How to capture MLflow Experiment description programmatically?

Question

I am trying to capture the MLflow experiment descriptions programmatically. I don't find any option as such in create or set experiments parameters. Is there a way to address this?

I tried to use the create/set_experiment when creating experiment in MLflow, but they don't have option to capture the experiment descriptions programmatically. Is there any different option available apart from having the description via UI?

Answer 1

Tag your experiment appropriately, in this case use mlflow.note.content tag.

experiment_id = mlflow.create_experiment(
   'my-experiment',
   tags={'mlflow.note.content':'this experiment is doing nothing'}
)

PS: How to figure it out, as this seems not well documented? Assign a description with GUI, and then inspect it in the backend :-)