arXiv：cs.CL· Rui Liu, Tong Zheng, Jindong Gu, Zhipeng Wang·· 3 小时前AI 评分43
CARE：为视觉-语言-动作模型推理加速提供认证
CARE: Certifying Acceleration for Vision-Language-Action Inference
AI 导读
针对 VLA 模型推理加速可能丢弃信息、导致原本可解任务失败的风险，研究者提出 CARE，通过相同初始条件下的配对 rollout 定义"加速诱发失败"，并在校准集上给出有限样本保证，使失败风险低于用户设定预算。
正文
Abstract:While vision-language-action (VLA) models have advanced rapidly, running them at every control step remains expensive. Prior work accelerates VLA inference using techniques like action chunking and visual-token pruning, typically evaluating based on latency and average task success. However, acceleration may discard information and break tasks the original policy would solve, a risk hidden by average metrics. Measuring these failures is challenging because action deviations compound over closed-loop trajectories, meaning task failure is only observable across full episodes. We therefore define an acceleration-induced failure via paired rollouts from identical initial conditions, tracking when the reference succeeds but the accelerated policy fails. To manage this, we introduce CARE, an approach for certified accelerator selection. CARE uses paired rollouts on a calibration set to provide finite-sample guarantees that acceleration-induced failure risk stays below a user-specified budget. It deploys the fastest certified candidate, falling back to the reference if none qualify. By relying only on terminal outcomes and measured compute, CARE applies unchanged across diverse acceleration mechanisms, while sequential testing and failure-triggered reference rollouts keep certification affordable. On four LIBERO suites with OpenVLA-OFT, CARE certifies $9.0$--$10.8\times$ speedups while guaranteeing (at $95\%$ confidence) that at least $85.8\%$ of reference-solved episodes are preserved. Under tight budgets, selectors without guarantees exceed the budget in up to $75\%$ of trials, whereas CARE stays within budget and its sequential form uses $78.9\%$ fewer rollouts than exhaustive evaluation. CARE further generalizes to flow-step reduction for $\pi_{0.5}$, and to Qwen3.5-9B and Llama-3.1-8B agents in Crafter.
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.08917 [cs.CL] |
| (or arXiv:2610.08917v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.08917 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Rui Liu [view email]
[v1]
Tue, 6 Oct 2026 18:00:05 UTC (2,105 KB)
来源：arXiv：cs.CL · arxiv.org