V-JEPA 2: Self-Supervised Video Models Enable Understanding,Prediction,Planning github.com 5 points by johlo 16 hours ago