sagemaker-notebook.ipynb

Kernel: conda_pytorch_p39

Going Production: Auto-scale Hugging Face Transformer Endpoints with Amazon SageMaker

Welcome to this getting started guide, we will use the new Hugging Face Inference DLCs and Amazon SageMaker Python SDK to deploy a transformer model for real-time inference. In this example we are going to deploy a trained Hugging Face Transformer model on to SageMaker for inference.

!pip install "sagemaker>=2.66.2" --upgrade

Deploy one of the 15 000+ Hugging Face Transformers to Amazon SageMaker for Inference

To deploy a model directly from the Hub to SageMaker we need to define 2 environment variables when creating the HuggingFaceModel . We need to define:

• HF_MODEL_ID: defines the model id, which will be automatically loaded from huggingface.co/models when creating or SageMaker Endpoint. The 🤗 Hub provides +10 000 models all available through this environment variable.

• HF_TASK: defines the task for the used 🤗 Transformers pipeline. A full list of tasks can be find here.

import sagemaker
import boto3

try:
    role = sagemaker.get_execution_role()
except ValueError:
    iam = boto3.client('iam')
    role = iam.get_role(RoleName='sagemaker_execution_role')['Role']['Arn']

print(f"sagemaker role arn: {role}")

from sagemaker.huggingface import HuggingFaceModel
from uuid import uuid4
import sagemaker 

# Hub Model configuration. https://huggingface.co/models
hub = {
  'HF_MODEL_ID':'yiyanghkust/finbert-tone', # model_id from hf.co/models
  'HF_TASK':'text-classification' # NLP task you want to use for predictions
}

# endpoint name
endpoint_name=f'{hub["HF_MODEL_ID"].split("/")[1]}-{str(uuid4())}' # model and endpoint name

# create Hugging Face Model Class
huggingface_model = HuggingFaceModel(
   env=hub,
   role=role, # iam role with permissions to create an Endpoint
   name=endpoint_name, # model and endpoint name
   transformers_version="4.26", # transformers version used
   pytorch_version="1.13", # pytorch version used
   py_version="py39", # python version of the DLC
)

# deploy model to SageMaker Inference
predictor = huggingface_model.deploy(
   initial_instance_count=1,
   instance_type="ml.c5.large"
)
# get aws region for dashboards
aws_region = predictor.sagemaker_session.boto_region_name

-----!

Architecture

The Hugging Face Inference Toolkit for SageMaker is an open-source library for serving Hugging Face transformer models on SageMaker. It utilizes the SageMaker Inference Toolkit for starting up the model server, which is responsible for handling inference requests. The SageMaker Inference Toolkit uses Multi Model Server (MMS) for serving ML models. It bootstraps MMS with a configuration and settings that make it compatible with SageMaker and allow you to adjust important performance parameters, such as the number of workers per model, depending on the needs of your scenario.

Deploying a model using SageMaker hosting services is a three-step process:

1. Create a model in SageMaker —By creating a model, you tell SageMaker where it can find the model components.

2. Create an endpoint configuration for an HTTPS endpoint —You specify the name of one or more models in production variants and the ML compute instances that you want SageMaker to launch to host each production variant.

3. Create an HTTPS endpoint —Provide the endpoint configuration to SageMaker. The service launches the ML compute instances and deploys the model or models as specified in the configuration

# example request, you always need to define "inputs"
data = {
   "inputs": "There is a shortage of capital for project SageMaker. We need extra financing"
}

# request
predictor.predict(data)

[{'label': 'negative', 'score': 0.9870443940162659}]

for i in range(500):
    predictor.predict(data)

Model Monitoring

print(f"https://console.aws.amazon.com/cloudwatch/home?region={aws_region}#metricsV2:graph=~(metrics~(~(~'AWS*2fSageMaker~'ModelLatency~'EndpointName~'finbert-tone-73d26f97-9376-4b3f-9334-a2-2021-10-29-12-18-52-365~'VariantName~'AllTraffic))~view~'timeSeries~stacked~false~start~'-PT15M~end~'P0D~region~'{aws_region}~stat~'SampleCount~period~30);query=~'*7bAWS*2fSageMaker*2cEndpointName*2cVariantName*7d*20{predictor.endpoint_name}")

Auto Scaling your Model

Amazon SageMaker is a fully managed service that provides every developer and data scientist with the ability to quickly build, train, and deploy machine learning (ML) models at scale.

Autoscaling is an out-of-the-box feature that monitors your workloads and dynamically adjusts the capacity to maintain steady and predictable performance at the possible lowest cost.

The following diagram is a sample architecture that showcases how a model is served as a endpoint with autoscaling enabled.

Reference Blog post Configuring autoscaling inference endpoints in Amazon SageMaker

Configure Autoscaling for our Endpoint

You can define minimum, desired, and maximum number of instances per endpoint and, based on the autoscaling configurations, instances are managed dynamically. The following diagram illustrates this architecture.

AWS offers many different ways to auto-scale your endpoints. One of them Simple-Scaling, where you scale the instance capacity based on CPUUtilization of the instances or SageMakerVariantInvocationsPerInstance.

In this example we are going to use SageMakerVariantInvocationsPerInstance to auto-scale our Endpoint

import boto3

# Let us define a client to play with autoscaling options
asg_client = boto3.client('application-autoscaling') # Common class representing Application Auto Scaling for SageMaker amongst other services

# he resource type is variant and the unique identifier is the resource ID.
# Example: endpoint/my-bert-fine-tuned/variant/AllTraffic .
resource_id=f"endpoint/{predictor.endpoint_name}/variant/AllTraffic"

# scaling configuration
response = asg_client.register_scalable_target(
    ServiceNamespace='sagemaker', #
    ResourceId=resource_id,
    ScalableDimension='sagemaker:variant:DesiredInstanceCount', 
    MinCapacity=1,
    MaxCapacity=4
)

Create Scaling Policy with configuration details, e.g. TargetValue when the instance should be scaled.

response = asg_client.put_scaling_policy(
    PolicyName=f'Request-ScalingPolicy-{predictor.endpoint_name}',
    ServiceNamespace='sagemaker',
    ResourceId=resource_id,
    ScalableDimension='sagemaker:variant:DesiredInstanceCount',
    PolicyType='TargetTrackingScaling',
    TargetTrackingScalingPolicyConfiguration={
        'TargetValue': 10.0, # Threshold
        'PredefinedMetricSpecification': {
            'PredefinedMetricType': 'SageMakerVariantInvocationsPerInstance',
        },
        'ScaleInCooldown': 300, # duration until scale in
        'ScaleOutCooldown': 60 # duration between scale out
    }
)

stress test the endpoint with threaded requests

import time

request_duration_in_seconds = 4*65
end_time = time.time() + request_duration_in_seconds

print(f"test will run {request_duration_in_seconds} seconds")

while time.time() < end_time:
    predictor.predict(data)

Monitor the InvocationsPerInstance in cloudwatch

print(f"https://console.aws.amazon.com/cloudwatch/home?region={aws_region}#metricsV2:graph=~(metrics~(~(~'AWS*2fSageMaker~'InvocationsPerInstance~'EndpointName~'{predictor.endpoint_name}~'VariantName~'AllTraffic))~view~'timeSeries~stacked~false~region~'{aws_region}~start~'-PT15M~end~'P0D~stat~'SampleCount~period~60);query=~'*7bAWS*2fSageMaker*2cEndpointName*2cVariantName*7d*20{predictor.endpoint_name}")

check the endpoint instance_count number

bt_sm = boto3.client('sagemaker')
response = bt_sm.describe_endpoint(EndpointName=predictor.endpoint_name)
print(f"Endpoint {response['EndpointName']} has \nCurrent Instance Count: {response['ProductionVariants'][0]['CurrentInstanceCount']}\nWith a desired instance count of {response['ProductionVariants'][0]['DesiredInstanceCount']}")

Endpoint finbert-tone-73d26f97-9376-4b3f-9334-a2-2021-10-29-12-18-52-365 has 
Current Instance Count: 4
With a desired instance count of 4

Clean up

# delete endpoint
predictor.delete_model()
predictor.delete_endpoint()