Deploy machine learning models in production

Cortex is an open source platform for deploying machine learning models as production web services.

install • tutorial • docs • examples • we're hiring • email us • chat with us

Key features

Multi framework: Cortex supports TensorFlow, PyTorch, scikit-learn, XGBoost, and more.
Autoscaling: Cortex automatically scales APIs to handle production workloads.
CPU / GPU support: Cortex can run inference on CPU or GPU infrastructure.
Spot instances: Cortex supports EC2 spot instances.
Rolling updates: Cortex updates deployed APIs without any downtime.
Log streaming: Cortex streams logs from deployed models to your CLI.
Prediction monitoring: Cortex monitors network metrics and tracks predictions.
Minimal configuration: Cortex deployments are defined in a single cortex.yaml file.

Spinning up a cluster

Cortex is designed to be self-hosted on any AWS account. You can spin up a cluster with a single command:

# install the CLI on your machine
$ bash -c "$(curl -sS https://raw.githubusercontent.com/cortexlabs/cortex/0.13/get-cli.sh)"

# provision infrastructure on AWS and spin up a cluster
$ cortex cluster up

aws region: us-west-2
aws instance type: p2.xlarge
spot instances: yes
min instances: 0
max instances: 10

￮ spinning up your cluster ...
your cluster is ready!

Deploying a model

Implement your predictor

# predictor.py

class PythonPredictor:
    def __init__(self, config):
        self.model = download_model()

    def predict(self, payload):
        return model.predict(payload["text"])

Configure your deployment

# cortex.yaml

- name: sentiment-classifier
  predictor:
    type: python
    path: predictor.py
  tracker:
    model_type: classification
  compute:
    gpu: 1
    mem: 4G

Deploy to AWS

$ cortex deploy

creating sentiment-classifier

Serve real-time predictions

$ curl http://***.amazonaws.com/sentiment-classifier \
    -X POST -H "Content-Type: application/json" \
    -d '{"text": "the movie was amazing!"}'

positive

Monitor your deployment

$ cortex get sentiment-classifier --watch

status   up-to-date   requested   last update   avg inference   2XX
live     1            1           8s            24ms            12

class     count
positive  8
negative  4

What is Cortex similar to?

Cortex is an open source alternative to serving models with SageMaker or building your own model deployment platform on top of AWS services like Elastic Kubernetes Service (EKS), Elastic Container Service (ECS), Lambda, Fargate, and Elastic Compute Cloud (EC2) and open source projects like Docker, Kubernetes, and TensorFlow Serving.

How does Cortex work?

The CLI sends configuration and code to the cluster every time you run cortex deploy. Each model is loaded into a Docker container, along with any Python packages and request handling code. The model is exposed as a web service using Elastic Load Balancing (ELB), TensorFlow Serving, and ONNX Runtime. The containers are orchestrated on Elastic Kubernetes Service (EKS) while logs and metrics are streamed to CloudWatch.

Examples of Cortex deployments

Sentiment analysis: deploy a BERT model for sentiment analysis.
Image classification: deploy an Inception model to classify images.
Search completion: deploy Facebook's RoBERTa model to complete search terms.
Text generation: deploy Hugging Face's DistilGPT2 model to generate text.
Iris classification: deploy a scikit-learn model to classify iris flowers.

juangon/cortex