[2210.01708] Exploring Parameter-Efficient Fine-Tuning to Enable Foundation Models in Federated Learning