Deploying to TensorFlow Serving Endpoints

Deploying from an Estimator

After a TensorFlow estimator has been fit, it saves a TensorFlow SavedModel bundle in the S3 location defined by output_path. You can call deploy on a TensorFlow estimator object to create a SageMaker Endpoint:

from sagemaker.tensorflow import TensorFlow

estimator = TensorFlow(
    entry_point="tf-train.py",
    ...,
    instance_count=1,
    instance_type="ml.c4.xlarge",
    framework_version="2.2",
    py_version="py37",
)

estimator.fit(inputs)

predictor = estimator.deploy(initial_instance_count=1, instance_type="ml.c5.xlarge")

The code block above deploys a SageMaker Endpoint with one instance of the type “ml.c5.xlarge”.

What happens when deploy is called

Calling deploy starts the process of creating a SageMaker Endpoint. This process includes the following steps.

  • Starts initial_instance_count EC2 instances of the type instance_type.

  • On each instance, it will do the following steps:

    • start a Docker container optimized for TensorFlow Serving, see SageMaker TensorFlow Serving containers.

    • start a TensorFlow Serving process configured to run your model.

    • start an HTTP server that provides access to TensorFlow Server through the SageMaker InvokeEndpoint API.

When the deploy call finishes, the created SageMaker Endpoint is ready for prediction requests. The Making predictions against a SageMaker Endpoint section will explain how to make prediction requests against the Endpoint.

Deploying directly from model artifacts

If you already have existing model artifacts in S3, you can skip training and deploy them directly to an endpoint:

from sagemaker.tensorflow import TensorFlowModel

model = TensorFlowModel(model_data='s3://mybucket/model.tar.gz', role='MySageMakerRole')

predictor = model.deploy(initial_instance_count=1, instance_type='ml.c5.xlarge')

Python-based TensorFlow serving on SageMaker has support for Elastic Inference, which allows for inference acceleration to a hosted endpoint for a fraction of the cost of using a full GPU instance. In order to attach an Elastic Inference accelerator to your endpoint provide the accelerator type to accelerator_type to your deploy call.

from sagemaker.tensorflow import TensorFlowModel

model = TensorFlowModel(model_data='s3://mybucket/model.tar.gz', role='MySageMakerRole')

predictor = model.deploy(initial_instance_count=1, instance_type='ml.c5.xlarge', accelerator_type='ml.eia1.medium')

Making predictions against a SageMaker Endpoint

Once you have the Predictor instance returned by model.deploy(...) or estimator.deploy(...), you can send prediction requests to your Endpoint.

The following code shows how to make a prediction request:

input = {
  'instances': [1.0, 2.0, 5.0]
}
result = predictor.predict(input)

The result object will contain a Python dict like this:

{
  'predictions': [3.5, 4.0, 5.5]
}

The formats of the input and the output data correspond directly to the request and response formats of the Predict method in the TensorFlow Serving REST API.

If your SavedModel includes the right signature_def, you can also make Classify or Regress requests:

# input matches the Classify and Regress API
input = {
  'signature_name': 'tensorflow/serving/regress',
  'examples': [{'x': 1.0}, {'x': 2.0}]
}

result = predictor.regress(input)  # or predictor.classify(...)

# result contains:
{
  'results': [3.5, 4.0]
}

You can include multiple instances in your predict request (or multiple examples in classify/regress requests) to get multiple prediction results in one request to your Endpoint:

input = {
  'instances': [
    [1.0, 2.0, 5.0],
    [1.0, 2.0, 5.0],
    [1.0, 2.0, 5.0]
  ]
}
result = predictor.predict(input)

# result contains:
{
  'predictions': [
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5]
  ]
}

If your application allows request grouping like this, it is much more efficient than making separate requests.

Other input formats

SageMaker’s TensforFlow Serving endpoints can also accept some additional input formats that are not part of the TensorFlow REST API, including a simplified json format, line-delimited json objects (“jsons” or “jsonlines”), and CSV data.

Simplified JSON Input

The Endpoint will accept simplified JSON input that doesn’t match the TensorFlow REST API’s Predict request format. When the Endpoint receives data like this, it will attempt to transform it into a valid Predict request, using a few simple rules:

  • python value, dict, or one-dimensional arrays are treated as the input value in a single ‘instance’ Predict request.

  • multidimensional arrays are treated as a multiple values in a multi-instance Predict request.

Combined with the client-side Predictor object’s JSON serialization, this allows you to make simple requests like this:

input = [
  [1.0, 2.0, 5.0],
  [1.0, 2.0, 5.0]
]
result = predictor.predict(input)

# result contains:
{
  'predictions': [
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5]
  ]
}

Or this:

# 'x' must match name of input tensor in your SavedModel graph
# for models with multiple named inputs, just include all the keys in the input dict
input = {
  'x': [1.0, 2.0, 5.0]
}

# result contains:
{
  'predictions': [
    [3.5, 4.0, 5.5]
  ]
}

Line-delimited JSON

The Endpoint will accept line-delimited JSON objects (also known as “jsons” or “jsonlines” data). The Endpoint treats each line as a separate instance in a multi-instance Predict request. To use this feature from your python code, you need to create a Predictor instance that does not try to serialize your input to JSON:

# create a Predictor without JSON serialization

predictor = Predictor('endpoint-name', serializer=None, content_type='application/jsonlines')

input = '''{'x': [1.0, 2.0, 5.0]}
{'x': [1.0, 2.0, 5.0]}
{'x': [1.0, 2.0, 5.0]}'''

result = predictor.predict(input)

# result contains:
{
  'predictions': [
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5]
  ]
}

This feature is especially useful if you are reading data from a file containing jsonlines data.

CSV (comma-separated values)

The Endpoint will accept CSV data. Each line is treated as a separate instance. This is a compact format for representing multiple instances of 1-d array data. To use this feature from your python code, you need to create a Predictor instance that can serialize your input data to CSV format:

# create a Predictor with JSON serialization

predictor = Predictor('endpoint-name', serializer=sagemaker.serializers.CSVSerializer())

# CSV-formatted string input
input = '1.0,2.0,5.0\n1.0,2.0,5.0\n1.0,2.0,5.0'

result = predictor.predict(input)

# result contains:
{
  'predictions': [
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5],
    [3.5, 4.0, 5.5]
  ]
}

You can also use python arrays or numpy arrays as input and let the CSVSerializer object convert them to CSV, but the client-size CSV conversion is more sophisticated than the CSV parsing on the Endpoint, so if you encounter conversion problems, try using one of the JSON options instead.

Specifying the output of a prediction request

The structure of the prediction result is determined at the end of the training process before SavedModel is created. For example, if you are using TensorFlow’s Estimator API for training, you control inference outputs using the export_outputs parameter of the tf.estimator.EstimatorSpec that you return from your model_fn.

More information on how to create export_outputs can be found in specifying the outputs of a custom model. You can also refer to TensorFlow’s Save and Restore documentation for other ways to control the inference-time behavior of your SavedModels.

Providing Python scripts for pre/post-processing

You can add your customized Python code to process your input and output data. This customized Python code must be named inference.py and specified through the entry_point parameter:

from sagemaker.tensorflow import TensorFlowModel

model = Model(entry_point='inference.py',
              model_data='s3://mybucket/model.tar.gz',
              role='MySageMakerRole')

How to implement the pre- and/or post-processing handler(s)

Your entry point file must be named inference.py and should implement

either a pair of input_handler and output_handler functions or a single handler function. Note that if handler function is implemented, input_handler and output_handler are ignored.

To implement pre- and/or post-processing handler(s), use the Context object that the Python service creates. The Context object is a namedtuple with the following attributes:

  • model_name (string): the name of the model to use for inference. For example, ‘half-plus-three’

  • model_version (string): version of the model. For example, ‘5’

  • method (string): inference method. For example, ‘predict’, ‘classify’ or ‘regress’, for more information on methods, please see Classify and Regress API and Predict API

  • rest_uri (string): the TFS REST uri generated by the Python service. For example, ‘http://localhost:8501/v1/models/half_plus_three:predict

  • grpc_uri (string): the GRPC port number generated by the Python service. For example, ‘9000’

  • custom_attributes (string): content of ‘X-Amzn-SageMaker-Custom-Attributes’ header from the original request. For example, ‘tfs-model-name=half*plus*three,tfs-method=predict’

  • request_content_type (string): the original request content type, defaulted to ‘application/json’ if not provided

  • accept_header (string): the original request accept type, defaulted to ‘application/json’ if not provided

  • content_length (int): content length of the original request

The following code example implements input_handler and output_handler. By providing these, the Python service posts the request to the TFS REST URI with the data pre-processed by input_handler and passes the response to output_handler for post-processing.

import json

def input_handler(data, context):
    """ Pre-process request input before it is sent to TensorFlow Serving REST API
    Args:
        data (obj): the request data, in format of dict or string
        context (Context): an object containing request and configuration details
    Returns:
        (dict): a JSON-serializable dict that contains request body and headers
    """
    if context.request_content_type == 'application/json':
        # pass through json (assumes it's correctly formed)
        d = data.read().decode('utf-8')
        return d if len(d) else ''

    if context.request_content_type == 'text/csv':
        # very simple csv handler
        return json.dumps({
            'instances': [float(x) for x in data.read().decode('utf-8').split(',')]
        })

    raise ValueError('{{"error": "unsupported content type {}"}}'.format(
        context.request_content_type or "unknown"))


def output_handler(data, context):
    """Post-process TensorFlow Serving output before it is returned to the client.
    Args:
        data (obj): the TensorFlow serving response
        context (Context): an object containing request and configuration details
    Returns:
        (bytes, string): data to return to client, response content type
    """
    if data.status_code != 200:
        raise ValueError(data.content.decode('utf-8'))

    response_content_type = context.accept_header
    prediction = data.content
    return prediction, response_content_type

You might want to have complete control over the request. For example, you might want to make a TFS request (REST or GRPC) to the first model, inspect the results, and then make a request to a second model. In this case, implement the handler method instead of the input_handler and output_handler methods, as demonstrated in the following code:

import json
import requests


def handler(data, context):
    """Handle request.
    Args:
        data (obj): the request data
        context (Context): an object containing request and configuration details
    Returns:
        (bytes, string): data to return to client, (optional) response content type
    """
    processed_input = _process_input(data, context)
    response = requests.post(context.rest_uri, data=processed_input)
    return _process_output(response, context)


def _process_input(data, context):
    if context.request_content_type == 'application/json':
        # pass through json (assumes it's correctly formed)
        d = data.read().decode('utf-8')
        return d if len(d) else ''

    if context.request_content_type == 'text/csv':
        # very simple csv handler
        return json.dumps({
            'instances': [float(x) for x in data.read().decode('utf-8').split(',')]
        })

    raise ValueError('{{"error": "unsupported content type {}"}}'.format(
        context.request_content_type or "unknown"))


def _process_output(data, context):
    if data.status_code != 200:
        raise ValueError(data.content.decode('utf-8'))

    response_content_type = context.accept_header
    prediction = data.content
    return prediction, response_content_type

You can also bring in external dependencies to help with your data processing. There are 2 ways to do this:

  1. If you included requirements.txt in your source_dir, the container installs the Python dependencies at runtime using pip install -r:

from sagemaker.tensorflow import TensorFlowModel

model = Model(entry_point='inference.py',
              source_dir='source/directory',
              model_data='s3://mybucket/model.tar.gz',
              role='MySageMakerRole')
  1. If you are working in a network-isolation situation or if you don’t want to install dependencies at runtime every time your endpoint starts or a batch transform job runs, you might want to put pre-downloaded dependencies under a lib directory and this directory as dependency. The container adds the modules to the Python path. Note that if both lib and requirements.txt are present in the model archive, the requirements.txt is ignored:

from sagemaker.tensorflow import TensorFlowModel

model = Model(entry_point='inference.py',
              dependencies=['/path/to/folder/named/lib'],
              model_data='s3://mybucket/model.tar.gz',
              role='MySageMakerRole')

For more information, see: https://github.com/aws/sagemaker-tensorflow-serving-container#prepost-processing

Deploying more than one model to your Endpoint

TensorFlow Serving Endpoints allow you to deploy multiple models to the same Endpoint when you create the endpoint.

To use this feature, you will need to:

  1. create a multi-model archive file

  2. create a SageMaker Model and deploy it to an Endpoint

  3. create Predictor instances that direct requests to a specific model

Creating a multi-model archive file

Creating an archive file that contains multiple SavedModels is simple, but involves a few steps:

  • obtaining some models

  • repackaging the models into a new archive file

  • uploading the new archive to S3

Obtaining model files

Let’s imagine you have already run two Tensorflow training jobs in SageMaker, and they exported SavedModels to s3://mybucket/models/model1.tar.gz and s3://mybucket/models/model2.tar.gz.

First, download the models and extract them:

aws s3 cp s3://mybucket/models/model1/model.tar.gz model1.tar.gz
aws s3 cp s3://mybucket/models/model2/model.tar.gz model2.tar.gz
mkdir -p multi/model1
mkdir -p multi/model2

tar xvf model1.tar.gz -C ./multi/model1
tar xvf model2.tar.gz -C ./multi/model2

Repackaging the models

Next, examine the directories in multi. If you trained the models using SageMaker’s TensorFlow containers, you are likely to have ./multi/model1/export/Servo/... and ./multi/model2/export/Servo/.... In both cases, “Servo” is the base name for the SaveModel files. When serving multiple models, each model needs a unique basename, so one or both of these will need to be changed. The /export/ part of the path isn’t needed either, so you can simplify the layout at the same time:

mv multi/model1/export/Servo/* multi/model1/
mv multi/model2/export/Servo/* multi/model2/
rm -fr multi/model1/export
rm -fr multi/model2/export

You should now have a directory structure like this:

└── multi
  ├── model1
  │   └── <version number>
  │       ├── saved_model.pb
  │       └── variables
  │           └── ...
  └── model2
      └── <version number>
          ├── saved_model.pb
          └── variables
              └── ...

To repackage the files into a new archive, use tar again:

tar -C "$PWD/multi/" -czvf multi.tar.gz multi/

The multi.tar.gz file is now ready to use.

Uploading the new archive to S3

aws s3 cp multi.tar.gz s3://mybucket/models/multi.tar.gz

Creating and Deploying a SageMaker Model

For the remaining steps, let’s return to python code using the SageMaker Python SDK.

from sagemaker.tensorflow import TensorFlowModel, TensorFlowPredictor

# change this to the name or ARN of your SageMaker execution role
role = 'SageMakerRole'

model_data = 's3://mybucket/models/multi.tar.gz'

# For multi-model endpoints, you should set the default model name in
# an environment variable. If it isn't set, the endpoint will work,
# but the model it will select as default is unpredictable.
env = {
  'SAGEMAKER_TFS_DEFAULT_MODEL_NAME': 'model1'
}

model = Model(model_data=model_data, role=role, framework_version='1.11', env=env)
predictor = model.deploy(initial_instance_count=1, instance_type='ml.c5.xlarge')

The predictor object returned by the deploy function is ready to use to make predictions using the default model (model1 in this example).

Creating Predictor instances for different models

The predictor returned by the model.deploy(...) function can only send requests to the default model. To use other models deployed to the same Endpoint, you need to create additional Predictor instances. Here’s how:

# ... continuing from the previous example

# get the endpoint name from the default predictor
endpoint = predictor.endpoint_name

# get a predictor for 'model2'
model2_predictor = Predictor(endpoint, model_name='model2')

# note: that will for actual SageMaker endpoints, but if you are using
# local mode you need to create the new Predictor this way:
#
# model2_predictor = Predictor(endpoint, model_name='model2'
#                              sagemaker_session=predictor.sagemaker_session)


# result is prediction from 'model2'
result = model2_predictor.predict(...)

Making predictions with the AWS CLI

The SageMaker Python SDK is not the only way to access your Endpoint. The AWS CLI is simple to use and a convenient way to test your endpoint. Here are a few examples that show how to use different features of SageMaker TensorFlow Serving Endpoints using the CLI.

Note: The invoke-endpoint command usually writes prediction results to a file. In the examples below, the >(cat) 1>/dev/null part is a shell trick to redirect the result to stdout so it can be seen.

# TensorFlow Serving REST API - predict request
aws sagemaker-runtime invoke-endpoint \
    --endpoint-name my-endpoint \
    --content-type 'application/json' \
    --body '{"instances": [1.0, 2.0, 5.0]}' \
    >(cat) 1>/dev/null

# Predict request for specific model name
aws sagemaker-runtime invoke-endpoint \
    --endpoint-name my-endpoint \
    --content-type 'application/json' \
    --body '{"instances": [1.0, 2.0, 5.0]}' \
    --custom-attributes 'tfs-model-name=other_model' \
    >(cat) 1>/dev/null

# TensorFlow Serving REST API - regress request
aws sagemaker-runtime invoke-endpoint \
    --endpoint-name my-endpoint \
    --content-type 'application/json' \
    --body '{"signature_name": "tensorflow/serving/regress","examples": [{"x": 1.0}]}' \
    --custom-attributes 'tfs-method=regress' \
    >(cat) 1>/dev/null

# Simple json request (2 instances)
aws sagemaker-runtime invoke-endpoint \
    --endpoint-name my-endpoint \
    --content-type 'application/json' \
    --body '[[1.0, 2.0, 5.0],[2.0, 3.0, 4.0]]' \
    >(cat) 1>/dev/null

# CSV request (2 rows)
aws sagemaker-runtime invoke-endpoint \
    --endpoint-name my-endpoint \
    --content-type 'text/csv' \
    --body "1.0,2.0,5.0"$'\n'"2.0,3.0,4.0" \
    >(cat) 1>/dev/null

# Line delimited JSON from an input file
aws sagemaker-runtime invoke-endpoint \
    --endpoint-name my-endpoint \
    --content-type 'application/jsons' \
    --body "$(cat input.jsons)" \
    results.json