!9441 modify example

From: @lijiaqi0612
Reviewed-by: @c_34,@zh_qh
Signed-off-by: @c_34

mindspore/nn/wrap/cell_wrapper.py

@@ -205,9 +205,9 @@ class TrainOneStepCell(Cell):
         >>> train_net = nn.TrainOneStepCell(loss_net, optim)
         >>>
         >>> #2) Using user-defined WithLossCell
-        >>>class MyWithLossCell(nn.cell):
+        >>>class MyWithLossCell(nn.Cell):
         >>>    def __init__(self, backbone, loss_fn):
-        >>>        super(WithLossCell, self).__init__(auto_prefix=False)
+        >>>        super(MyWithLossCell, self).__init__(auto_prefix=False)
         >>>        self._backbone = backbone
         >>>        self._loss_fn = loss_fn
         >>>
@@ -215,6 +215,10 @@ class TrainOneStepCell(Cell):
         >>>        out = self._backbone(x, y)
         >>>        return self._loss_fn(out, label)
         >>>
+        >>>    @property
+        >>>    def backbone_network(self):
+        >>>        return self._backbone
+        >>>
         >>> loss_net = MyWithLossCell(net, loss_fn)
         >>> train_net = nn.TrainOneStepCell(loss_net, optim)
     """