Mlflow in practice

You must determine 3 aspects depending on the usecase:

1. Backend store If you didn’t specify - - backend-store-uri when starting mlflow, it defaults to the local file system.
2. Artifact store defaults to the local file system if - - default-artifact-root flag
3. Tracking server

mlflow ui

It just a UI presentation of the backend artifacts and backend store but not a server

I noticed that in case of local backend store and local artifact store the experiments ids become much longer because it’s expected to have much more descriptive data about the experiment.