diff --git a/python/trainer_config.yaml b/python/trainer_config.yaml
index d2ac3bca79..dd76dbf9a4 100644
--- a/python/trainer_config.yaml
+++ b/python/trainer_config.yaml
@@ -130,12 +130,6 @@ Ball3DHardBrain:
     gamma: 0.995
     beta: 0.001
 
-BouncerBrain:
-    normalize: true
-    max_steps: 5.0e5
-    num_layers: 2
-    hidden_units: 56
-
 TennisBrain:
     normalize: true
 
diff --git a/python/unitytrainers/ppo/trainer.py b/python/unitytrainers/ppo/trainer.py
index 549c5e9d63..482d40180a 100755
--- a/python/unitytrainers/ppo/trainer.py
+++ b/python/unitytrainers/ppo/trainer.py
@@ -204,6 +204,9 @@ def generate_intrinsic_rewards(self, curr_info, next_info):
         :return: Intrinsic rewards for all agents.
         """
         if self.use_curiosity:
+            if curr_info.agents != next_info.agents:
+                raise UnityTrainerException("Training with Curiosity-driven exploration"
+                                            " and On-Demand Decision making is currently not supported.")
             feed_dict = {self.model.batch_size: len(curr_info.vector_observations), self.model.sequence_length: 1}
             if self.is_continuous_action:
                 feed_dict[self.model.output] = next_info.previous_vector_actions