Option to specify a csv file where all HPO/repeated runs with all metrics and params are saved.
This might be an alternative to the already implemented saving of min/max metrics at the of each run and save it to the csv. However, I would like both options.
This would make it easier to analyse the data in python instead of the MLflow UI.
Optionally, we could even think of some plotting utils to visualze the runs (however, it don't know if these utils would generalize or if they are specific to users).
Option to specify a csv file where all HPO/repeated runs with all metrics and params are saved.
This might be an alternative to the already implemented saving of min/max metrics at the of each run and save it to the csv. However, I would like both options.
This would make it easier to analyse the data in python instead of the MLflow UI.
Optionally, we could even think of some plotting utils to visualze the runs (however, it don't know if these utils would generalize or if they are specific to users).