一、1.模型搭建
import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as F # 用Relu函数
import torch.optim as optim # 优化器优化
batch_size = 64
# transform预处理,把图像转化成图像张量
'''ToTensor:将一个’PIL Image‘or‘numpy.ndarray’转化为tensor格式,PIL Image or numpy.ndarray的shape为(H x W x C),范围是[0, 255]
转化为shape为(C x H x W)范围在[0.0, 1.0]'''
'''Normalize:标准化函数,使用均值和标准差对tensor进行标准化
output[channel] = (input[channel] - mean[channel]) / std[channel]
'''
transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])
# mnist数据集为1*28*28的单通道图像
# ./表示当前目录 ../表示父级目录 /表示根目录
train_dataset = datasets.MNIST(root='../dataset/mnist',
train=True,
download=True,
transform=transform) # 训练数据集
train_loader = DataLoader(train_dataset,
shuffle=True,
batch_size=batch_size)
test_dataset = datasets.MNIST(root='../dataset/mnist',
train=False,
download=True,
transform=transform)
test_loader = DataLoader(test_dataset,
shuffle=False,
batch_size=batch_size)
class Net(torch.nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = torch.nn.Conv2d(1, 10, kernel_size=5)
self.conv2 = torch.nn.Conv2d(10, 20, kernel_size=5)
self.pooling = torch.nn.MaxPool2d(2)
self.fc = torch.nn.Linear(320, 10)
def forward(self, x):
x = self.pooling(F.relu(self.conv1(x)))
x = self.pooling(F.relu(self.conv2(x)))
# 此时x是维度为4的tensor,即(batchsize,C,H,W),x.size(0)指batchsize的值
# Flatten层用来将输入“压平”,即把多维的输入一维化,常用在从卷积层到全连接层的过渡。
x = x.view(x.size(0), -1) # -1是自动计算,为C*H*W
x = self.fc(x) # x应用全连接网络
return x
model = Net()
#将模型迁移到GPU上
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model.to(device)
criterion = torch.nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)
# 因为网络模型已经有点大了,所以梯度下降里面要用更好的优化算法,比如用带冲量的(momentum),来优化训练过程
# 把一轮循环封装到函数里面
def train(epoch):
running_loss = 0.0
# 通过函数enumerate返回每一批数据data,以及索引index(batch_idx),因为start=0所以index从0开始
for batch_idx, data in enumerate(train_loader, 0):
inputs, targets = data # 将数据分为图像以及所对应的标签
inputs, targets = inputs.to(device), targets.to(device) # 迁移到GPU上
optimizer.zero_grad() # 将历史损失梯度清零
# forward
y_pred = model(inputs) # 将训练图片输入网络得到输出
# backward
loss = criterion(y_pred, targets)
loss.backward()
# update
optimizer.step() # 参数更新
running_loss += loss.item()
if batch_idx % 300 == 299: # 每300个mini-batches打印一次
print('[%d,%5d] loss:%.3f' % (epoch + 1, batch_idx + 1, running_loss / 300))
running_loss = 0.0
def test():
correct = 0 # 正确多少
total = 0 # 总数多少
with torch.no_grad(): # 测试不用算梯度
for data in test_loader: # 从test_loader拿数据
images, labels = data #将数据分为图像以及所对应的标签
images, labels = images.to(device), labels.to(device) #迁移到GPU上
outputs = model(images) # 拿完数据做预测
_, predicted = torch.max(outputs.data, dim=1) # 沿着第一个维度找最大值的下标,返回值有两个,因为是10列嘛,返回值
# 返回值一个是每一行的最大值,另一个是最大值的下标(每一个样本就是一行,每一行有10个量)(行是第0个维度,列是第1个维度)
total += labels.size(0) # 取size元组的第0个元素(N,1),累加后可得测试样本的总数
# 推测出来的分类与label是否相等,真就是1,假就是0,求完和之后把标量拿出来,循环完成后可得预测正确的样本数
correct += (predicted == labels).sum().item()
print("accuracy on test set:%d %% [%d/%d]" % (100 * correct / total, correct, total))
# 训练
if __name__ == '__main__':
for epoch in range(10):
train(epoch)
test() # 训练一轮,测试一轮
2.结果展示
二、1.模型搭建(加深)
import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as F # 用Relu函数
import torch.optim as optim # 优化器优化
batch_size = 64
# transform预处理,把图像转化成图像张量
'''ToTensor:将一个’PIL Image‘or‘numpy.ndarray’转化为tensor格式,PIL Image or numpy.ndarray的shape为(H x W x C),范围是[0, 255]
转化为shape为(C x H x W)范围在[0.0, 1.0]'''
'''Normalize:标准化函数,使用均值和标准差对tensor进行标准化
output[channel] = (input[channel] - mean[channel]) / std[channel]
'''
transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])
# mnist数据集为1*28*28的单通道图像
# ./表示当前目录 ../表示父级目录 /表示根目录
train_dataset = datasets.MNIST(root='../dataset/mnist',
train=True,
download=True,
transform=transform) # 训练数据集
train_loader = DataLoader(train_dataset,
shuffle=True,
batch_size=batch_size)
test_dataset = datasets.MNIST(root='../dataset/mnist',
train=False,
download=True,
transform=transform)
test_loader = DataLoader(test_dataset,
shuffle=False,
batch_size=batch_size)
class Net(torch.nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = torch.nn.Conv2d(1, 10, kernel_size=3)
self.conv2 = torch.nn.Conv2d(10, 20, kernel_size=3, padding=1)
self.conv3 = torch.nn.Conv2d(20, 32, kernel_size=3)
self.pooling = torch.nn.MaxPool2d(2)
self.relu = torch.nn.ReLU()
self.fc1 = torch.nn.Linear(128, 64)
self.fc2 = torch.nn.Linear(64, 32)
self.fc3 = torch.nn.Linear(32, 10)
def forward(self, x):
# batch_size = x.size(0)
x = self.pooling(self.relu(self.conv1(x)))
x = self.pooling(self.relu(self.conv2(x)))
x = self.pooling(self.relu(self.conv3(x)))
# 此时x是维度为4的tensor,即(batchsize,C,H,W),x.size(0)指batchsize的值
# Flatten层用来将输入“压平”,即把多维的输入一维化,常用在从卷积层到全连接层的过渡。
x = x.view(x.size(0), -1)
x = self.fc1(x) # x应用全连接网络
x = self.fc2(x)
x = self.fc3(x)
return x
model = Net()
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model.to(device)
criterion = torch.nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)
# 因为网络模型已经有点大了,所以梯度下降里面要用更好的优化算法,比如用带冲量的(momentum),来优化训练过程
# 把一轮循环封装到函数里面
def train(epoch):
running_loss = 0.0
# 通过函数enumerate返回每一批数据data,以及索引index(batch_idx),因为start=0所以index从0开始
for batch_idx, data in enumerate(train_loader, 0):
inputs, targets = data # 将数据分为图像以及所对应的标签
inputs, targets = inputs.to(device), targets.to(device) # 迁移到GPU上
optimizer.zero_grad() # 将历史损失梯度清零
# forward
y_pred = model(inputs) # 将训练图片输入网络得到输出
# backward
loss = criterion(y_pred, targets)
loss.backward()
# update
optimizer.step() # 参数更新
running_loss += loss.item()
if batch_idx % 300 == 299: # 每300个mini-batches打印一次
print('[%d,%5d] loss:%.3f' % (epoch + 1, batch_idx + 1, running_loss / 300))
running_loss = 0.0
def test():
correct = 0 # 正确多少
total = 0 # 总数多少
with torch.no_grad(): # 测试不用算梯度
for data in test_loader: # 从test_loader拿数据
images, labels = data #将数据分为图像以及所对应的标签
images, labels = images.to(device), labels.to(device) #迁移到GPU上
outputs = model(images) # 拿完数据做预测
_, predicted = torch.max(outputs.data, dim=1) # 沿着第一个维度找最大值的下标,返回值有两个,因为是10列嘛,返回值
# 返回值一个是每一行的最大值,另一个是最大值的下标(每一个样本就是一行,每一行有10个量)(行是第0个维度,列是第1个维度)
total += labels.size(0) # 取size元组的第0个元素(N,1),累加后可得测试样本的总数
# 推测出来的分类与label是否相等,真就是1,假就是0,求完和之后把标量拿出来,循环完成后可得预测正确的样本数
correct += (predicted == labels).sum().item()
print("accuracy on test set:%d %% [%d/%d]" % (100 * correct / total, correct, total))
# 训练
if __name__ == '__main__':
for epoch in range(10):
train(epoch)
test() # 训练一轮,测试一轮
2.结果展示
reference
《PyTorch深度学习实践》完结合集_哔哩哔哩_bilibili