V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning Paper โข 2506.09985 โข Published Jun 11 โข 29 โข 2