal anchor, while multi-vector outcomes vary with anchor and scale. Retain-aware gradient baselines provide data-dependent comparators but require removal-time data and optimization; subtraction is data- and gradient-free only at edit time, assuming precomputed expert deltas. Finally, a single-skill relearning probe is consistent with behavioral masking, not certified unlearning. These results characterize task-vector subtraction as a fast but brittle intervention and underscore the need for closed-loop target-and-control evaluation when assessing locality in embodied model editing.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。