V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning Paper • 2603.14482 • Published 21 days ago • 26