In [1]: import torch
In [2]: import torch.nn as nn

In [3]: conv = nn.Conv2d(in_channels=2, out_channels=1, kernel_size=3)

# 此時weight以及bias已由nn.Conv2d初始化
conv.weight, conv.bias
Out[4]: 
(Parameter containing:
 tensor([[[[-0.0335,  0.0855, -0.0708],
           [-0.1672,  0.0902, -0.0077],
           [-0.0838, -0.1539, -0.0933]],
 
          [[-0.0496,  0.1807, -0.1477],
           [ 0.0397,  0.1963,  0.0932],
           [-0.2018, -0.0436,  0.1971]]]], requires_grad=True),
 Parameter containing:
 tensor([-0.1963], requires_grad=True))

# 手動設(shè)定
# conv.weight.data 以及 conv.bias.data屬性為torch.tensor
# 因此只要獲取conv.weight.data以及conv.bias.data屬性，后續(xù)調(diào)用torch.tensor的不同方法即可進行修改
# 例如：全部修改為0
In [5]: conv.weight.data.zero_(), conv.bias.data.zero_()

In [6]: conv.weight, conv.bias
Out[6]: 
(Parameter containing:
 tensor([[[[0., 0., 0.],
           [0., 0., 0.],
           [0., 0., 0.]],
 
          [[0., 0., 0.],
           [0., 0., 0.],
           [0., 0., 0.]]]], requires_grad=True),
 Parameter containing:
 tensor([0.], requires_grad=True))

Conv2d(in_channels, out_channels, kernel_size, stride=1,padding=0, dilation=1, groups=1,bias=True, padding_mode=‘zeros')

class CNN(nn.Module):
    def __init__(self,in_channels:int,out_channels:int):
        """
        創(chuàng)建一個卷積神經(jīng)網(wǎng)絡(luò)
        網(wǎng)絡(luò)只有兩層
        :param in_channels: 輸入通道數(shù)量
        :param out_channels: 輸出通道數(shù)量
        """
        super(CNN).__init__()
        self.conv1=nn.Conv2d(in_channels,10,3,stride=1,padding=1)
        self.pool1=nn.MaxPool2d(kernel_size=2,stride=1)
        self.conv2=nn.Conv2d(10,out_channels,3,stride=1,padding=1)
        self.pool2=nn.MaxPool2d(kernel_size=2,stride=1)
    def forward(self,x):
        """
        前向傳播函數(shù)
        :param x:  輸入，tensor 類型
        :return: 返回結(jié)果
        """
        out=self.conv1(x)
        out=self.pool1(out)
        out=self.conv2(out)
        out=self.pool2(out)
        return out