Amazon SageMaker提供了一种在批量转换作业中监控模型质量的方法。您可以使用Amazon SageMaker Model Monitor来监视批处理转换作业中的模型性能并检测模型性能的漂移。以下是使用Amazon SageMaker Model Monitor监控批处理转换作业的步骤示例：

1. 定义批处理转换作业的输入和输出数据格式。

from sagemaker.model_monitor.dataset_format import DatasetFormat

data_capture_format = DatasetFormat(
    csv_format={
        "header": True,
        "delimiter": ",",
        "quoteChar": '"',
        "doubleQuote": True,
        "escapeChar": "\\",
        "nullValue": "?",
    }
)

output_kms_key = "alias/aws/sagemaker"

2. 开始数据捕获

from sagemaker.model_monitor import DataCaptureConfig

data_capture_config = DataCaptureConfig(
    enable_capture=True,
    sampling_percentage=100,
    capture_options=["REQUEST", "RESPONSE"],
    destination_s3_uri=f"s3://{s3_bucket}/{s3_prefix}",
    kms_key_id=output_kms_key,
    csv_content_types=["text/csv"],
    json_content_types=["application/json"],
    inference_response_capture_config=None,
    probability_capture_config=None,
    probability_threshold=0.0,
    anchor_threshold=None,
    capture_mode=None,
    max_request_size=None,
    max_response_size=None,
    sagemaker_capture_duration=None,
    enable_capture_upload=True,
    sampling_direction="bottom",
    json_format_for_request=None,
    json_format_for_response=None,
    dataset_format_type=None,
    dataset_format_options=None,
    csv_format_options=None,
)

transformer = model.transformer(
    instance_count=1,
    instance_type="ml.m5.large",
    strategy="SingleRecord",
    output_path=output_path,
    output_kms_key=output_kms_key,
    accept="text/csv",
    env={
        "SAGEMAKER_MODEL_SERVER_WAIT_TIME_SECONDS": "60",
    },
    max_payload=6,
    tags=[{"Key": "stack_name", "Value":