run_batch_prediction_job.py

import sagemaker as sage
sess = sage.Session()

transformer = sagemaker.transformer.Transformer(
    base_transform_job_name='Batch-Transform',
    model_name='sagemaker-demo-[...]',  # insert model name here
    instance_count=1,
    instance_type='ml.c4.xlarge',
    output_path='s3://quilt-example/quilt_sagemaker_demo/outputs',
    sagemaker_session=sess
)

# start an example prediction job
transformer.transform(
    's3://quilt-example/quilt/fashion_mnist/fashion-mnist_train.csv', 
    content_type='text/csv', 
    split_type='Line'
)

# wait until prediction job is completed
transformer.wait()

# get the data in the read the resulting file
import boto3
s3_client = boto3.resource('s3')
s3_client.download_file('s3://quilt-example/', 'quilt_sagemaker_demo/model/[...]')