torch.cuda:
该包增加了对CUDA张量类型的支持,实现了与CPU张量相同的功能,但使用GPU进行计算。
它是延迟的初始化,所以你可以随时导入它,并使用is_available()来确定系统是否支持CUDA。
使用GPU训练:
import torch
import torchvision
from torch.utils.tensorboard import SummaryWriter
from torch import nn
from torch.utils.data import DataLoader
#准备数据集
train_data = torchvision.datasets.CIFAR10(root="../data", train=True, transform=torchvision.transforms.ToTensor(), download=True)
test_data = torchvision.datasets.CIFAR10(root="../data", train=False, transform=torchvision.transforms.ToTensor(), download=True)
#length 长度
train_data_size = len(train_data)
test_data_size = len(test_data)
#如果train_data_size=10,训练数据集的长度为:10
print("训练数据集的长度为:{}".format(train_data_size))
print("测试数据集的长度为:{}".format(test_data_size))
#利用DataLoader来加载数据集
train_dataloader = DataLoader(train_data, batch_size=64)
test_dataloader = DataLoader(test_data, batch_size=64)
#创建网络模型
class Test(nn.Module):
def __init__(self):
super(Test, self).__init__()
self.model = nn.Sequential(
nn.Conv2d(3, 32, 5, 1, 2),
nn.MaxPool2d(2),
nn.Conv2d(32, 32, 5, 1, 2),
nn.MaxPool2d(2),
nn.Conv2d(32, 64, 5, 1, 2),
nn.MaxPool2d(2),
nn.Flatten(),
nn.Linear(64*4*4, 64),
nn.Linear(64, 10)
)
def forward(self, x):
x = self.model(x)
return x
test = Test()
#将网络模型转入cuda
if torch.cuda.is_available():
test = test.cuda()
#损失函数
loss_fn = nn.CrossEntropyLoss()
loss_fn = loss_fn.cuda()
#优化器
#le-2 = 1×(10)(-2) = 1 / 100 =0.01
learning_rate = 1e-2
optimizer = torch.optim.SGD(test.parameters(), lr=learning_rate)
#设置训练网络的一些参数
#记录训练的次数
total_train_step = 0
#记录测试的次数
total_test_step = 0
#训练的轮数
epoch = 10
#添加tensorboard
writer = SummaryWriter("./logs_train")
for i in range(epoch):
print("------第{}轮训练开始------".format(i+1))
#训练步骤开始
#调用51行,使网络进入训练状态
test.train()
for data in train_dataloader:
imgs, targets = data
if torch.cuda.is_available():
imgs = imgs.cuda()
targets = targets.cuda()
outputs = test(imgs)
#获得损失值
loss = loss_fn(outputs, targets)
#优化器优化模型
#利用优化器将梯度清0
optimizer.zero_grad()
#利用反向传播得到每一个参数结点的梯度
loss.backward()
#对参数进行优化
optimizer.step()
#记录训练次数
total_train_step = total_train_step + 1
if total_train_step % 100 == 0:
#item()把tensor数据类型转化为真实的数字
print("训练次数:{},Loss:{}".format(total_train_step, loss.item()))
writer.add_scalar("train_loss", loss.item(), total_train_step)
#测试步骤开始
total_test_loss = 0
#计算正确率
total_accuracy = 0
with torch.no_grad():
for data in test_dataloader:
imgs, targets = data
if torch.cuda.is_available():
imgs = imgs.cuda()
targets = targets.cuda()
outputs = test(imgs)
loss = loss_fn(outputs, targets)
total_test_loss = total_test_loss + loss
accuracy = (outputs.argmax(1) == targets).sum()
total_accuracy = total_accuracy + accuracy
print("整体测试集上的Loss:{}".format(total_test_loss))
print("整体测试集上的正确率:{}".format(total_accuracy/test_data_size))
writer.add_scalar("test_loss", total_test_loss, total_test_step)
writer.add_scalar("test_accuracy", total_accuracy/test_data_size, total_test_step)
total_test_step = total_test_step + 1
#torch.save(test.state_dict(), "test_{}.pth".format(i))
torch.save(test, "test_{}.pth".format(i))
print("模型已保存")
writer.close()
如果想对比GPU和CPU训练的速度
可以在开始训练前和每次训练结束时使用time函数获取时间
#添加tensorboard
writer = SummaryWriter("./logs_train")
start_time = time.time()
#记录训练次数
total_train_step = total_train_step + 1
if total_train_step % 100 == 0:
end_time = time.time()
print(end_time - start_time)
#item()把tensor数据类型转化为真实的数字
print("训练次数:{},Loss:{}".format(total_train_step, loss.item()))
writer.add_scalar("train_loss", loss.item(), total_train_step)
结果为GPU处理速度比CPU快将近十倍
对于没有GPU的用户来说,可以使用google colab体验免费的GPU
在修改中选择硬件加速,选择GPU
torch.device
torch.device的一个实例是一个对象,该对象代表的是张量torch.Tensor所在
的设备或将要分配到的设备.
torch.device包含了一个设备类型(‘cpu’ 或者 ‘cuda’),以及该设备类型的可选设备序数.如果该设备序数不存在,那么这个对象所代表的将总是该设备类型的当前设备,尽管已经调用了torch.cuda.set_device()函数;例如,使用设备’cuda’来创建一个torch.Tensor 对象等效于使用设备’cuda:X’来创建torch.Tensor对象,其中X是函数torch.cuda.current_device()的返回结果.
torch.device(‘cuda’) 与 torch.device(‘cuda:0’)在进行计算时,对于单卡计算机而言,没有任何区别,都是唯一的那一张GPU。
代码实战:
import time
import torch
import torchvision
from torch.utils.tensorboard import SummaryWriter
# from model import *
from torch import nn
from torch.utils.data import DataLoader
#定义训练的设备
# device = torch.device("cpu")
# device = torch.device("cuda")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print(device)
#准备数据集
train_data = torchvision.datasets.CIFAR10(root="../data", train=True, transform=torchvision.transforms.ToTensor(), download=True)
test_data = torchvision.datasets.CIFAR10(root="../data", train=False, transform=torchvision.transforms.ToTensor(), download=True)
#length 长度
train_data_size = len(train_data)
test_data_size = len(test_data)
#如果train_data_size=10,训练数据集的长度为:10
print("训练数据集的长度为:{}".format(train_data_size))
print("测试数据集的长度为:{}".format(test_data_size))
#利用DataLoader来加载数据集
train_dataloader = DataLoader(train_data, batch_size=64)
test_dataloader = DataLoader(test_data, batch_size=64)
#创建网络模型
class Test(nn.Module):
def __init__(self):
super(Test, self).__init__()
self.model = nn.Sequential(
nn.Conv2d(3, 32, 5, 1, 2),
nn.MaxPool2d(2),
nn.Conv2d(32, 32, 5, 1, 2),
nn.MaxPool2d(2),
nn.Conv2d(32, 64, 5, 1, 2),
nn.MaxPool2d(2),
nn.Flatten(),
nn.Linear(64*4*4, 64),
nn.Linear(64, 10)
)
def forward(self, x):
x = self.model(x)
return x
test = Test()
#将网络模型转入device
test = test.to(device)
#损失函数
loss_fn = nn.CrossEntropyLoss()
loss_fn = loss_fn.to(device)
#优化器
#le-2 = 1×(10)(-2) = 1 / 100 =0.01
learning_rate = 1e-2
optimizer = torch.optim.SGD(test.parameters(), lr=learning_rate)
#设置训练网络的一些参数
#记录训练的次数
total_train_step = 0
#记录测试的次数
total_test_step = 0
#训练的轮数
epoch = 10
#添加tensorboard
writer = SummaryWriter("./logs_train")
start_time = time.time()
for i in range(epoch):
print("------第{}轮训练开始------".format(i+1))
#训练步骤开始
#调用51行,使网络进入训练状态
test.train()
for data in train_dataloader:
imgs, targets = data
if torch.cuda.is_available():
imgs = imgs.to(device)
targets = targets.to(device)
outputs = test(imgs)
#获得损失值
loss = loss_fn(outputs, targets)
#优化器优化模型
#利用优化器将梯度清0
optimizer.zero_grad()
#利用反向传播得到每一个参数结点的梯度
loss.backward()
#对参数进行优化
optimizer.step()
#记录训练次数
total_train_step = total_train_step + 1
if total_train_step % 100 == 0:
end_time = time.time()
print(end_time - start_time)
#item()把tensor数据类型转化为真实的数字
print("训练次数:{},Loss:{}".format(total_train_step, loss.item()))
writer.add_scalar("train_loss", loss.item(), total_train_step)
#测试步骤开始
total_test_loss = 0
#计算正确率
total_accuracy = 0
with torch.no_grad():
for data in test_dataloader:
imgs, targets = data
if torch.cuda.is_available():
imgs = imgs.to(device)
targets = targets.to(device)
outputs = test(imgs)
loss = loss_fn(outputs, targets)
total_test_loss = total_test_loss + loss
accuracy = (outputs.argmax(1) == targets).sum()
total_accuracy = total_accuracy + accuracy
print("整体测试集上的Loss:{}".format(total_test_loss))
print("整体测试集上的正确率:{}".format(total_accuracy/test_data_size))
writer.add_scalar("test_loss", total_test_loss, total_test_step)
writer.add_scalar("test_accuracy", total_accuracy/test_data_size, total_test_step)
total_test_step = total_test_step + 1
#torch.save(test.state_dict(), "test_{}.pth".format(i))
torch.save(test, "test_{}.pth".format(i))
print("模型已保存")
writer.close()