【Pytorch入門到精通】nn.Linear

<code>torch.nn.Linear（in_features，out_features，bias = True ）/<code>

模塊用途：執行線性變換 y = A x+ b

參數說明：

in_features :每個輸入sample特徵維度

out_features :每個輸出樣本的維度

bias : 如果設置為False，則該層不學習偏差參數。默認值：True

應用Linear實現MLP：

<code>import torch
from torch import nn


class MLP(nn.Module):
    def __init__(self, **kwargs):
        super(MLP, self).__init__(**kwargs)
        self.hidden = nn.Linear(784, 256)
        self.act = nn.ReLU()
        self.output = nn.Linear(256, 10)

    def forward(self, x):
        a = self.act(self.hidden(x))
        return self.output(a)


X = torch.rand(2, 784)
net = MLP()
output = net(X)
print(net)
print(output)
print(output.size())/<code>

執行輸出：

<code>(base) PS D:\\workspace\\Git\\python> conda activate base
(base) PS D:\\workspace\\Git\\python>  ${env:DEBUGPY_LAUNCHER_PORT}='14659'; & 'C:\\ProgramData\\Anaconda3\\python.exe' 'c:\\Users\\renhao\\.vscode\\extensions\\ms-python.python-2020.3.71659\\pythonFiles\\lib\\python\\debugpy\\wheels\\debugpy\\launcher' 'd:\\workspace\\Git\\python\\Pytorch\\torchModuleLearn.py'
MLP(
  (hidden): Linear(in_features=784, out_features=256, bias=True)
  (act): ReLU()
)
tensor([[-0.0893,  0.0537,  0.1864, -0.0603,  0.1481,  0.2578,  0.1449,  0.0449,
         -0.0525, -0.0684],
        [-0.1521, -0.0388,  0.0725,  0.0723,  0.0686,  0.1950, -0.0641,  0.0938,
         -0.0604,  0.0821]], grad_fn=<addmmbackward>)
(base) PS D:\\workspace\\Git\\python>  ${env:DEBUGPY_LAUNCHER_PORT}='3318'; & 'C:\\ProgramData\\Anaconda3\\python.exe' 'c:\\Users\\renhao\\.vscode\\extensions\\ms-python.python-2020.3.71659\\pythonFiles\\lib\\python\\debugpy\\wheels\\debugpy\\launcher' 'd:\\workspace\\Git\\python\\Pytorch\\torchModuleLearn.py' 
MLP(
  (hidden): Linear(in_features=784, out_features=256, bias=True)
  (act): ReLU()
  (output): Linear(in_features=256, out_features=10, bias=True)
)
tensor([[-0.1295,  0.0865, -0.0277,  0.1044,  0.2579,  0.1228, -0.1272,  0.0310,
         -0.2329, -0.0463],
        [-0.1160,  0.0768,  0.1460,  0.2687,  0.0801, -0.0196, -0.0890,  0.0590,
         -0.1971,  0.0714]], grad_fn=<addmmbackward>)
torch.Size([2, 10])/<addmmbackward>/<addmmbackward>/<code>

閱讀更多 深度視野 的文章

關鍵字: 入門 精通 Python