diff --git a/neuralprophet/forecaster.py b/neuralprophet/forecaster.py
index 852fc297b..da28758a0 100644
--- a/neuralprophet/forecaster.py
+++ b/neuralprophet/forecaster.py
@@ -2734,7 +2734,12 @@ def _train(
         metrics_df = pd.DataFrame(self.metrics_logger.history)
         return metrics_df
 
-    def restore_trainer(self):
+    def restore_trainer(self, accelerator: Optional[str] = None):
+        """
+        If no accelerator was provided, use accelerator stored in model.
+        """
+        if accelerator is None:
+            accelerator = self.accelerator
         """
         Restore the trainer based on the forecaster configuration.
         """
@@ -2743,7 +2748,7 @@ def restore_trainer(self):
             config=self.trainer_config,
             metrics_logger=self.metrics_logger,
             early_stopping=self.early_stopping,
-            accelerator=self.accelerator,
+            accelerator=accelerator,
             metrics_enabled=bool(self.metrics),
         )
 
diff --git a/neuralprophet/utils.py b/neuralprophet/utils.py
index bf16f1ed7..ea245dc7f 100644
--- a/neuralprophet/utils.py
+++ b/neuralprophet/utils.py
@@ -92,10 +92,12 @@ def load(path: str, map_location=None):
         >>> from neuralprophet import load
         >>> model = load("test_save_model.np")
     """
+    torch_map_location = None
     if map_location is not None:
-        map_location = torch.device(map_location)
-    m = torch.load(path, map_location=map_location)
-    m.restore_trainer()
+        torch_map_location = torch.device(map_location)
+
+    m = torch.load(path, map_location=torch_map_location)
+    m.restore_trainer(accelerator=map_location)
     return m