Does VLA Even Know the Basics? Measuring Commonsense and World Knowledge Retention in Vision-Language-Action Models Paper • 2606.19297 • Published 17 days ago • 65