def val(data_loader, net, criterion):
losses = tools.AverageMeter()
acc = tools.AverageMeter()
net.eval()
loop = len(data_loader)
for i, (images, labels) in enumerate(data_loader):
images = Variable(images.view(-1, 14951)).cuda()
labels = Variable(labels).cuda()
outputs = net(images)
loss = criterion(outputs, labels)
prec1, prec1 = tools.Accuracy(outputs, labels, topk=(1, 1))
losses.update(loss.data[0], images.size(0))
acc.update(prec1[0], images.size(0))
#if (i + 1) % 10 == 0:
print('Step [%d], Loss: %.4f'
% (i + 1, loss.data[0]))
print('Accuracy : ', (prec1.cpu().data.numpy()[0]))
print('Accuracy (All): ', acc.avg.cpu().data.numpy()[0])
print('Losses (All): ', losses.avg)
print()
return acc.avg
def train(train_loader, model, criterion, optimizer, epoch):
batch_time = tools.AverageMeter()
losses = tools.AverageMeter()
acc = tools.AverageMeter()
end = time.time()
for i, (img, label, og_label, key, linenumber) in enumerate(train_loader):
target = label.cuda(async=True)
img = Variable(img).cuda()
label = Variable(label).cuda()
output = model(img)
loss = criterion(output,label)
optimizer.zero_grad()
loss.backward()
optimizer.step()
# measure accuracy and record loss
prec1, prec1 = tools.Accuracy(output.data, target, topk=(1, 1))
losses.update(loss.data[0], img.size(0))
acc.update(prec1[0], img.size(0))
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if i % 50 == 0:
print('\tEpoch: [{0}][{1}/{2}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
'Accuracy {acc.val:.3f} ({acc.avg:.3f})'.format(
epoch, i, len(train_loader), batch_time=batch_time, loss=losses, acc=acc))
return acc.avg
def val(data_loader, net, criterion, scheduler, epoch):
net.train(False)
losses = tools.AverageMeter()
acc = tools.AverageMeter()
net.eval()
loop = len(data_loader)
for i, (images, labels) in enumerate(data_loader):
#images, labels = data_loader[i]
# Convert torch tensor to Variable
images = Variable(images.view(-1, 14951)).cuda()
labels = Variable(labels).cuda()
outputs = net(images)
loss = criterion(outputs, labels)
prec1, prec1 = tools.Accuracy(outputs, labels, topk=(1, 1))
losses.update(loss.data[0], images.size(0))
acc.update(prec1[0], images.size(0))
if (i + 1) % 100 == 0:
print('Epoch [%d/%d], Step [%d/%d], Loss: %.4f' %
(epoch + 1, num_epochs, i + 1,
len(val_dataset) // batch_size, loss.data[0]))
print('Accuracy : ', (prec1.cpu().data.numpy()[0]))
print('Accuracy (All): ', acc.avg.cpu().data.numpy()[0])
print('Losses (All): ', losses.avg)
print()
return acc.avg
def train(data_loader, net, criterion, optimizer, scheduler, epoch):
losses = tools.AverageMeter()
acc = tools.AverageMeter()
scheduler.step()
net.train(True)
loop = len(data_loader)
for i, (images, labels) in enumerate(data_loader):
# Convert torch tensor to Variable
images = Variable(images.view(-1, 14951)).cuda()
labels = Variable(labels).cuda()
# Forward + Backward + Optimize
optimizer.zero_grad() # zero the gradient buffer
outputs = net(images)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
prec1, prec1 = tools.Accuracy(outputs, labels, topk=(1, 1))
losses.update(loss.data[0], images.size(0))
acc.update(prec1[0], images.size(0))
if (i + 1) % 100 == 0:
print('Epoch [%d/%d], Step [%d/%d], Loss: %.4f' %
(epoch + 1, num_epochs, i + 1,
len(train_dataset) // batch_size, loss.data[0]))
print('Accuracy : ', (prec1.cpu().data.numpy()[0]))
print('Accuracy (All): ', acc.avg.cpu().data.numpy()[0])
print('Losses (All): ', losses.avg)
print()
def val(val_loader, model, criterion):
batch_time = tools.AverageMeter()
losses = tools.AverageMeter()
acc = tools.AverageMeter()
end = time.time()
model.eval()
for i, (img, label) in enumerate(val_loader):
target = label.cuda(async=True)
img = Variable(img).cuda()
label = Variable(label).cuda()
output = model(img)
loss = criterion(output,label)
# measure accuracy and record loss
prec1, prec1 = tools.Accuracy(output.data, target, topk=(1, 1))
losses.update(loss.data[0], img.size(0))
acc.update(prec1[0], img.size(0))
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if i % 100 == 0:
print('\tVal: [{0}/{1}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
'Accuracy {acc.val:.3f} ({acc.avg:.3f})'.format(
i, len(val_loader), batch_time=batch_time, loss=losses, acc=acc))
print(' * Accuracy {acc.avg:.3f}'.format(acc=acc))
return acc.avg
def test(data_loader, net, criterion): losses = tools.AverageMeter() acc = tools.AverageMeter() net.eval() loop = len(data_loader)