Mikhail Knyazev
04/15/2022, 1:23 PMextract_data -> prepare_data -> train_model
. The job in typically launched as a whole with a pre-defined config via the "Launchpad" or a sensor. But in some cases I want to tweak config of the train_model
op specifically while reusing the output of prepare_data
from the previous run.
Of course I can split this job in two, explicitly passing prepared data between jobs, but it feels like repeating LocalArtifactStorage
logic...sandy
04/15/2022, 7:58 PM