Tensorflow Model Server Docker Image

You can make your Tensorflow model a Docker image by making changes to Dockerfile and deploy to Kubernetes.

Details at the Wiki.

Requirement

Install Docker

Install Kubernetes

Creation of Docker Image

1. Step

• Copy the model file into the 'models' folder.

2. Step

• Replace MODEL_NAME in Dockerfile with your own model name.

3. Step (This step may take some time!)

  $ docker build -t IMAGE_NAME .

4. Step

  $ docker run -p 8501:8501 IMAGE_NAME  

5. Step

• You can use the url.

  localhost:8501

Deploy Kubernetes

1. Step

  $ cd kubernetesFiles

2. Step

  $ kubectl apply –f deployment.yaml
  $ kubectl apply –f service.yaml