Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models Paper • 2608.27550 • Published 7 days ago • 86