MCPcopy Create free account
hub / github.com/PaddlePaddle/Paddle / create_parameter

Method create_parameter

python/paddle/nn/layer/layers.py:1012–1071  ·  view source on GitHub ↗

Create parameters for this layer. Parameters: shape(list): Shape of the parameter. The data type in the list must be int. attr(ParamAttr, optional): Parameter attribute of weight. Please refer to :ref:`api_paddle_ParamAttr`. Default: None. dtype(str, opti

(
        self,
        shape: ShapeLike,
        attr: ParamAttrLike | None = None,
        dtype: DTypeLike | None = None,
        is_bias: bool = False,
        default_initializer: Initializer | None = None,
        device: PlaceLike | None = None,
    )

Source from the content-addressed store, hash-verified

1010 return hook_remove_helper
1011
1012 def create_parameter(
1013 self,
1014 shape: ShapeLike,
1015 attr: ParamAttrLike | None = None,
1016 dtype: DTypeLike | None = None,
1017 is_bias: bool = False,
1018 default_initializer: Initializer | None = None,
1019 device: PlaceLike | None = None,
1020 ) -> Tensor:
1021 """Create parameters for this layer.
1022
1023 Parameters:
1024 shape(list): Shape of the parameter. The data type in the list must be int.
1025 attr(ParamAttr, optional): Parameter attribute of weight. Please refer to :ref:`api_paddle_ParamAttr`. Default: None.
1026 dtype(str, optional): Data type of this parameter.
1027 If set str, it can be "bool", "float16", "float32", "float64",
1028 "int8", "int16", "int32", "int64", "uint8" or "uint16". Default: "float32".
1029 is_bias(bool, optional): if this is a bias parameter. Default: False.
1030 default_initializer(Initializer, optional): the default initializer for this parameter.
1031 If set None, default initializer will be set to paddle.nn.initializer.Xavier and paddle.nn.initializer.Constant
1032 for non-bias and bias parameter, respectively. Default: None.
1033 device(PlaceLike, optional): the device place for the parameter. Default: None.
1034
1035 Returns:
1036 :Tensor, created parameter.
1037
1038 Examples:
1039 .. code-block:: pycon
1040
1041 >>> import paddle
1042 >>> paddle.seed(2023)
1043
1044 >>> class MyLayer(paddle.nn.Layer):
1045 ... def __init__(self):
1046 ... super().__init__()
1047 ... self._linear = paddle.nn.Linear(1, 1)
1048 ... w_tmp = self.create_parameter([1, 1])
1049 ... self.add_parameter("w_tmp", w_tmp)
1050 ...
1051 ... def forward(self, input):
1052 ... return self._linear(input)
1053 >>> mylayer = MyLayer()
1054 >>> for name, param in mylayer.named_parameters():
1055 ... print(name, param) # will print w_tmp,_linear.weight,_linear.bias
1056 w_tmp Parameter containing:
1057 Tensor(shape=[1, 1], dtype=float32, place=Place(cpu), stop_gradient=False,
1058 [[0.06979191]])
1059 _linear.weight Parameter containing:
1060 Tensor(shape=[1, 1], dtype=float32, place=Place(cpu), stop_gradient=False,
1061 [[1.26729357]])
1062 _linear.bias Parameter containing:
1063 Tensor(shape=[1], dtype=float32, place=Place(cpu), stop_gradient=False,
1064 [0.])
1065 """
1066 temp_attr = copy.deepcopy(attr)
1067 if isinstance(temp_attr, str) and temp_attr == "":
1068 temp_attr = None
1069 return self._helper.create_parameter(

Callers 15

__init__Method · 0.95
test_create_parameterMethod · 0.95
applyMethod · 0.45
removeMethod · 0.45
applyMethod · 0.45
__init__Method · 0.45
__init__Method · 0.45
__init__Method · 0.45
__init__Method · 0.45
__init__Method · 0.45
__init__Method · 0.45

Calls 1

deepcopyMethod · 0.45

Tested by 3

__init__Method · 0.76
test_create_parameterMethod · 0.76