BentoML

Steps:

Build & Train Model

• downloaded small sample dataset
• train_model.py > generate model&scaler pkl files

Bentoml service

• sample Wine service with one endpoint > service.py
• try and serve locally to check everything is fine

bentoml serve service:Wine

• bentofile.yaml with required pip installs
• build bentoml

bentoml build

• containerize (on mac so i need to specify platform)

bentoml containerize --platform=linux/amd64 wine:latest

• image will be saved to the local registery
• run docker image locally with port 3000 exposed

docker run --rm -p 3000:3000 wine:latest