Template Repository to easily get started with CD pipeline for model serving with tensorflow serving and Heroku
-
Create an app on your Heroku account (Create Heroku account if you don't have one)
-
in your account, go to settings and copy your heroku api key
-
on your github repo, go to settings > secrets and add the HEROKU API KEY:
HEROKU_API_KEY
as the key and the api ke you copied from heroku as value -
in the repo, go the
.github/workflows/main.yaml
and uncomment the code -
add your email address to the yaml file and also your app name for example:
name: Deploy on: push: branches: - main jobs: build: runs-on: ubuntu-latest steps: - uses: actions/checkout@v2 - uses: akhileshns/[email protected] with: heroku_api_key: ${{secrets.HEROKU_API_KEY}} heroku_app_name: "scalable-example" heroku_email: "[email protected]" usedocker: true
-
in the models folder, you can add in your saved tensorflow model (notice the strcture of the default model)
-
for every model added in, you must update the
Dockerfile
andmodels.conf
file