fix(pt): pass device and dtype in SeZMDeNSFittingNet.deserialize

safe_numpy_to_tensor takes `device` and `dtype` as required keyword-only
arguments, so SeZMDeNSFittingNet.deserialize raised TypeError. Rebuild the
state dict from the instantiated module's own state_dict, matching the
other sezm_nn deserialize implementations.

Signed-off-by: Anai-Guo <antai12232931@outlook.com>

Author: Anai-Guo

deepmd/pt/model/descriptor/sezm_nn/dens.py

@@ -746,6 +746,12 @@ def deserialize(cls, data: dict[str, Any]) -> SeZMDeNSFittingNet:
         config = data.pop("config")
         variables = data.pop("@variables")
         obj = cls(**config)
-        state = {key: safe_numpy_to_tensor(value) for key, value in variables.items()}
+        template = obj.state_dict()
+        state = {
+            key: safe_numpy_to_tensor(
+                value, device=template[key].device, dtype=template[key].dtype
+            )
+            for key, value in variables.items()
+        }
         obj.load_state_dict(state)
         return obj