From da7fb4b2272c2551bcdfe1e32f5bba2ccc556375 Mon Sep 17 00:00:00 2001
From: jspieler <75164246+jspieler@users.noreply.github.com>
Date: Thu, 9 Mar 2023 20:49:52 +0100
Subject: [PATCH] Fixed DDPG example (#667)

---
 examples/deep_deterministic_policy_gradient.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/examples/deep_deterministic_policy_gradient.py b/examples/deep_deterministic_policy_gradient.py
index 6305cac9cb..30be069abd 100644
--- a/examples/deep_deterministic_policy_gradient.py
+++ b/examples/deep_deterministic_policy_gradient.py
@@ -173,12 +173,12 @@ class DeepDeterministicPolicyGradient:
     for param, target_param in zip(
         optim.get_parameters(self.actor), optim.get_parameters(self.target_actor)
     ):
-      target_param.assign(param * tau + target_param * (1.0 - tau))
+      target_param.assign(param.detach() * tau + target_param * (1.0 - tau))
 
     for param, target_param in zip(
         optim.get_parameters(self.critic), optim.get_parameters(self.target_critic)
     ):
-      target_param.assign(param * tau + target_param * (1.0 - tau))
+      target_param.assign(param.detach() * tau + target_param * (1.0 - tau))
 
   def choose_action(self, state: Tensor, evaluate: bool = False) -> NDArray:
     mu = self.actor.forward(state, self.max_action)