Pytorch官网的例子是一个LeNet网络：
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(3, 6, 5)
self.pool = nn.MaxPool2d(2, 2)
self.conv2 = nn.Conv2d(6, 16, 5)
self.fc1 = nn.Linear(16 * 5 * 5, 120)
self.fc2 = nn.Linear(120, 84)
self.fc3 = nn.Linear(84, 10)
def forward(self, x):
x = self.pool(F.relu(self.conv1(x)))
x = self.pool(F.relu(self.conv2(x)))
x = x.view(-1, 16 * 5 * 5)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
net = Net()从这个例子可以看到，Net这个类继承了父类（nn.Module）,而这个类主要包含两个函数：
- __init__(self): 初始化函数
- forward(self,x): 前传函数
在初始化函数中super(Net, self).__init__()这个初始化表达方式是为了防止多次初始化。self.conv1 = Conv2d(3,6,5)就是初始化一个输入是3，输出是6的5*5卷积。然后再forward函数时调用这个self.conv1.这个，也是我想讲的第一种初始化形式。
第二种初始化形式，是我再vgg16中看到的
class VGG(nn.Module):
def __init__(self, features, num_classes=1000, init_weights=True):
super(VGG, self).__init__()
self.features = nn.Sequential(
nn.Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False),
nn.Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False),
nn.Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False),
nn.Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2, stride=1, padding=1, dilation=1, ceil_mode=False),
nn.Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)),
nn.ReLU(),
nn.MaxPool2d(kernel_size=2, stride=1, padding=0, dilation=1, ceil_mode=False))
self.classifier = nn.Sequential(
nn.Linear(512 * 7 * 7, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, num_classes),
)
if init_weights:
self._initialize_weights()
def forward(self, x):
x = self.features(x)
x = x.view(x.size(0), -1)
x = self.classifier(x)
return x
我将features部分拓展开写了，当然官网给出的形式更为简洁，vgg16中定义了两个nn.Sequential，一个是features就是特征提取部分，一个是classfier分类部分。如果数据只是串行的形式，我是觉得nn.Sequential这样的形式看起来更整齐，而且在调用某一部分时也更清晰，比如在我的训练优化其中，vgg16部分学习率为10e-5, 其他部分是10e-4，我的代码实现如下：
gparam = list(map(id, net.features.parameters()))
base_params = filter(lambda p : id(p) not in gparam, net.parameters())
# define a optimizer with different learning rate
optimizer = optim.SGD([
{'params': base_params},
{'params': net.features.parameters(), 'lr': 0.00001}], lr=0.001, momentum=0.9)用net.features.parameters()就很好的区分出vgg16的features部分。
nn.Module这个主要函数分两个：在net=Net()时只调用了初始化部分，也就是创建了一部分网络。只有当output=Net(input)时才调用了forward进行前传。
网络有两种简单的创建形式：第一种简单定义self.conv1= Conv2d(3,6,5) 第二种以nn.Sequential. 当然更复杂的网络可以利用更多的形式去简化创建代码。这里可以参见pytorch中给出的vgg部分代码如下：
class VGG(nn.Module):
def __init__(self, features, num_classes=1000, init_weights=True):
super(VGG, self).__init__()
self.features = features
self.classifier = nn.Sequential(
nn.Linear(512 * 7 * 7, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, num_classes),
)
if init_weights:
self._initialize_weights()
def forward(self, x):
x = self.features(x)
x = x.view(x.size(0), -1)
x = self.classifier(x)
return x
def _initialize_weights(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
m.weight.data.normal_(0, math.sqrt(2. / n))
if m.bias is not None:
m.bias.data.zero_()
elif isinstance(m, nn.BatchNorm2d):
m.weight.data.fill_(1)
m.bias.data.zero_()
elif isinstance(m, nn.Linear):
m.weight.data.normal_(0, 0.01)
m.bias.data.zero_()
def make_layers(cfg, batch_norm=False):
layers = []
in_channels = 3
for v in cfg:
if v == 'M':
layers += [nn.MaxPool2d(kernel_size=2, stride=2)]
else:
conv2d = nn.Conv2d(in_channels, v, kernel_size=3, padding=1)
if batch_norm:
layers += [conv2d, nn.BatchNorm2d(v), nn.ReLU(inplace=True)]
else:
layers += [conv2d, nn.ReLU(inplace=True)]
in_channels = v
return nn.Sequential(*layers)
cfg = {
'A': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],
'B': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],
'D': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],
'E': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M', 512, 512, 512, 512, 'M'],
}
def vgg16(pretrained=False, **kwargs):
"""VGG 16-layer model (configuration "D")
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
if pretrained:
kwargs['init_weights'] = False
model = VGG(make_layers(cfg['D']), **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['vgg16']))
return model