[docs] Update custom saving and loading

docs/tutorial_toolbox/saving_and_loading.ipynb

@@ -245,9 +245,23 @@
     }
    },
    "source": [
-    "You can make your own saving and loading functions easily. Beacause all variables in the model can be easily collected through ``.vars()``. Therefore, saving variables is just transforming these variables to numpy.ndarray and then storing them into the disk. Similarly, to load variables, you just need read the numpy arrays from the disk and then transform these arrays as instances of [Variables](../tutorial_math/variables.ipynb). \n",
+    "You can make your own saving and loading functions easily. Beacause all variables in the model can be easily collected through ``.vars()``.\n",
     "\n",
-    "The only gotcha to pay attention to is to avoid saving duplicated variables. "
+    "For customizing the saving, users can use:\n",
+    "\n",
+    "```python\n",
+    "class YourClass(bp.BrainPyObject):\n",
+    "   def __save_state__(self):\n",
+    "     ...\n",
+    "```\n",
+    "\n",
+    "For customizing the loading, users can use:\n",
+    "\n",
+    "```python\n",
+    "class YourClass(bp.BrainPyObject):\n",
+    "   def __load_state__(self, state_dict):\n",
+    "     ...\n",
+    "```"
    ]
   }
  ],