x = torch.empty(5,3)
print(x)
1.1.3、创建一个随机初始化的5×3张量
x = torch.rand(5, 3)
print(x)
1.1.4、创建一个5×3的0张量,类型为long
x = torch.zeros(5, 3, dtype=torch.long)
print(x)
1.1.5、直接从数组创建张量
x = torch.tensor([5.5, 3])
print(x)
1.1.6、创建一个5×3的单位张量,类型为double
x = torch.ones(5, 3, dtype=torch.double)
print(x)
1.1.7、从已有的张量创建相同维度的新张量,并且重新定义类型为float
x = torch.ones(5, 3)
x = torch.randn_like(x, dtype=torch.float)
print(x)
1.1.8、打印一个张量的维度
x = torch.ones(5, 3)
print(x.size())
1.1.9、两个张量相加
# 方法一
x = torch.ones(5, 3)
y = torch.randn(5, 3)
print(x + y)
# 方法二
# print(torch.add(x, y)
# 方法三
# result = torch.empty(5, 3)
# torch.add(x, y, out=result)
# print(result)
# 方法四
# y.add_(x)
# print(y)
1.1.10、取向量的第一列
x = torch.randn(5, 3)
print(x[:, 1])
1.1.11、将一个4×4的张量resize成一个一维张量
x = torch.randn(4, 4)
y = x.view(16)
print(x.size(), y.size())
1.1.12、将一个4×4的张量,resize成一个2×8的张量
# 方法一
x = torch.randn(4, 4)
y = x.view(2, 8)
print(x.size(), y.size())
# 方法二
# import torch
#
# x = torch.randn(4, 4)
# y = x.view(-1, 8) # 确定一个维度,-1的维度会被自动计算
# print(x.size(), y.size())
1.1.13、从张量中取出数字
x = torch.randn(1)
print(x)
print(x.item())
1.2、Numpy的操作
1.2.1、将张量装换成numpy数组
a = torch.ones(5)
print(a)
b = a.numpy()
print(b)
1.2.2、将张量+1,并观察上提中numpy数组的变化
a = torch.ones(5)
b = a.numpy()
a.add_(1)
print(a)
print(b)
1.2.3、从numpy数组创建张量
a = np.ones(5)
b = torch.from_numpy(a)
print(a)
print(b)
1.2.4、将numpy数组+1并观察上题中张量的变化
a = np.ones(5)
b = torch.from_numpy(a)
np.add(a, 1, out=a)
print(a)
print(b)
二、自动微分
2.1、张量的自动微分
2.1.1、新建一个张量,并设置requires_grad=True
x = torch.ones(2, 2, requires_grad=True)
print(x)
2.1.2、对张量进行任意操作(y = x + 2)
x = torch.ones(2, 2)
y = x + 2
print(y)
print(y.grad_fn) # y就多了一个AddBackward
2.1.3、再对y进行任意操作
x = torch.ones(2, 2)
y = x + 2
print(y)
print(y.grad_fn) # y就多了一个AddBackward
z = y * y * 3
out = z.mean()
print(z) # z多了MulBackward
print(out) # out多了MeanBackward
2.2、梯度
2.2.1、对out进行反向传播
out.backward()
2.2.2、打印梯度
print(x.grad) # out=0.25*Σ3(x+2)^2
2.2.3、创建一个结果为矢量的计算过程()
x = torch.rand(3, requires_grad=True)
y = x * 2
while y.data.norm() < 1000:
y = y * 2
print(y)
2.2.4、计算 v = [0. 1,1.0,0.0001]处的梯度
v = torch.tensor([0.1, 1.0 , 0.0001], dtype=torch.float)
y.backward(v)
print(x.grad)
2.2.5、关闭梯度功能
print(x.requires_grad)
print((x ** 2).requires_grad)
with torch.no_grad():
print((x ** 2).requires_grad)
# 方法二
# print(x.requires_grad)
# y = x.detach()
# print(y.requires_grad)
# print(x.eq(y).all())
三、神经网络
3.1、这部分会实现LeNet5,结构如下所示
3.1.1、定义网络
import torch
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
# 26.定义①的卷积层,输入为32x32的图像,卷积核大小5x5卷积核种类6
self.conv1 = nn.Conv2d(3, 6, 5)
# 27.定义③的卷积层,输入为前一层6个特征,卷积核大小5x5,卷积核种类16
self.conv2 = nn.Conv2d(6, 16, 5)
# 28.定义⑤的全链接层,输入为16*5*5,输出为120
self.fc1 = nn.Linear(16 * 5 * 5, 120) # 6*6 from image dimension
# 29.定义⑥的全连接层,输入为120,输出为84
self.fc2 = nn.Linear(120, 84)
# 30.定义⑥的全连接层,输入为84,输出为10
self.fc3 = nn.Linear(84, 10)
def forward(self, x):
# 31.完成input-S2,先卷积+relu,再2x2下采样
x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
# 32.完成S2-S4,先卷积+relu,再2x2下采样
x = F.max_pool2d(F.relu(self.conv2(x)), 2) # 卷积核方形时,可以只写一个
# 33.将特征向量扁平成列向量
x = x.view(-1, 16 * 5 * 5)
# 34.使用fc1+relu
x = F.relu(self.fc1(x))
# 35.使用fc2+relu
x = F.relu(self.fc2(x))
# 36.使用fc3
x = self.fc3(x)
return x
net = Net()
print(net)
3.1.2、打印网络的参数
params = list(net.parameters())
# print(params)
print(len(params))
3.1.3、打印某一层参数的形状
print(params[0].size())
3.1.4、随机输入一个向量,查看前项传播输出
input = torch.randn(1, 1, 32, 32)
out = net(input)
print(out)
3.1.5、将梯度初始化
net.zero_grad()
3.1.6、随机一个梯度进行反向传播
out.backward(torch.randn(1, 10))
3.2、损失函数
3.2.1、用自带的MSELoss()定义损失函数
criterion = nn.MSELoss()
3.2.2、随机一个真值,并用随机的输入计算损失
target = torch.randn(10) # 随机真值
target = target.view(1, -1) # 变成列向量
output = net(input) # 用随机输入计算输出
loss = criterion(output, target) # 计算损失
print(loss)
3.2.3、将梯度初始化,计算上一步中loss的反向传播
net.zero_grad()
print('conv1.bias.grad before backward')
print(net.conv1.bias.grad)
3.2.4、计算3.2.2中loss的反向传播
loss.backward()
print('conv1.bias.grad after backward')
print(net.conv1.bias.grad)
3.3、更新权重
3.3.1、定义SGD优化器算法,学习率设置为0.01
import torch.optim as optim
optimizer = optim.SGD(net.parameters(), lr=0.01)
3.3.2、使用优化器更新权重
optimizer.zero_grad()
output = net(input)
loss = criterion(output, target)
loss.backward()
# 更新权重
optimizer.step()
3.4、训练一个分类器
3.4.1、读取CIFAR10数据,做标准化
构造一个transform,将三通道(0,1)区间的数据转化成(-1,1)的数据
import torchvision
import torchvision.transforms as transforms
transform = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
读取数据集
trainset = cifar(root = '/home/kesci/input/cifar10', segmentation='train', transforms=transform)
testset = cifar(root = '/home/kesci/input/cifar10', segmentation='test', transforms=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,shuffle=True, num_workers=2)
testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size,shuffle=False, num_workers=2)
classes = ('plane', 'car', 'bird', 'cat',
'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
3.4.2、建立网络
这部分沿用前面的网络
net2 = Net()
3.4.3、定义损失函数和优化器
定义交叉熵损失函数
criterion2 = nn.CrossEntropyLoss()
定义SGD优化器算法,学习率设置为0.001,momentum=0.9
optimizer2 = optim.SGD(net2.parameters(), lr=0.001, momentum=0.9)
3.4.4、训练网络
for epoch in range(2):
running_loss = 0.0
for i, data in enumerate(trainloader, 0):
# 获取X,y对
inputs, labels = data
# 51.初始化梯度
optimizer2.zero_grad()
# 52.前馈
outputs = net2(inputs)
# 53.计算损失
loss = criterion2(outputs, labels)
# 54.计算梯度
loss.backward()
# 55.更新权值
optimizer2.step()
# 每2000个数据打印平均代价函数值
running_loss += loss.item()
if i % 2000 == 1999: # print every 2000 mini-batches
print('[%d, %5d] loss: %.3f' %
(epoch + 1, i + 1, running_loss / 2000))
running_loss = 0.0
print('Finished Training')
3.4.5、使用模型预测
取一些数据
dataiter = iter(testloader)
images, labels = dataiter.next()
# print images
imshow(torchvision.utils.make_grid(images))
print('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4))
使用模型预测
outputs = net2(images)
_, predicted = torch.max(outputs, 1)
print('Predicted: ', ' '.join('%5s' % classes[predicted[j]]
for j in range(4)))
在测试集上进行打分
correct = 0
total = 0
with torch.no_grad():
for data in testloader:
images, labels = data
outputs = net2(images)
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
print('Accuracy of the network on the 10000 test images: %d %%' % ( 100 * correct / total))
3.4.6、存取模型
保存训练好的模型
PATH = './cifar_net.pth'
torch.save(net.state_dict(), PATH)
读取保存的模型
pretrained_net = torch.load(PATH)
加载模型
net3 = Net()
net3.load_state_dict(pretrained_net)