简单实现lenet-5：
#coding=utf-8
import torch
from torch.autograd import Variable
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
# 建立了两个卷积层，第一层1 个通道输入, 6个输出通道, 5x5 卷积核
self.conv1 = nn.Conv2d(1, 6, 5)
self.conv2 = nn.Conv2d(6, 16, 5)
#三个全连接层，y = Wx + b 这里没有做激活/非线性操作
self.fc1 = nn.Linear(16 * 5 * 5, 120)
self.fc2 = nn.Linear(120, 84)
self.fc3 = nn.Linear(84, 10)
def forward(self, x): #2D卷积层的输入data维数是 batchsize*channel*height*width
# 最大池化
x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
x = F.max_pool2d(F.relu(self.conv2(x)), 2)
x = x.view(-1, self.num_flat_features(x))
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
def num_flat_features(self, x):
size = x.size()[1:] # all dimensions except the batch dimension
num_features = 1
for s in size:
num_features *= s
return num_features
net =Net()
print net
#网络的可学习参数通过net.parameters()返回
params = list(net.parameters())
print len(params)
print params[0].size()
input = Variable(torch.randn(1,1,32,32))
print input
out = net(input)
print out
#将网络中所有的参数梯度清零
net.zero_grad()
out.backward(torch.randn(1,10))
#损失函数
output = net(input)
target = Variable(torch.arange(1,11))
criterion = nn.MSELoss()
loss = criterion(output,target)
print '---' * 10
print 'loss:',loss
#反向传播
net.zero_grad() #将之前的梯度清零
print 'net.conv1.bias.grad brfore backward:'
print net.conv1.bias.grad
loss.backward()
print 'net.conv1.bias.grad after backward:'
print net.conv1.bias.grad
import torch.optim as optim
#新建一个优化器，指定要调整的参数和学习率
optimizer = optim.SGD(net.parameters(),lr = 0.01)
optimizer.zero_grad()
output = net(input)
loss = criterion(output,target)
loss.backward()
optimizer.step() #更新参数
结果：
Net (
(conv1): Conv2d(1, 6, kernel_size=(5, 5), stride=(1, 1))
(conv2): Conv2d(6, 16, kernel_size=(5, 5), stride=(1, 1))
(fc1): Linear (400 -> 120)
(fc2): Linear (120 -> 84)
(fc3): Linear (84 -> 10)
)
10
(6L, 1L, 5L, 5L)
Variable containing:
(0 ,0 ,.,.) =
0.5187 -0.7634 -0.7736 ... 0.7010 0.7504 0.4978
1.0905 0.7255 -1.4743 ... 1.8054 -1.3483 0.5112
-0.0973 0.3014 0.7019 ... 0.3816 -0.2391 -1.6550
... ⋱ ...
-0.4778 0.3382 -0.0205 ... 0.1119 -0.1873 0.3986
0.4438 -0.3040 -2.1828 ... -0.4498 -0.5209 -1.2561
-1.5232 1.4791 0.1453 ... 0.5937 0.1218 -0.1614
[torch.FloatTensor of size 1x1x32x32]
Variable containing:
-0.0099 0.0765 -0.1277 0.1134 -0.0274 -0.0051 -0.0470 -0.0740 -0.1531 0.1007
[torch.FloatTensor of size 1x10]
------------------------------
loss: Variable containing:
38.7571
[torch.FloatTensor of size 1]
net.conv1.bias.grad brfore backward:
Variable containing:
0
0
0
0
0
0
[torch.FloatTensor of size 6]
net.conv1.bias.grad after backward:
Variable containing:
-0.0227
-0.0524
-0.0884
0.0287
0.1252
-0.1181
[torch.FloatTensor of size 6]