AI RESEARCH
LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels
arXiv CS.AI
•
ArXi:2603.19312v1 Announce Type: cross Joint Embedding Predictive Architectures (JEPAs) offer a compelling framework for learning world models in compact latent spaces, yet existing methods remain fragile, relying on complex multi-term losses, exponential moving averages, pre-trained encoders, or auxiliary supervision to avoid representation collapse. In this work, we