diff --git a/docarray/typing/tensor/torch_tensor.py b/docarray/typing/tensor/torch_tensor.py
index 32d1cbff7b..dc5b0587c0 100644
--- a/docarray/typing/tensor/torch_tensor.py
+++ b/docarray/typing/tensor/torch_tensor.py
@@ -130,7 +130,7 @@ def validate(
         field: 'ModelField',
         config: 'BaseConfig',
     ) -> T:
-        if isinstance(value, TorchTensor):
+        if isinstance(value, cls):
             return cast(T, value)
         elif isinstance(value, torch.Tensor):
             return cls._docarray_from_native(value)
@@ -195,10 +195,7 @@ def _docarray_from_native(cls: Type[T], value: torch.Tensor) -> T:
         :param value: the native `torch.Tensor`
         :return: a `TorchTensor`
         """
-        if cls.__unparametrizedcls__:  # This is not None if the tensor is parametrized
-            value.__class__ = cls.__unparametrizedcls__  # type: ignore
-        else:
-            value.__class__ = cls
+        value.__class__ = cls
         return cast(T, value)
 
     @classmethod
@@ -254,11 +251,11 @@ def __torch_function__(cls, func, types, args=(), kwargs=None):
         # this tells torch to treat all of our custom tensors just like
         # torch.Tensor's. Otherwise, torch will complain that it doesn't
         # know how to handle our custom tensor type.
-        docarray_torch_tensors = TorchTensor.__subclasses__()
+        docarray_torch_tensors = cls.__subclasses__() + [cls]
         types_ = tuple(
             torch.Tensor if t in docarray_torch_tensors else t for t in types
         )
-        return super().__torch_function__(func, types_, args, kwargs)
+        return torch.Tensor.__torch_function__(func, types_, args, kwargs)
 
     @classmethod
     def _docarray_from_ndarray(cls: Type[T], value: np.ndarray) -> T: