torch.nn.Patameter()详解

torch.nn.Parameter是torch.Tensor的子类。
其主要作用是将不可训练的tensor变成可训练的参数，并将其绑定到module里面。
它与torch.Tensor的区别就是nn.Parameter的对象会被被加入到module自带的parameter()这个迭代器中去（见后面Linear例子）；而module中非nn.Parameter对象的普通tensor是不在parameter()这个迭代器中的。
注意：nn.Parameter的对象的requires_grad属性的默认值是True，即可被训练的，这与torth.Tensor对象的默认值相反。

用法：
torch.nn.Parameter(Tensor data, bool requires_grad)
其中：data为传入Tensor类型参数，requires_grad默认值为True，表示可训练，False表示不可训练。

在nn.Module类中，也是使用nn.Parameter来对每一个module的参数进行初始化的，以torch.nn.Linear类为例：

class Linear(Module):
    r"""Applies a linear transformation to the incoming data: :math:`y = xA^T + b`
    Args:
        in_features: size of each input sample
        out_features: size of each output sample
        bias: If set to ``False``, the layer will not learn an additive bias.
            Default: ``True``
    Shape:
        - Input: :math:`(N, *, H_{in})` where :math:`*` means any number of
          additional dimensions and :math:`H_{in} = \text{in\_features}`
        - Output: :math:`(N, *, H_{out})` where all but the last dimension
          are the same shape as the input and :math:`H_{out} = \text{out\_features}`.
    Attributes:
        weight: the learnable weights of the module of shape
            :math:`(\text{out\_features}, \text{in\_features})`. The values are
            initialized from :math:`\mathcal{U}(-\sqrt{k}, \sqrt{k})`, where
            :math:`k = \frac{1}{\text{in\_features}}`
        bias:   the learnable bias of the module of shape :math:`(\text{out\_features})`.
                If :attr:`bias` is ``True``, the values are initialized from
                :math:`\mathcal{U}(-\sqrt{k}, \sqrt{k})` where
                :math:`k = \frac{1}{\text{in\_features}}`
    Examples::
        >>> m = nn.Linear(20, 30)
        >>> input = torch.randn(128, 20)
        >>> output = m(input)
        >>> print(output.size())
        torch.Size([128, 30])
    """
    __constants__ = ['in_features', 'out_features']
 
    def __init__(self, in_features, out_features, bias=True):
        super(Linear, self).__init__()
        self.in_features = in_features
        self.out_features = out_features
        self.weight = Parameter(torch.Tensor(out_features, in_features))
        if bias:
            self.bias = Parameter(torch.Tensor(out_features))
        else:
            self.register_parameter('bias', None)
        self.reset_parameters()
 
    def reset_parameters(self):
        init.kaiming_uniform_(self.weight, a=math.sqrt(5))
        if self.bias is not None:
            fan_in, _ = init._calculate_fan_in_and_fan_out(self.weight)
            bound = 1 / math.sqrt(fan_in)
            init.uniform_(self.bias, -bound, bound)
 
    def forward(self, input):
        return F.linear(input, self.weight, self.bias)
 
    def extra_repr(self):
        return 'in_features={}, out_features={}, bias={}'.format(
            self.in_features, self.out_features, self.bias is not None
        )

可见，在初始化函数__init__()中，代码：
self.weight = Parameter(torch.Tensor(out_features, in_features))
通过nn.Parameter()对象对weights进行了初始化，bias也是一样。
（与torch.tensor([1,2,3],requires_grad=True)的区别，这个只是将参数变成可训练的，并没有绑定在module的parameter()列表中。)

我们生成一个Linear对象模型，看看其parameter列表中是否含有weigs和bias这两个可训练参数：

import torch.nn as nn

my_linear = nn.Linear(2,3)  # 生成一个线性层对象（这也是一个完整的module）
for i in my_linear.named_parameters():  # 查看该module包含多少个可训练的参数(也可用*.parameters())
    print(i)

输出：
('weight', Parameter containing:
tensor([[-0.2682, -0.5847],
        [ 0.6815, -0.4122],
        [ 0.2677,  0.2913]], requires_grad=True))
('bias', Parameter containing:
tensor([ 0.0536, -0.3389, -0.6452], requires_grad=True))

参考：
https://zhuanlan.zhihu.com/p/368808794