trainer.py

from torch.autograd import Variable
import torch.nn.functional as F
import torch


def train(epoch,model,optimizer,train_loader,scheduler=None,cuda = True,log_interval = 100):
    model.train()
    avg_loss = 0.
    train_acc = 0.
    for batch_idx, (data, target) in enumerate(train_loader):
        if cuda:
            data, target = data.cuda(), target.cuda()
        data, target = Variable(data), Variable(target)
        optimizer.zero_grad()
        output = model(data)
        loss = F.cross_entropy(output, target)
        avg_loss += loss.item()
        pred = output.data.max(1, keepdim=True)[1]
        train_acc += pred.eq(target.data.view_as(pred)).cpu().sum()
        loss.backward()
        optimizer.step()
        if batch_idx % log_interval == 0:
            print('Train Epoch: {} [{}/{} ({:.1f}%)]\tLoss: {:.6f}'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                100. * batch_idx / len(train_loader), loss.item()))
    if scheduler:
        scheduler.step(avg_loss)


def test(model,test_loader,cuda=True):
    model.eval()
    correct = 0
    with torch.no_grad():
        for data, target in test_loader:
            if cuda:
                data, target = data.cuda(), target.cuda()
            data, target = Variable(data), Variable(target)
            output = model(data)
            pred = output.data.max(1, keepdim=True)[1] # get the index of the max log-probability
            correct += pred.eq(target.data.view_as(pred)).cpu().sum()

        print('\nTest set: Accuracy: {}/{} ({:.1f}%)\n'.format(
            correct, len(test_loader.dataset), 100. * correct / len(test_loader.dataset)))
        return correct / float(len(test_loader.dataset))