class mindspore.ops.MultitypeFuncGraph(name, read_value=False)[source]

Generates overloaded functions.

MultitypeFuncGraph is a class used to generate overloaded functions, considering different types as inputs. Initialize an MultitypeFuncGraph object with name, and use register with input types as the decorator for the function to be registered. And the object can be called with different types of inputs, and work with HyperMap and Map.

  • name (str) – Operator name.

  • read_value (bool) – If the registered function not need to set value on Parameter, and all inputs will pass by value, set read_value to True. Default: False.


ValueError – If failed to find find a matching function for the given arguments.

Supported Platforms:

Ascend GPU CPU


>>> # `add` is a metagraph object which will add two objects according to
>>> # input type using ".register" decorator.
>>> from mindspore import Tensor
>>> from mindspore.ops import Primitive, operations as P
>>> from mindspore import dtype as mstype
>>> tensor_add = P.Add()
>>> add = MultitypeFuncGraph('add')
>>> @add.register("Number", "Number")
... def add_scala(x, y):
...     return x + y
>>> @add.register("Tensor", "Tensor")
... def add_tensor(x, y):
...     return tensor_add(x, y)
>>> output = add(1, 2)
>>> print(output)
>>> output = add(Tensor([0.1, 0.6, 1.2], dtype=mstype.float32), Tensor([0.1, 0.6, 1.2], dtype=mstype.float32))
>>> print(output)
[0.2 1.2 2.4]

Register a function for the given type string.


type_names (Union[str, mindspore.dtype]) – Inputs type names or types list.


decorator, a decorator to register the function to run, when called under the types described in type_names.