2 ms·VL-JEPA: Joint Embedding Predictive Architecture for Vision-Language2 points by andsoitis 9mo ago