Graph out-of-distribution (OOD) generalization remains a major challenge in graph neural networks (GNNs). Invariant learning, aiming to extract invariant features across varied distributions, has ...