DBAAS-3012: clean up tags in start_run (#31)

Author: Epstein

splicemachine/ml/management.py

@@ -284,7 +284,7 @@ def set_active_run(self, run_id):
         """
         self.active_run = self.get_run(run_id)
 
-    def start_run(self, tags=None, run_name=None, experiment_id=None, nested=False):
+    def start_run(self, tags={}, run_name=None, experiment_id=None, nested=False):
         """
         Create a new run in the active experiment and set it to be active
         :param tags: a dictionary containing metadata about the current run.
@@ -311,9 +311,6 @@ def start_run(self, tags=None, run_name=None, experiment_id=None, nested=False):
                 raise MlflowException(
                     "There are no experiements available yet. Please create an experiment before starting a run")
 
-        if not tags:
-            tags = {}
-
         tags['mlflow.user'] = _get_user()
 
         self.active_run = super(MLManager, self).create_run(new_run_exp_id, tags=tags)