Missing ML model informations

Hi Community!
I trained a machine learning model in a python processor and saved it to ONE DATA. Now when I navigate to the Models tab and open one saved model I only see informations about the model name and creation/modification dates.
However, in this thread under point 4 (Serve Models to MLeap / Docker Containers : ONE LOGIC Service Center) there is a screenshot of exactly this model view, where one can select and look at different model versions with parameter informations.

Is this only available for Spark/MLeap machine learning models trained using the custom ONE DATA “Train model” processor and not for models originating in python?
Thanks in advance!

Short answer: Yes.

Longer answer: This was a deliberate decision during the implementation of the Python model feature to keep it as generic as possible. Especially to avoid limiting the functionality to work with just a single version of Python ML libraries for which we parsed the resulting pickles.

There was the idea for “the future” to allow the user to add arbitrary meta-information in the “add_model”-call to od_output, but this was not followed due to time/budget constraints.

1 Like