编写的代码需能够接收服务端下发的下列参数：

• type：表示任务类型，取值"training"、"evaluation"。training表示训练任务， evaluation表示评估任务。
• train_url：训练模型输出路径。
• data_url：数据集路径。
• model_path：common model路径。
• subModelLearningRate：学习率。
• subModelIterateTimes：迭代次数。

def parse_args() -> argparse.Namespace:
    """
    Parse args on edge side, DO NOT modify.
    args:
    type: training or testing, directed by server
    model_path: path to store common model
    data_url: path to fetch data, usually in local
    train_url: path to save model after training
    subModelLearningRate: learning rate
    subModelIterateTimes: number of batches per local train
    """

    parser = argparse.ArgumentParser(description="Federated Learning",
                                     formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    parser.add_argument("--type", type=str, default="training", choices=["training", "evaluation"])
    parser.add_argument("--model_path", type=str, help="the path to store common model")
    parser.add_argument("--data_url", type=str, help="the path to fetch data")
    parser.add_argument("--train_url", type=str, help="the path to save model after training")
    parser.add_argument("--subModelLearningRate", type=float, default=0.01)
    parser.add_argument("--subModelIterateTimes", type=int, default=1)
    args, _ = parser.parse_known_args()

    return args