Spaces:
Sleeping
Sleeping
import torch | |
from torch import nn | |
class LNN(nn.Module): | |
# 创建一个全连接网络用于手写数字识别,并通过一个参数dim控制中间层的维度 | |
def __init__(self, dim=32): | |
super(LNN, self).__init__() | |
self.fc1 = nn.Linear(28 * 28, dim) | |
self.fc2 = nn.Linear(dim, 10) | |
def forward(self, x): | |
x = x.view(-1, x.shape[1] * x.shape[2] * x.shape[3]) | |
x = torch.relu(self.fc1(x)) | |
x = self.fc2(x) | |
return x | |
if __name__ == "__main__": | |
input = torch.randn(2, 1, 28, 28) | |
model = LNN() | |
output = model(input) | |
assert output.shape == (2, 10) | |