V-JEPA 2: Self-Supervised Video Models Enable Understanding,Prediction,Planning github.com 4 points by johlo 8 hours ago