MWE: ```python import openml from sklearn.tree import DecisionTreeClassifier run = openml.runs.run_model_on_task(DecisionTreeClassifier(), 59) print('Run is not yet uploaded:', run.id == None) run.setup_string = "test.py -o --setup-string" run.publish() print('local run:', run.id, run.setup_string) server_run = openml.runs.get_run(run.id) print('server run:', server_run.id, server_run.setup_string) ``` current output: ``` Run is not yet uploaded: True local run: 10559751 test.py -o --setup-string server run: 10559751 None ``` expected output: ``` Run is not yet uploaded: True local run: 10559751 test.py -o --setup-string server run: 10559751 test.py -o --setup-string ``` I imagine adding the field to `OpenMLRun._to_dict` is enough. @mfeurer this is just an oversight, right? I couldn't find/think of a reason as to why this would not be supported.
Activity
mfeurer commentedon Dec 21, 2020
I'd think so