6

I am trying to move a run in MLflow from one experiment to another. Does anybody know if its possible? If yes, how? (I use Python API)

Kıvanç Yüksel
  • 701
  • 7
  • 17

3 Answers3

3

https://github.com/mlflow/mlflow-export-import

You can copy a run from one experiment to another - either in the same tracking server or between two tracking servers. Caveats apply if they are Databricks MLflow tracking servers.

mirekphd
  • 4,799
  • 3
  • 38
  • 59
Andre
  • 304
  • 1
  • 2
  • The package supports a 3-month old MLflow Server (or newer)... to new for many prod environments... But fixing the link (it was moved to "mlflow" namespace). – mirekphd Jul 20 '23 at 14:57
1

For local users, I directly move run folders and modify the experiment id in meta.yaml.

Rémy Hosseinkhan Boucher
  • 145
  • 9
  • Your answer could be improved with additional supporting information. Please [edit] to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers [in the help center](/help/how-to-answer). – Community Oct 23 '22 at 08:12
0

For local users, I directly move run folders and modify the experiment id in meta.yaml.

To help elaborate, every run has its own folder, which basically contains more folders like "artifacts", "metrics", "params", and "tags", you could find a "meta.yaml" file in that directory.

folder example

Open the meta.yaml file and you should find the experiment_id, simply change the number behind it should do the trick.

meta.yaml example

Sunny Lee
  • 31
  • 5