A new contrastive learning framework called VMCL improves knowledge graph embedding for unseen entities and shows smooth, predictable gains as model capacity scales up.
Results that may be inaccessible to you are currently showing.
Hide inaccessible results