SOURCE-LINKED INTELLIGENCE

What Makes an Efficient VLA? Navigating Action-Head Design, Scaling, and Latency

arXiv · AI, language, vision and robotics · article · Sep 12, 2026 · UTC

Vision-Language-Action (VLA) models combine a pretrained vision encoder, a language backbone, and an action head, but their relative contribution has not been established under controlled, latency-paired conditions. We fix the backbone families (SigLIP2 and Qwen2.5) and the training pipeline, sweep action-head design and module scale, and pair each configuration with measured on-device latency. The study yields three findings. First, action-head performance is governed primarily by initialization rather than decoder architecture, loss, or inference budget: copying the last transformer layers o

Read original source ↗ Open in workspace

recordType

paper

region

Global