MNIST训练

数据下载&训练代码

### mnistGPU.py

import torch
from torch import nn
import torchvision
from torch.utils.data import DataLoader
from torch.autograd import Variable
import time

data_tf = torchvision.transforms.Compose(
                [torchvision.transforms.ToTensor(),
                 torchvision.transforms.Normalize([0.5],[0.5])])

#准备训练数据
train_set = torchvision.datasets.MNIST('./data', train=True,  transform=data_tf, download=True)
#准备测试数据
test_set = torchvision.datasets.MNIST('./data', train=False,  transform=data_tf, download=True)
batchSize = 64
train_data = DataLoader(train_set, batch_size=batchSize, shuffle=True) # 64
test_data = DataLoader(test_set, batch_size=128, shuffle=False) # 128

#示例网络1
class fc_net_2layer(nn.Module):
    def __init__(self):
        super(fc_net_2layer, self).__init__()
        self.fc = nn.Sequential(
                    nn.Linear(28 * 28, 10),
                    nn.ReLU(),
                    nn.Linear(10, 10) #最后输出10个分类
                )
    def forward(self, x):
        x = x.view(x.size(0), -1)
        x = self.fc(x)

        return x

#示例网络2
class fc_net_4layer(nn.Module):
    def __init__(self):
        super(fc_net_4layer, self).__init__()
        self.fc = nn.Sequential(
                    nn.Linear(28 * 28, 400),
                    nn.ReLU(),
                    nn.Linear(400, 200),
                    nn.ReLU(),
                    nn.Linear(200, 100),
                    nn.ReLU(),
                    nn.Linear(100, 10) #最后输出10个分类
                )
    def forward(self, x):
        x = x.view(x.size(0), -1)
        x = self.fc(x)

        return x

#示例网络3
class CNN(nn.Module):
    def __init__(self):
        super(CNN,self).__init__()

        self.layer1 = nn.Sequential(
                nn.Conv2d(1,16,kernel_size=3), # 16, 26 ,26
                nn.BatchNorm2d(16),
                nn.ReLU(inplace=True))

        self.layer2 = nn.Sequential(
                nn.Conv2d(16,32,kernel_size=3),# 32, 24, 24
                nn.BatchNorm2d(32),
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=2,stride=2)) # 32, 12,12     (24-2) /2 +1

        self.layer3 = nn.Sequential(
                nn.Conv2d(32,64,kernel_size=3), # 64,10,10
                nn.BatchNorm2d(64),
                nn.ReLU(inplace=True))

        self.layer4 = nn.Sequential(
                nn.Conv2d(64,128,kernel_size=3),  # 128,8,8
                nn.BatchNorm2d(128),
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=2,stride=2))  # 128, 4,4

        self.fc = nn.Sequential(
                nn.Linear(128 * 4 * 4,1024),
                nn.ReLU(inplace=True),
                nn.Linear(1024,128),
                nn.ReLU(inplace=True),
                nn.Linear(128,10))

    def forward(self,x):
        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)
        x = x.view(x.size(0),-1)
        x = self.fc(x)

        return x

#网络参数数量
def get_parameter_number(net):
    total_num = sum(p.numel() for p in net.parameters())
    trainable_num = sum(p.numel() for p in net.parameters() if p.requires_grad)
    return {'Total': total_num, 'Trainable': trainable_num}
#选择示例网络
# net = fc_net_2layer()
# net = fc_net_4layer()
net = CNN()
print(net)
use_gpu = torch.cuda.is_available()
print(get_parameter_number(net))
#设置损失函数
criterion = nn.CrossEntropyLoss()
#设置网络优化方式
learingrate = 0.01
optimizer = torch.optim.SGD(net.parameters(), lr=learingrate) #学习率0.1 0.01

if(use_gpu):
    net = net.cuda()
    criterion = criterion.cuda()

losses = []
acces = []
eval_losses = []
eval_acces = []
trainloss = []
start = time.time()
#开始训练
for e in range(20):
    train_loss = 0
    train_acc = 0
    net.train()
    for im, label in train_data:
        if (use_gpu):
            im, label = im.cuda(),label.cuda()
        im = Variable(im)
        label = Variable(label)
        # 前向传播
        out = net(im)
        loss = criterion(out, label)
        # 反向传播
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
        # 记录误差
        train_loss += loss.item()
        # 计算分类的准确率
        _, pred = out.max(1)
        num_correct = (pred == label).sum().item()
        acc = num_correct / im.shape[0]
        train_acc += acc
        #print('epoch: {}, Batch Train Loss: {:.6f}, Bacth Train Acc: {:.6f}'.format(e, loss.item(), acc))

    losses.append(train_loss / len(train_data))
    acces.append(train_acc / len(train_data))
    # 在测试集上检验效果
    eval_loss = 0
    eval_acc = 0
    net.eval() # 将模型改为预测模式
    for im, label in test_data:
        if (use_gpu):
            im, label = im.cuda(),label.cuda()
        im = Variable(im)
        label = Variable(label)
        out = net(im)
        loss = criterion(out, label)
        # 记录误差
        eval_loss += loss.item()
        # 记录准确率
        _, pred = out.max(1)
        num_correct = (pred == label).sum().item()
        acc = num_correct / im.shape[0]
        eval_acc += acc
        #print('epoch: {}, Batch Evaluate Loss: {:.6f}, Bacth Evaluate Acc: {:.6f}'.format(e, loss.item(), acc))

    eval_losses.append(eval_loss / len(test_data))
    eval_acces.append(eval_acc / len(test_data))
    print('***** One epoch has finished ******')
    print('epoch: {}, Train Loss: {:.6f}, Train Acc: {:.6f}, Eval Loss: {:.6f}, Eval Acc: {:.6f}'
          .format(e, train_loss / len(train_data), train_acc / len(train_data),
                     eval_loss / len(test_data), eval_acc / len(test_data)))
    trainloss.append(train_loss / len(train_data))
end = time.time()
print('The code run{:.2f}s'.format(end - start))

filename = 'epochs.txt'
with open(filename, 'a') as name:
    name.write('CNN,')
    for i in range(len(trainloss)-1):
        name.write(str(trainloss[i])+',')
    name.write(str(trainloss[-1]) + '\n')
name.close()

运行：

python mnistGPU.py

结果

1. CNN结构

2. fc_net_4layer

3. fc_net_2layer

4. 三种网络对比

net	总para量	runtime	last epoch tranloss	acc	eval loss	eval acc
CNN	2328298	606.21s	0.002778	0.999883	0.020380	0.99374
fc_net_4layer	415310	263.78s	0.069752	0.979761	0.100663	0.967168
fc_net_2layer	7960	265.27s	0.254767	0.927056	0.260187	0.923853

遇到的问题

1. conda install tensorflow 报错

解决： 用pip install。。。。

ps：查看tensowflow版本：

import tensorflow as tf
print(tf.__version__)

2. unindent does not match any outer indentation level
解决：缩进问题。。没对齐