pytorch-----过拟合(Trick)-交叉验证

俩个问题
1.怎么检测过拟合
2.怎么减少过拟合

如何检测overfitting

train -val-test


import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchvision import datasets,transforms

batch_size = 200
learning_rate = 0.01
epochs = 10

####train = True 是用来分开train与test的
train_db = datasets.MNIST('../data', train=True, download=True,
                          transform=transforms.Compose([
                              transforms.ToTensor(),
                              transforms.Normalize((0.1307,), (0.3081,))
                          ]))
train_loader = torch.utils.data.DataLoader(
    train_db,
    batch_size=batch_size,shuffle = True
)
test_db = datasets.MNIST('../data', train=False, transform=transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize((0.1307,), (0.3081,))
]))
test_loader = torch.utils.data.DataLoader(
    test_db,
    batch_size=batch_size, shuffle=True
)
print('train:', len(train_db), 'test:', len(test_db))
train_db, val_db = torch.utils.data.random_split(train_db, [50000, 10000])
print('db1:', len(train_db), 'db2:', len(val_db))
train_loader = torch.utils.data.DataLoader(
    train_db,
    batch_size=batch_size, shuffle=True)
val_loader = torch.utils.data.DataLoader(
    val_db,
    batch_size=batch_size, shuffle=True)


class MLP(nn.Module):

    def __init__(self):
        super(MLP, self).__init__()

        self.model = nn.Sequential(
            nn.Linear(784, 200),
            nn.LeakyReLU(inplace=True),
            nn.Linear(200, 200),
            nn.LeakyReLU(inplace=True),
            nn.Linear(200, 10),
            nn.LeakyReLU(inplace=True),
        )

    def forward(self, x):
        x = self.model(x)

        return x

device = torch.device('cuda:0')
net = MLP().to(device)
optimizer = optim.SGD(net.parameters(),lr = learning_rate)
criteon = nn.CrossEntropyLoss().to(device)

for epoch in range(epochs):

    for batch_idx, (data, target) in enumerate(train_loader):
        data = data.view(-1, 28 * 28)
        data, target = data.to(device), target.cuda()

        logits = net(data)
        loss = criteon(logits, target)

        optimizer.zero_grad()
        loss.backward()
        # print(w1.grad.norm(), w2.grad.norm())
        optimizer.step()

        if batch_idx % 100 == 0:
            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                       100. * batch_idx / len(train_loader), loss.item()))

    test_loss = 0
    correct = 0
    for data, target in val_loader:
        data = data.view(-1, 28 * 28)
        data, target = data.to(device), target.cuda()
        logits = net(data)
        test_loss += criteon(logits, target).item()

        pred = logits.data.max(1)[1]
        correct += pred.eq(target.data).sum()

    test_loss /= len(val_loader.dataset)
    print('\nVAL set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(val_loader.dataset),
        100. * correct / len(val_loader.dataset)))

###这里选保存为过拟合处的模型参数,设置,然后对惊醒test


test_loss = 0
correct = 0
for data, target in test_loader:
    data = data.view(-1, 28 * 28)
    data, target = data.to(device), target.cuda()
    logits = net(data)
    test_loss += criteon(logits, target).item()

    pred = logits.data.max(1)[1]
    correct += pred.eq(target.data).sum()

test_loss /= len(test_loader.dataset)
print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
    test_loss, correct, len(test_loader.dataset),
    100. * correct / len(test_loader.dataset)))

 

评论 1
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值