pytorch使用cuda,调用GPU进行训练
Pytorch允许把在GPU上训练的模型加载到CPU上,也允许把在CPU上训练的模型加载到GPU上。
首先需要判断自己的pytorch是否能够使用GPU计算:
print(torch.cuda.is_available())
如果输出False的话,要重新配置cuda环境,这里就不仔细说明了。
然后,明确什么东西可以使用GPU训练,一般来说包括网络模型、数据(输入、标注)、损失函数,主要有以下两种方法:
方法1:使用.to(device)方法
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
net.to(device)
inputs, labels = inputs.to(device), labels.to(device)
loss_fn = loss_fn.to(device)
方法2:使用.cuda()方法
if torch.cuda.is_available():
net.cuda()
if torch.cuda.is_available():
inputs, labels = inputs.cuda(), labels.cuda()
if torch.cuda.is_available():
loss_fn = loss_fn.cuda()
使用方法2调用cuda进行训练的一个案例,方法1同理。
import time
import torch
import torchvision.datasets
from torch import nn
from torch.nn import Sequential, Conv2d, MaxPool2d, Flatten, Linear
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter
# from model import *
# 准备数据集
train_data = torchvision.datasets.CIFAR10("./dataset_cifar10/train", train=True,
transform=torchvision.transforms.ToTensor(),
download=True)
test_data = torchvision.datasets.CIFAR10("./dataset_cifar10/test", train=False,
transform=torchvision.transforms.ToTensor(),
download=True)
# 利用DataLoader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 Lyon(nn.Module):
def __init__(self):
super(Lyon, self).__init__()
self.model1 = Sequential(
Conv2d(3, 32, 5, padding=2),
MaxPool2d(2),
Conv2d(32, 32, 5, padding=2),
MaxPool2d(2),
Conv2d(32, 64, 5, padding=2),
MaxPool2d(2),
Flatten(),
Linear(1024, 64),
Linear(64, 10)
)
def forward(self, x):
x = self.model1(x)
return x
lyon = Lyon()
if torch.cuda.is_available():
lyon = lyon.cuda()
# 损失函数
loss_fn = nn.CrossEntropyLoss()
if torch.cuda.is_available():
loss_fn = loss_fn.cuda()
# 优化器
learning_rate = 0.01
optimizer = torch.optim.SGD(lyon.parameters(), lr=learning_rate)
# 设置训练网络的一些参数
# 记录训练次数
total_train_step = 0
# 记录训练次数
total_test_step = 0
# 训练轮数
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
lyon.train() # 可以不写
for data in train_dataloader:
imgs, targets = data
if torch.cuda.is_available():
imgs = imgs.cuda()
targets = targets.cuda()
output = lyon(imgs)
loss = loss_fn(output, targets)
# 优化器优化模型
optimizer.zero_grad()
loss.backward()
optimizer.step()
total_train_step = total_train_step + 1
if total_train_step % 100 == 0:
end_time = time.time()
print(end_time - start_time)
print("训练次数:{},Loss:{}".format(total_train_step, loss))
writer.add_scalar("train_loss", loss.item(), total_train_step)
# 测试步骤开始
lyon.eval() # 评估步骤开始,可以不写
total_test_loss = 0
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 = lyon(imgs)
loss = loss_fn(outputs, targets)
accuracy = (outputs.argmax(1) == targets).sum() # outputs.argmax(1)将输出结果转换为targets的模式
total_test_loss = total_test_loss + loss.item()
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))
writer.add_scalar("test_loss", total_test_loss, total_test_step)
total_test_step = total_test_step + 1
torch.save(lyon, "./train_model/lyon_{}.pth".format(i))
# 官方推荐模型保存方式
# torch.save(lyon.state_dict(), "./lyon_{}.pth".format(i))
print("模型已保存!")
writer.close()