李沐深度学习记录5：13.Dropout

Dropout从零开始实现

import torch
from torch import nn
from d2l import torch as d2l# 定义Dropout函数
def dropout_layer(X, dropout):assert 0 <= dropout <= 1# 在本情况中，所有元素都被丢弃if dropout == 1:return torch.zeros_like(X)# 在本情况中，所有元素都被保留if dropout == 0:return X#torch.rand生成0-1之间的均匀分布随机数，将其值与dropout概率作比较，得到布尔类型结果由mask存储#布尔类型为0的则为随机丢弃置0的隐藏层单元，留下的则进行值的替换h-->h/(1-p)mask = (torch.rand(X.shape) > dropout).float()return mask * X / (1.0 - dropout)# 测试dropout函数
# X= torch.arange(16, dtype = torch.float32).reshape((2, 8))
# print(X)
# print(dropout_layer(X, 0.))
# print(dropout_layer(X, 0.5))
# print(dropout_layer(X, 1.))#定义模型参数
num_inputs, num_outputs, num_hiddens1, num_hiddens2 = 784, 10, 256, 256#定义模型
dropout1, dropout2 = 0.2, 0.5class Net(nn.Module):  #写一个模型类继承nn.Moduledef __init__(self, num_inputs, num_outputs, num_hiddens1, num_hiddens2,is_training = True):super(Net, self).__init__()self.num_inputs = num_inputsself.training = is_training#定义三个全连接层和激活函数self.lin1 = nn.Linear(num_inputs, num_hiddens1)self.lin2 = nn.Linear(num_hiddens1, num_hiddens2)self.lin3 = nn.Linear(num_hiddens2, num_outputs)self.relu = nn.ReLU()def forward(self, X):H1 = self.relu(self.lin1(X.reshape((-1, self.num_inputs)))) #第一层全连接层加激活函数# 只有在训练模型时才使用dropoutif self.training == True:# 在第一个全连接层之后添加一个dropout层H1 = dropout_layer(H1, dropout1)H2 = self.relu(self.lin2(H1))if self.training == True:# 在第二个全连接层之后添加一个dropout层H2 = dropout_layer(H2, dropout2)out = self.lin3(H2)return outnet = Net(num_inputs, num_outputs, num_hiddens1, num_hiddens2)#训练和测试
num_epochs, lr, batch_size = 10, 0.5, 256
loss = nn.CrossEntropyLoss(reduction='none')
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)
trainer = torch.optim.SGD(net.parameters(), lr=lr)
d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, trainer)

Dropout简洁实现

import torch
from torch import nn
from d2l import torch as d2l#定义模型参数
num_inputs, num_outputs, num_hiddens1, num_hiddens2 = 784, 10, 256, 256#定义模型
dropout1, dropout2 = 0.2, 0.5#定义模型
net=nn.Sequential(nn.Flatten(),nn.Linear(784,256),nn.ReLU(),#第一个全连接层之后添加一个Dropout层nn.Dropout(dropout1),nn.Linear(256,256),nn.ReLU(),#第二个全连接层之后添加一个Dropout层nn.Dropout(dropout2),nn.Linear(256,10))
#参数初始化
def init_weights(m):if type(m)==nn.Linear:nn.init.normal_(m.weight,std=0.01)net.apply(init_weights)

#读取数据
batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)#训练测试
num_epochs,lr=10,0.5
loss = nn.CrossEntropyLoss(reduction='none')
trainer=torch.optim.SGD(net.parameters(),lr=lr)
d2l.train_ch3(net,train_iter,test_iter,loss,num_epochs,trainer)