Search

Tag: #world-models97 results

C-JEPA Learns World Models via Object Masking

C-JEPA Learns World Models via Object Masking

C-JEPA extends masked joint embedding prediction to object-centric representations with object-level masking, inducing latent interventions for interaction reasoning. It boosts counterfactual VQA by 20% and enables efficient agent planning using 1% of latent features. Code is on GitHub.

ArXiv AIResearchFeb 13#research#c-jepa#world-models
Page 10 of 10