diff --git a/python/ray/rllib/ddpg/apex.py b/python/ray/rllib/ddpg/apex.py
index 1799e71a8..7a3b5c598 100644
--- a/python/ray/rllib/ddpg/apex.py
+++ b/python/ray/rllib/ddpg/apex.py
@@ -45,6 +45,6 @@ class ApexDDPGAgent(DDPGAgent):
         # Ape-X updates based on num steps trained, not sampled
         if self.optimizer.num_steps_trained - self.last_target_update_ts > \
                 self.config["target_network_update_freq"]:
-            self.local_evaluator.update_target()
+            self.local_evaluator.for_policy(lambda p: p.update_target())
             self.last_target_update_ts = self.optimizer.num_steps_trained
             self.num_target_updates += 1
diff --git a/python/ray/rllib/dqn/apex.py b/python/ray/rllib/dqn/apex.py
index 34e6ecd91..bfb58d02a 100644
--- a/python/ray/rllib/dqn/apex.py
+++ b/python/ray/rllib/dqn/apex.py
@@ -56,6 +56,6 @@ class ApexAgent(DQNAgent):
         # Ape-X updates based on num steps trained, not sampled
         if self.optimizer.num_steps_trained - self.last_target_update_ts > \
                 self.config["target_network_update_freq"]:
-            self.local_evaluator.update_target()
+            self.local_evaluator.for_policy(lambda p: p.update_target())
             self.last_target_update_ts = self.optimizer.num_steps_trained
             self.num_target_updates += 1