Researchers show that the frozen predictor from Joint-Embedding Predictive Architectures (JEPAs) can be transferred across encoder families to improve occluded feature completion. In experiments, pair…
Elon Litman published a technical walkthrough of Joint Embedding Predictive Architectures (JEPA), Yann LeCun's self-supervised learning method that trains models to predict representations of masked i…
Researchers introduced Neural Voxel Dynamics, a self-supervised framework that learns implicit 3D physics from video by lifting 2D features into a volumetric latent space. The method achieves long-ter…
Chinese physical AI start-up Fysics AI launched Fysiverse, a physics-based world model that embeds real-world physical laws into its code, diverging from data-driven approaches used by OpenAI and Meta…