diff --git a/ts_benchmark/baselines/deep_forecasting_model_base.py b/ts_benchmark/baselines/deep_forecasting_model_base.py
index 51c4929..fa7df90 100644
--- a/ts_benchmark/baselines/deep_forecasting_model_base.py
+++ b/ts_benchmark/baselines/deep_forecasting_model_base.py
@@ -715,6 +715,7 @@ def forecast(
                         output = output[:, -config.horizon:, :series_dim]
                         break
 
+
                 column_num = output.shape[-1]
                 temp = output.cpu().numpy().reshape(-1, column_num)[-config.horizon:]
 
@@ -873,7 +874,7 @@ def _perform_rolling_predictions(
         """
         rolling_time = 0
         input_np, target_np, input_mark_np, target_mark_np = self._get_rolling_data(
-            input_np, None, all_mark, rolling_time
+            input_np, None, None, all_mark, rolling_time
         )
         if exog_future is not None:
             rolling_time_sum = horizon // self.config.horizon + 1
@@ -933,12 +934,13 @@ def _perform_rolling_predictions(
                     break
                 rolling_time += 1
                 output = output.cpu().numpy()[:, -self.config.horizon :, :]
+                # 这里的_get_rolling_data函数需要修改：
                 (
                     input_np,
                     target_np,
                     input_mark_np,
                     target_mark_np,
-                ) = self._get_rolling_data(input_np, output, all_mark, rolling_time)
+                ) = self._get_rolling_data(input_np, output, exog_future, all_mark, rolling_time)
 
         answers = np.concatenate(answers, axis=1)
         return answers[:, -horizon:, :]
@@ -947,6 +949,7 @@ def _get_rolling_data(
         self,
         input_np: np.ndarray,
         output: Optional[np.ndarray],
+        exog_future: Optional[torch.Tensor],
         all_mark: np.ndarray,
         rolling_time: int,
     ) -> Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray]:
@@ -960,6 +963,15 @@ def _get_rolling_data(
         :return: Updated input data, target data, input marks, and target marks for rolling prediction.
         """
         if rolling_time > 0:
+            if exog_future!= None:
+                exog_output = exog_future[
+                    :,
+                    rolling_time
+                    * self.config.horizon : (rolling_time + 1)
+                    * self.config.horizon,
+                    :,
+                ].cpu().numpy()
+                output = np.concatenate((output, exog_output), axis=2)
             input_np = np.concatenate((input_np, output), axis=1)
             input_np = input_np[:, -self.config.seq_len :, :]
         target_np = np.zeros(