diff --git a/src/llmtuner/tuner/dpo/trainer.py b/src/llmtuner/tuner/dpo/trainer.py
index 1004bf60..572ce13d 100644
--- a/src/llmtuner/tuner/dpo/trainer.py
+++ b/src/llmtuner/tuner/dpo/trainer.py
@@ -42,7 +42,7 @@ class DPOPeftTrainer(PeftModelMixin, DPOTrainer):
 
         if ref_model is not None:
             if self.is_deepspeed_enabled:
-                self.ref_model = self.accelerator._prepare_deepspeed(self.ref_model)
+                self.ref_model, = self.accelerator._prepare_deepspeed(self.ref_model)
                 self.ref_model.eval()
             else:
                 self.ref_model = self.accelerator.prepare_model(self.ref_model, evaluation_mode=True)