ActTune: Action-Aware Precision and GPU Operating-Point Adaptation for Energy-Efficient Vision-Language-Action Inference

2026-10-06

AI Summary

This study addresses the high inference energy consumption of Vision-Language-Action (VLA) models and the task failures often caused by naive power reduction. To minimize total GPU energy per successful task, this work proposes ActTune, a framework that combines action-aware, layer-wise precision allocation with workload-driven GPU operating point selection to enable asynchronous dynamic adaptation. To enhance efficiency, ActTune incorporates decision tree learning for configuration error modeling, lookup table calibration for frequency–power mapping, and a shared resident quantized weight repository. Evaluated on the LIBERO benchmark, ActTune improves task success rate by 2.3%, achieves a 2.02× inference speedup, and reduces energy consumption per successful task by 76.8%.

Abstract

Vision-language-action (VLA) policies repeatedly invoke inference to control robots, making graphics processing unit (GPU) energy a recurring cost of task execution. Reducing energy per inference call, however, may not reduce energy per successful task if numerical errors increase failures or slower inference prolongs execution. We therefore target GPU energy per successful task while preserving task success and keeping the inference-latency increase within 10\%. Our approach builds on two observations: quantization sensitivity varies across action classes, model layers, and weights versus activations; and numerical precision changes the workload, shifting favorable GPU operating points. We introduce ActTune, an action-aware framework that connects layer-wise precision allocation with workload-dependent GPU operating-point selection over requested frequency--power-cap pairs. A lightweight decision tree learns its splits and leaf precision configurations directly from configuration action errors, then selects precision before each policy call. The controller forecasts the next workload and applies the selected GPU operating point asynchronously using a lookup table calibrated under a latency budget. A shared resident quantized weight bank enables configuration switching without weight reconstruction or additional policy evaluations. On LIBERO, a benchmark for lifelong robot learning, ActTune improves mean task success by up to 2.3\% relative to state of the art. Relative to the original BF16 implementations, it delivers up to $2.02\times$ faster inference and, with GPU operating-point adaptation, reduces energy per successful task by up to 76.8\%.

Problem

• Vision-Language-Action
• Energy Efficiency
• GPU Inference
• Quantization
• Robot Learning

Innovation

• Vision-Language-Action
• Quantization
• GPU Operating-Point Adaptation
• Energy Efficiency
• Decision Tree