diff --git a/vertexai/preview/_workflow/serialization_engine/serializers.py b/vertexai/preview/_workflow/serialization_engine/serializers.py
index e4dca7e3a5..cf49f1f2b8 100644
--- a/vertexai/preview/_workflow/serialization_engine/serializers.py
+++ b/vertexai/preview/_workflow/serialization_engine/serializers.py
@@ -144,6 +144,7 @@ def _is_valid_gcs_path(path: str) -> bool:
 
 def _load_torch_model(path: str, map_location: "torch.device") -> "torch.nn.Module":
     import torch
+
     try:
         return torch.load(path, map_location=map_location)
     except Exception:
@@ -434,7 +435,9 @@ class TorchModelSerializer(serializers_base.Serializer):
         serializers_base.SerializationMetadata(serializer="TorchModelSerializer")
     )
 
-    def serialize(self, to_serialize: "torch.nn.Module", gcs_path: str, **kwargs) -> str:
+    def serialize(
+        self, to_serialize: "torch.nn.Module", gcs_path: str, **kwargs
+    ) -> str:
         """Serializes a torch.nn.Module to a gcs path.
 
         Args:
@@ -450,6 +453,7 @@ def serialize(self, to_serialize: "torch.nn.Module", gcs_path: str, **kwargs) ->
             ValueError: if `gcs_path` is not a valid GCS uri.
         """
         import torch
+
         del kwargs
         if not _is_valid_gcs_path(gcs_path):
             raise ValueError(f"Invalid gcs path: {gcs_path}")
@@ -500,11 +504,18 @@ def deserialize(self, serialized_gcs_path: str, **kwargs) -> "torch.nn.Module":
         except ImportError as e:
             raise ImportError("torch is not installed.") from e
 
-        map_location = (
-            torch._GLOBAL_DEVICE_CONTEXT.device
-            if torch._GLOBAL_DEVICE_CONTEXT
-            else None
-        )
+        # Get the default device in the local torch environment.
+        # If `set_default_device` hasn't been called, _GLOBAL_DEVICE_CONTEXT
+        # should be None, then we set map_location to None as well.
+        map_location = None
+        # In torch 2.3.0, get_default_device is introduced
+        if hasattr(torch._GLOBAL_DEVICE_CONTEXT, "device_context") and hasattr(
+            torch, "get_default_device"
+        ):
+            map_location = torch.get_default_device()
+        # For older versions, we get default device from _GLOBAL_DEVICE_CONTEXT
+        elif hasattr(torch._GLOBAL_DEVICE_CONTEXT, "device"):
+            map_location = torch._GLOBAL_DEVICE_CONTEXT.device
 
         if serialized_gcs_path.startswith("gs://"):
             with tempfile.NamedTemporaryFile() as temp_file:
@@ -731,7 +742,9 @@ class TorchDataLoaderSerializer(serializers_base.Serializer):
         serializers_base.SerializationMetadata(serializer="TorchDataLoaderSerializer")
     )
 
-    def _serialize_to_local(self, to_serialize: "torch.utils.data.DataLoader", path: str):
+    def _serialize_to_local(
+        self, to_serialize: "torch.utils.data.DataLoader", path: str
+    ):
         """Serializes a torch.utils.data.DataLoader to a local path.
 
         Args:
@@ -778,6 +791,7 @@ def _serialize_to_local(self, to_serialize: "torch.utils.data.DataLoader", path:
         # for default batch sampler we store batch_size, drop_last, and sampler object
         # but not batch sampler object.
         import torch
+
         if isinstance(to_serialize.batch_sampler, torch.utils.data.BatchSampler):
             pass_through_args["batch_size"] = to_serialize.batch_size
             pass_through_args["drop_last"] = to_serialize.drop_last
@@ -797,7 +811,9 @@ def _serialize_to_local(self, to_serialize: "torch.utils.data.DataLoader", path:
         with open(f"{path}/pass_through_args.json", "w") as f:
             json.dump(pass_through_args, f)
 
-    def serialize(self, to_serialize: "torch.utils.data.DataLoader", gcs_path: str, **kwargs) -> str:
+    def serialize(
+        self, to_serialize: "torch.utils.data.DataLoader", gcs_path: str, **kwargs
+    ) -> str:
         """Serializes a torch.utils.data.DataLoader to a gcs path.
 
         Args:
@@ -883,7 +899,9 @@ def _deserialize_from_local(self, path: str) -> "torch.utils.data.DataLoader":
 
         return torch.utils.data.DataLoader(**kwargs)
 
-    def deserialize(self, serialized_gcs_path: str, **kwargs) -> "torch.utils.data.DataLoader":
+    def deserialize(
+        self, serialized_gcs_path: str, **kwargs
+    ) -> "torch.utils.data.DataLoader":
         """Deserialize a torch.utils.data.DataLoader given the gcs path.
 
         Args: