diff --git a/python/ray/rllib/agents/ddpg/ddpg_policy.py b/python/ray/rllib/agents/ddpg/ddpg_policy.py
index 5aa1aa710..95e4bd121 100644
--- a/python/ray/rllib/agents/ddpg/ddpg_policy.py
+++ b/python/ray/rllib/agents/ddpg/ddpg_policy.py
@@ -231,17 +231,15 @@ class DDPGTFPolicy(DDPGPostprocessing, TFPolicy):
         if config["l2_reg"] is not None:
             for var in self.policy_vars:
                 if "bias" not in var.name:
-                    self.actor_loss += (
-                        config["l2_reg"] * 0.5 * tf.nn.l2_loss(var))
+                    self.actor_loss += (config["l2_reg"] * tf.nn.l2_loss(var))
             for var in self.q_func_vars:
                 if "bias" not in var.name:
-                    self.critic_loss += (
-                        config["l2_reg"] * 0.5 * tf.nn.l2_loss(var))
+                    self.critic_loss += (config["l2_reg"] * tf.nn.l2_loss(var))
             if self.config["twin_q"]:
                 for var in self.twin_q_func_vars:
                     if "bias" not in var.name:
                         self.critic_loss += (
-                            config["l2_reg"] * 0.5 * tf.nn.l2_loss(var))
+                            config["l2_reg"] * tf.nn.l2_loss(var))
 
         # update_target_fn will be called periodically to copy Q network to
         # target Q network