pytorch-mnist神经网络训练

在net.py里面构造网络，网络的结构为输入为28*28,第一层隐藏层的输出为300, 第二层输出的输出为100, 最后一层的输出层为10,

net.py

import torch
from torch import nn
 
class Batch_Net(nn.Module):
    def __init__(self, in_dim, n_hidden_1, n_hidden_2, out_dim):
        super(Batch_Net, self).__init__()
        self.layer_1 = nn.Sequential(nn.Linear(in_dim, n_hidden_1), nn.BatchNorm1d(n_hidden_1), nn.ReLU(True))
        self.layer_2 = nn.Sequential(nn.Linear(n_hidden_1, n_hidden_2), nn.BatchNorm1d(n_hidden_2), nn.ReLU(True))
        self.output = nn.Sequential(nn.Linear(n_hidden_2, out_dim))
 
    def forward(self, x):
        x = self.layer_1(x)
        x = self.layer_2(x)
        x = self.output(x)
        return x

main.py 进行网络的训练

import torch
from torch import nn, optim
from torch.autograd import Variable
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
 
import net
 
batch_size = 128  # 每一个batch_size的大小
learning_rate = 1e-2 # 学习率的大小
num_epoches = 20  # 迭代的epoch值
 # 表示data将数据变成0, 1之间，0.5, 0.5表示减去均值处以标准差
data_tf = transforms.Compose([transforms.ToTensor(), transforms.Normalize([0.5], [0.5])])  # 表示均值和标准差
# 获得训练集的数据
train_dataset = datasets.MNIST(root='./data', train=True, transform=data_tf, download=True)
# 获得测试集的数据
test_dataset = datasets.MNIST(root='./data', train=False, transform=data_tf, download=True)
# 获得训练集的可迭代队列
train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
# 获得测试集的可迭代队列
test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)
# 构造模型的网络
model = net.Batch_Net(28*28, 300, 100, 10)
if torch.cuda.is_available(): # 如果有cuda就将模型放在GPU上
    model.cuda()
 
criterion = nn.CrossEntropyLoss() # 构造交叉损失函数
optimizer = optim.SGD(model.parameters(), lr=learning_rate) # 构造模型的优化器
 
for epoch in range(num_epoches): # 迭代的epoch
    train_loss = 0 # 训练的损失值
    test_loss = 0 # 测试的损失值
    eval_acc = 0 # 测试集的准确率
    for data in train_loader:  # 获得一个batch的样本
        img, label = data # 获得图片和标签
        img = img.view(img.size(0), -1) # 将图片进行img的转换
        if torch.cuda.is_available(): # 如果存在torch
            img = Variable(img).cuda() # 将图片放在torch上
            label = Variable(label).cuda() # 将标签放在torch上
        else:
            img = Variable(img)  # 构造img的变量
            label = Variable(label)
        optimizer.zero_grad() # 消除optimizer的梯度
        out = model.forward(img) # 进行前向传播
        loss = criterion(out, label) # 计算损失值
        loss.backward() # 进行损失值的后向传播
        optimizer.step() # 进行优化器的优化
        train_loss += loss.data #
    for data in test_loader:
        img, label = data
        img = img.view(img.size(0), -1)
        if torch.cuda.is_available():
            img = Variable(img, volatile=True).cuda()
            label = Variable(label, volatile=True).cuda()
        else:
            img = Variable(img, volatile=True)
            label = Variable(label, volatile=True)
        out = model.forward(img)
        loss = criterion(out, label)
        test_loss += loss.data
        top_p, top_class = out.topk(1, dim=1) # 获得输出的每一个样本的最大损失
        equals = top_class == label.view(*top_class.shape) # 判断两组样本的标签是否相等
        accuracy = torch.mean(equals.type(torch.FloatTensor)) # 计算准确率
        eval_acc += accuracy
    print('train_loss{:.6f}, test_loss{:.6f}, Acc:{:.6f}'.format(train_loss / len(train_loader), test_loss / len(test_loader), eval_acc / len(test_loader)))