In this PR, I'm fixing a design issue inherited from the original mlflow platform. In this case, when someone instantiates an experiment with default values, instead of letting the server set the artifact_location, we are setting it as ./mlruns to avoid non-absolute directories (breaking logging artifacts).
48
In this PR, I'm fixing a design issue inherited from the original
mlflow
platform. In this case, when someone instantiates an experiment with default values, instead of letting the server set theartifact_location
, we are setting it as./mlruns
to avoid non-absolute directories (breaking logging artifacts).