Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models Paper • 2608.27550 • Published Aug 27 • 84