V-JEPA 2: Self-Supervised Video Models Enable Understanding,Prediction,Planning github.com 6 points by johlo 3 days ago