AI RESEARCH
Subspace-Decomposed JEPAs: Disentangling Progression and Content in Latent World Models
arXiv CS.LG
•
ArXi:2605.31111v1 Announce Type: new Joint-Embedding Predictive Architectures (JEPAs) learn compact latent world models by predicting future embeddings, but no single coordinate of the latent is designated to encode task progression. We carve the JEPA latent into two orthogonal subspaces with disjoint roles: a low-dimensional progression subspace shaped by a cosine-margin triplet loss, and a high-dimensional content subspace regularised by the existing SIGReg objective of LeWM.