When PR containing metrics of automated training is completed, it should trigger deployment of containers with the new trained model. about aifororcas-livesystem HOT 2 OPEN

@Herman-Wu see https://github.com/orcasound/aifororcas-livesystem/tree/main/InferenceSystem#deploying-your-updated-container-to-azure-container-instances for how we deploy to ACI usually. The README.md is outdated the container-group is now called live-inference-system-aci-3gb-new I believe. Please cross-check with Azure portal to situate yourself.

from aifororcas-livesystem.

Thanks, @prakruti . Added new GitHub action workflow [GitHub Actions Deploy AI4Orca ML ACI] and submitted PR.
Since there is no event for model training completed yet, the action will be triggered by updating files in "config/Production/" folder.
To going forward, we will need the following additional tasks to complete the [automatic training-finish -> deploy inferencing docker to ACI] scenario.

1. Need another workflow to create and register new Docker images automatically. ( Or re-write who docker images get a new model and preprocess data)
2. Event or Manual trigger that kick-off the process
3. Smoke testing on the new images in ACI.

from aifororcas-livesystem.