diff --git a/benchmark_marl.py b/benchmark_marl.py
index 614424f9..3c6a0e70 100644
--- a/benchmark_marl.py
+++ b/benchmark_marl.py
@@ -4,9 +4,9 @@
 
 def parse_args():
     parser = argparse.ArgumentParser("Run an MARL demo.")
-    parser.add_argument("--method", type=str, default="vdac")
-    parser.add_argument("--env", type=str, default="sc2")
-    parser.add_argument("--env-id", type=str, default="3m")
+    parser.add_argument("--method", type=str, default="ippo")
+    parser.add_argument("--env", type=str, default="mpe")
+    parser.add_argument("--env-id", type=str, default="simple_spread_v3")
     parser.add_argument("--seed", type=int, default=10)
     parser.add_argument("--test", type=int, default=0)
     parser.add_argument("--device", type=str, default="cuda:0")
diff --git a/xuance/environment/pettingzoo/pettingzoo_env.py b/xuance/environment/pettingzoo/pettingzoo_env.py
index 3eeca9b7..4609e5f1 100644
--- a/xuance/environment/pettingzoo/pettingzoo_env.py
+++ b/xuance/environment/pettingzoo/pettingzoo_env.py
@@ -9,7 +9,8 @@ class PettingZoo_Env(ParallelEnv):
     def __init__(self, env_name: str, env_id: str, seed: int, **kwargs):
         super(PettingZoo_Env, self).__init__()
         scenario = importlib.import_module('pettingzoo.' + env_name + '.' + env_id)
-        self.env = scenario.parallel_env(continuous_actions=kwargs["continuous"],
+        self.continuous_actions = kwargs["continuous"]
+        self.env = scenario.parallel_env(continuous_actions=self.continuous_actions,
                                          render_mode=kwargs["render_mode"])
         self.scenario_name = env_name + "." + env_id
         self.n_handles = len(AGENT_NAME_DICT[self.scenario_name])
@@ -53,8 +54,9 @@ def reset(self, seed=None, options=None):
         return observations, reset_info
 
     def step(self, actions):
-        for k, v in actions.items():
-            actions[k] = np.clip(v, self.action_spaces[k].low, self.action_spaces[k].high)
+        if self.continuous_actions:
+            for k, v in actions.items():
+                actions[k] = np.clip(v, self.action_spaces[k].low, self.action_spaces[k].high)
         observations, rewards, terminations, truncations, infos = self.env.step(actions)
         for k, v in rewards.items():
             self.individual_episode_reward[k] += v