import torch import torch.nn as nn import torch.nn.functional as F import torch.optim as optim from torchvision import datasets, transforms print("PyTorch Version: ",torch.__version__)
PyTorch Version: 1.0.0
首先我们定义一个基于ConvNet的简单神经网络
In [4]:
class Net(nn.Module): def __init__(self): super(Net, self).__init__() self.conv1 = nn.Conv2d(1, 20, 5, 1) self.conv2 = nn.Conv2d(20, 50, 5, 1) self.fc1 = nn.Linear(4*4*50, 500) self.fc2 = nn.Linear(500, 10) def forward(self, x): x = F.relu(self.conv1(x)) x = F.max_pool2d(x, 2, 2) x = F.relu(self.conv2(x)) x = F.max_pool2d(x, 2, 2) x = x.view(-1, 4*4*50) x = F.relu(self.fc1(x)) x = self.fc2(x) return F.log_softmax(x, dim=1)
NLL loss的定义
ℓ(x,y)=L={l1,…,lN}⊤,ln=−wynxn,yn,wc=weight[c]⋅1{c≠ignore\_index}ℓ(x,y)=L={l1,…,lN}⊤,ln=−wynxn,yn,wc=weight[c]⋅1{c≠ignore\_index}
In [7]:
def train(model, device, train_loader, optimizer, epoch, log_interval=100): model.train() for batch_idx, (data, target) in enumerate(train_loader): data, target = data.to(device), target.to(device) optimizer.zero_grad() output = model(data) loss = F.nll_loss(output, target) loss.backward() optimizer.step() if batch_idx % log_interval == 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() ))
In [8]:
def test(model, device, test_loader): model.eval() test_loss = 0 correct = 0 with torch.no_grad(): for data, target in test_loader: data, target = data.to(device), target.to(device) output = model(data) test_loss += F.nll_loss(output, target, reduction='sum').item() # sum up batch loss pred = output.argmax(dim=1, keepdim=True) # get the index of the max log-probability correct += pred.eq(target.view_as(pred)).sum().item() 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)))
In [13]:
torch.manual_seed(53113) use_cuda = torch.cuda.is_available() device = torch.device("cuda" if use_cuda else "cpu") batch_size = test_batch_size = 32 kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {} train_loader = torch.utils.data.DataLoader( datasets.MNIST('./mnist_data', train=True, download=True, transform=transforms.Compose([ transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,)) ])), batch_size=batch_size, shuffle=True, **kwargs) test_loader = torch.utils.data.DataLoader( datasets.MNIST('./mnist_data', train=False, transform=transforms.Compose([ transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,)) ])), batch_size=test_batch_size, shuffle=True, **kwargs) lr = 0.01 momentum = 0.5 model = Net().to(device) optimizer = optim.SGD(model.parameters(), lr=lr, momentum=momentum) epochs = 2 for epoch in range(1, epochs + 1): train(model, device, train_loader, optimizer, epoch) test(model, device, test_loader) save_model = True if (save_model): torch.save(model.state_dict(),"mnist_cnn.pt")
Train Epoch: 1 [0/60000 (0.000000%)] Loss: 2.297938 Train Epoch: 1 [3200/60000 (5.333333%)] Loss: 0.567845 Train Epoch: 1 [6400/60000 (10.666667%)] Loss: 0.206370 Train Epoch: 1 [9600/60000 (16.000000%)] Loss: 0.094653 Train Epoch: 1 [12800/60000 (21.333333%)] Loss: 0.180530 Train Epoch: 1 [16000/60000 (26.666667%)] Loss: 0.041645 Train Epoch: 1 [19200/60000 (32.000000%)] Loss: 0.135092 Train Epoch: 1 [22400/60000 (37.333333%)] Loss: 0.054001 Train Epoch: 1 [25600/60000 (42.666667%)] Loss: 0.111863 Train Epoch: 1 [28800/60000 (48.000000%)] Loss: 0.059039 Train Epoch: 1 [32000/60000 (53.333333%)] Loss: 0.089227 Train Epoch: 1 [35200/60000 (58.666667%)] Loss: 0.186015 Train Epoch: 1 [38400/60000 (64.000000%)] Loss: 0.093208 Train Epoch: 1 [41600/60000 (69.333333%)] Loss: 0.077090 Train Epoch: 1 [44800/60000 (74.666667%)] Loss: 0.038075 Train Epoch: 1 [48000/60000 (80.000000%)] Loss: 0.036247 Train Epoch: 1 [51200/60000 (85.333333%)] Loss: 0.052358 Train Epoch: 1 [54400/60000 (90.666667%)] Loss: 0.013201 Train Epoch: 1 [57600/60000 (96.000000%)] Loss: 0.036660 Test set: Average loss: 0.0644, Accuracy: 9802/10000 (98%) Train Epoch: 2 [0/60000 (0.000000%)] Loss: 0.054402 Train Epoch: 2 [3200/60000 (5.333333%)] Loss: 0.032239 Train Epoch: 2 [6400/60000 (10.666667%)] Loss: 0.092350 Train Epoch: 2 [9600/60000 (16.000000%)] Loss: 0.058544 Train Epoch: 2 [12800/60000 (21.333333%)] Loss: 0.029762 Train Epoch: 2 [16000/60000 (26.666667%)] Loss: 0.012521 Train Epoch: 2 [19200/60000 (32.000000%)] Loss: 0.101891 Train Epoch: 2 [22400/60000 (37.333333%)] Loss: 0.127773 Train Epoch: 2 [25600/60000 (42.666667%)] Loss: 0.009259 Train Epoch: 2 [28800/60000 (48.000000%)] Loss: 0.013482 Train Epoch: 2 [32000/60000 (53.333333%)] Loss: 0.039676 Train Epoch: 2 [35200/60000 (58.666667%)] Loss: 0.016707 Train Epoch: 2 [38400/60000 (64.000000%)] Loss: 0.168691 Train Epoch: 2 [41600/60000 (69.333333%)] Loss: 0.056318 Train Epoch: 2 [44800/60000 (74.666667%)] Loss: 0.008174 Train Epoch: 2 [48000/60000 (80.000000%)] Loss: 0.075149 Train Epoch: 2 [51200/60000 (85.333333%)]