adding weight decay for experimentation

ml-agents/mlagents/trainers/ppo/optimizer_torch.py

         )
 
         self.optimizer = torch.optim.Adam(
-            params, lr=self.trainer_settings.hyperparameters.learning_rate
+            params,
+            lr=self.trainer_settings.hyperparameters.learning_rate,
+            weight_decay=1e-6,
         )
         self.stats_name_to_update_name = {
             "Losses/Value Loss": "value_loss",

ml-agents/mlagents/trainers/sac/optimizer_torch.py

             self.trainer_settings.max_steps,
         )
         self.policy_optimizer = torch.optim.Adam(
-            policy_params, lr=hyperparameters.learning_rate
+            policy_params, lr=hyperparameters.learning_rate, weight_decay=1e-6
-            value_params, lr=hyperparameters.learning_rate
+            value_params, lr=hyperparameters.learning_rate, weight_decay=1e-6
         )
         self.entropy_optimizer = torch.optim.Adam(
             self._log_ent_coef.parameters(), lr=hyperparameters.learning_rate