Python 类型错误：属性类型的对象不是 JSON 可序列化的(Sagemaker 管道)

我正在try 建立一个Sagemaker管道，有2个步骤:预处理，然后训练一个射频模型. 第一步产生3个输出:scaled_data.csv、train.csv和est.csv.第二步是训练和测试CSV来训练射频模型.在运行步骤2时出现错误，指出"TypeError:类型属性的对象不是JSON Serializable".

以下是我设置管道步骤的代码:

# upload data from local path to default bucket with prefix raw_data
WORK_DIRECTORY = "data"

input_data = sagemaker_session.upload_data(
    path="{}/{}".format(WORK_DIRECTORY, "dataset.csv"),
    bucket=bucket,
    key_prefix="{}/{}".format(prefix, "input_data"),
)

1. 设置第一步(扩展步骤)

scaling_processor = SKLearnProcessor(
    framework_version=FRAMEWORK_VERSION,
    instance_type="ml.m5.4xlarge",
    instance_count=processing_instance_count,
    base_job_name="data-process",
    role=role,
    sagemaker_session=pipeline_session,
)

scaling_processor_args = scaling_processor.run(
    inputs=[
        ProcessingInput(source=input_data, destination="/opt/ml/processing/input"),
    ],
    outputs=[
        ProcessingOutput(output_name="scaled_data", source="/opt/ml/processing/output/scaled_data/"),
        ProcessingOutput(output_name="train", source="/opt/ml/processing/output/train/"),
        ProcessingOutput(output_name="test", source="/opt/ml/processing/output/test/")    ],
    
    code="scripts/preprocess.py",
)

step_process = ProcessingStep(name="DataProcess", step_args=scaling_processor_args)

2. 设置第二步(RF训练-BYO模式)，这里是出现错误的地方:

estimator_cls = sagemaker.sklearn.SKLearn
FRAMEWORK_VERSION = "0.23-1"

rf_processor = FrameworkProcessor(
    estimator_cls,
    FRAMEWORK_VERSION,
    role = role,
    instance_count=1,
    instance_type='ml.m5.2xlarge',
    base_job_name='rf-modelling'
)

rf_processor_args = rf_processor.run(
    inputs=[
        ProcessingInput(source=step_process.properties.ProcessingOutputConfig.Outputs["train"].S3Output.S3Uri,
                        destination="/opt/ml/processing/input"),
        ProcessingInput(source=step_process.properties.ProcessingOutputConfig.Outputs["test"].S3Output.S3Uri,
                        destination="/opt/ml/processing/input"),    ],
    outputs=[
        ProcessingOutput(output_name="rf_model",source = "/opt/ml/processing/output/")
   ],
    
    code="scripts/train.py",
)

step_train = ProcessingStep(name="RFTrain", step_args=rf_processor_args)

在运行步骤2时出现错误，指出"TypeError:类型属性的对象不是JSON Serializable".问题出在我为RF_PROCESSOR_ARGS中的第二步设置处理输入的那一行. 你知道是什么原因导致这个错误吗？