A new contrastive learning framework called VMCL improves knowledge graph embedding for unseen entities and shows smooth, predictable gains as model capacity scales up.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results