diff --git a/verl/trainer/ppo/ray_trainer.py b/verl/trainer/ppo/ray_trainer.py
index 02ac4d67185..5f33f6e3af3 100644
--- a/verl/trainer/ppo/ray_trainer.py
+++ b/verl/trainer/ppo/ray_trainer.py
@@ -1105,9 +1105,7 @@ def fit(self):
                             batch = batch.union(reward_tensor)
 
                         if self.config.reward_model.launch_reward_fn_async:
-                            future_reward = compute_reward_async.remote(
-                                data=batch, config=self.config, tokenizer=self.tokenizer
-                            )
+                            future_reward = compute_reward_async.remote(data=batch, reward_fn=self.reward_fn)
                         else:
                             reward_tensor, reward_extra_infos_dict = compute_reward(batch, self.reward_fn)