Aiden-Jeon/model-registry-tutorial

tutorial code for model reigstry

Model Registry Tutorial

tutorial code for model registry

Model registry

1. Run model registry with docker compose command

   docker compose -f docker/model-registry/docker-compose.yaml up -d

2. Check whether you can access your model registry: http://localhost:5001

Train model

1. Install packages:

   pip install -r src/requirements.txt

2. Train model through script you can choose scikit-learn native model or pyfunc custom model

   • scikit-learn model

     python src/train.py

   • pyfunc custom model

     python src/train_custom_model.py

API serving

1. Check your run id from model registry and download model.

   python docker/api-serving/download_model.py --run-id <YOUR_RUN_ID_HERE>

2. Build docker image.

   docker build ./docker/api-serving -f ./docker/api-serving/Dockerfile -t api-serving:latest

3. Run your image.

   docker run -p 8000:8000 api-serving:latest

4. Check your swagger: http://localhost:8000