Model-generalization

Less Data Can Be Better for Domain Generalization
Generalization to unknown test distribution improves when a model is