AWS Documentation » Amazon SageMaker » Developer Guide » Get Started » Step 6: Deploy the Model to Amazon SageMaker » Step 6.2: Deploy the Model with Batch Transform

The AWS Documentation website is getting a new look!
Try it now and let us know what you think. Switch to the new look >>

You can return to the original look by selecting English in the language selector above.

Step 6.2: Deploy the Model with Batch Transform

To get inference for an entire dataset, use batch transform. Amazon SageMaker stores the results in Amazon S3.

For information about batch transforms, see Get Inferences for an Entire Dataset with Batch Transform. For an example that uses batch transform, see the batch transform sample notebook at https://github.com/awslabs/amazon-sagemaker-examples/tree/master/sagemaker_batch_transform/introduction_to_batch_transform.

Deploy a Model with Batch Transform (Amazon SageMaker High-level Python Library)

The following code creates a sagemaker.transformer.Transformer object from the model that you trained in Create and Run a Training Job (Amazon SageMaker Python SDK). Then it calls that object's transform method to create a transform job. When you create the sagemaker.transformer.Transformer object, you specify the number and type of ML instances to use to perform the batch transform job, and the location in Amazon S3 where you want to store the inferences.

Paste the following code in a cell in the Jupyter notebook you created in Step 3: Create a Jupyter Notebook and run the cell.

batch_input =
                's3://{}/{}/test/examples'.format(bucket, prefix) # The location of the test dataset

batch_output = 's3://{}/{}/batch-inference'.format(bucket, prefix) # The location to store the
results of the batch transform job

transformer = xgb_model.transformer(instance_count=1, instance_type='ml.m4.xlarge', output_path=batch_output)

transformer.transform(data=batch_input, data_type='S3Prefix', content_type='text/csv', split_type='Line')

transformer.wait()

For more information, see https://sagemaker.readthedocs.io/en/stable/transformer.html.

Next Step

Step 7: Validate the Model

Deploy a Model with Batch Transform (SDK for Python (Boto 3))

To run a batch transform job, call the create_transform_job. method using the model that you trained in Create and Run a Training Job (AWS SDK for Python (Boto 3)).

To create a batch transform job (SDK for Python (Boto 3))

For each of the following steps, paste the code in a cell in the Jupyter notebook you created in Step 3: Create a Jupyter Notebook and run the cell.

1. Name the batch transform job and specify where the input data (the test dataset) is stored and where to store the job's output.

   batch_job_name = 'xgboost-mnist-batch' + strftime("%Y-%m-%d-%H-%M-%S", gmtime())
   
   batch_input = 's3://{}/{}/test/examples'.format(bucket, prefix)
   print(batch_input)
   
   batch_output = 's3://{}/{}/batch-inference'.format(bucket, prefix)

2. Configure the parameters that you pass when you call the create_transform_job method.

   request = \
   {
       "TransformJobName": batch_job_name,
       "ModelName": model_name,
       "BatchStrategy": "MultiRecord",
       "TransformOutput": {
           "S3OutputPath": batch_output
       },
       "TransformInput": {
           "DataSource": {
               "S3DataSource": {
                   "S3DataType": "S3Prefix",
                   "S3Uri": batch_input 
               }
           },
           "ContentType": "text/csv",
           "SplitType": "Line",
           "CompressionType": "None"
       },
       "TransformResources": {
               "InstanceType": "ml.m4.xlarge",
               "InstanceCount": 1
       }
   }

   For more information about the parameters, see CreateTransformJob.

3. Call the create_transform_job method, passing in the parameters that you configured in the previous step. Then call the describe_transform_job method in a loop until it completes.

   Paste the following code in a cell in the Jupyter notebook you created in Step 3: Create a Jupyter Notebook and run the cell.

   sm.create_transform_job(**request)
                               
   while(True):
       response = sm.describe_transform_job(TransformJobName=batch_job_name)
       status = response['TransformJobStatus']
       if  status == 'Completed':
           print("Transform job ended with status: " + status)
           break
       if status == 'Failed':
           message = response['FailureReason']
           print('Transform failed with the following error: {}'.format(message))
           raise Exception('Transform job failed') 
       print("Transform job is still in status: " + status)    
       time.sleep(30) 

Next Step

Step 7: Validate the Model