diff --git a/.DS_Store b/.DS_Store
new file mode 100644
index 0000000..746009c
Binary files /dev/null and b/.DS_Store differ
diff --git a/agents/.DS_Store b/agents/.DS_Store
new file mode 100644
index 0000000..ede64cb
Binary files /dev/null and b/agents/.DS_Store differ
diff --git a/agents/ppo/agent.py b/agents/ppo/agent.py
index ebc7bba..b7215c0 100644
--- a/agents/ppo/agent.py
+++ b/agents/ppo/agent.py
@@ -102,16 +102,17 @@ class Agent:
                 self.critic.optimizer.zero_grad()
                 self.total_loss.backward()
 
+                T.nn.utils.clip_grad_norm_(
+                    self.actor.parameters(), max_norm=2)
+                T.nn.utils.clip_grad_norm_(
+                    self.critic.parameters(), max_norm=2)
+
                 # Calculate the gradient norms for both networks
                 actor_grad_norm = T.nn.utils.clip_grad_norm_(
                     self.actor.parameters(), max_norm=1)
                 critic_grad_norm = T.nn.utils.clip_grad_norm_(
                     self.critic.parameters(), max_norm=1)
 
-                T.nn.utils.clip_grad_norm_(
-                    self.actor.parameters(), max_norm=1)
-                T.nn.utils.clip_grad_norm_(
-                    self.critic.parameters(), max_norm=1)
                 # Log or print the gradient norms
                 print(f"Actor Gradient Norm: {actor_grad_norm}")
                 print(f"Critic Gradient Norm: {critic_grad_norm}")
diff --git a/agents/ppo/brain.py b/agents/ppo/brain.py
index aa0728c..6dc02b9 100644
--- a/agents/ppo/brain.py
+++ b/agents/ppo/brain.py
@@ -96,9 +96,9 @@ class CriticNetwork(nn.Module):
 
         self.critic = nn.Sequential(
             nn.Linear(input_dims, fc1_dims),
-            nn.ReLU(),
+            nn.LeakyReLU(),
             nn.Linear(fc1_dims, fc2_dims),
-            nn.ReLU(),
+            nn.LeakyReLU(),
             nn.Linear(fc2_dims, 1)
         )