Start of run in MLflow should not depends on the value of '--print_report' parameter.
As for now in case '--print_report'=False, the run for inference process is not created in MLflow.
And vice-versa, the run in created as far as '--print_report'=True
Regarding the training process started with --print_report=True, the only one run (TRAIN) should be created. The run should contain the reports and metrics as well.
Start of run in MLflow should not depends on the value of '--print_report' parameter. As for now in case '--print_report'=False, the run for inference process is not created in MLflow. And vice-versa, the run in created as far as '--print_report'=True
Regarding the training process started with --print_report=True, the only one run (TRAIN) should be created. The run should contain the reports and metrics as well.