The Loss Tracker helps track model training and experiments. The project consists of a Python interface which can send data to a backend. The backend streams this data to the frontend for visualization:
Experiments can be stored and later retrieved via the specified name in the Python logging request.
cd client && npm run dev
cd server && npx nodemon
Documentation about the Python interface can be found in tracker/.
In the Github Actions workflow the project gets build for arm64. These images can be used to deploy on a configured Kubernetes cluster. The configuration files for the Kubernetes Ingress and Deployment can be found in server/production and client/production.