def validate(val_loader,
model,
criterion,
epoch,
print_freq,
save_history=True,
ngpu=1):
device = torch.device("cuda:0" if (
torch.cuda.is_available() and ngpu > 0) else "cpu")
batch_time = metrics.AverageMeter()
losses = metrics.AverageMeter()
top1 = metrics.AverageMeter()
top5 = metrics.AverageMeter()
history = {'epoch': [], 'loss': [], 'acc_topk1': [], 'acc_topk5': []}
# switch to evaluate mode
model.eval()
with torch.no_grad():
end = time.time()
for i, (input, target) in enumerate(val_loader):
input = input.to(device)
target = target.to(device)
# compute output
output = model(input)
loss = criterion(output, target)
# measure metrics.accuracy and record loss
acc1, acc5 = metrics.accuracy(output, target, topk=(1, 5))
losses.update(loss.item(), input.size(0))
top1.update(acc1[0], input.size(0))
top5.update(acc5[0], input.size(0))
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if i % print_freq == 0:
history['epoch'].append(epoch)
history['loss'].append(losses.avg)
history['acc_topk1'].append(top1.avg)
history['acc_topk5'].append(top5.avg)
print('Test: [{0}/{1}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
'Acc@1 {top1.val:.3f} ({top1.avg:.3f})\t'
'Acc@5 {top5.val:.3f} ({top5.avg:.3f})'.format(
i,
len(val_loader),
batch_time=batch_time,
loss=losses,
top1=top1,
top5=top5))
print(' * Acc@1 {top1.avg:.3f} Acc@5 {top5.avg:.3f}'.format(top1=top1,
top5=top5))
return history
def val(model, loader, loss_criterion, label_names, cuda=True):
model.eval()
loader = tqdm(loader, ncols=100)
losses = metrics.AverageMeter()
cm = metrics.ConfusionMatrixMeter(label_names, cmap='Blues')
with torch.no_grad():
for i, batch in enumerate(loader, start=0):
if cuda: batch = batch.to('cuda:0')
batch_size = len(batch.batch.unique())
outputs = model(batch)
out_device = outputs.device
gt = batch.y.to(out_device)
loss = loss_criterion(outputs, gt)
batch_size = len(torch.unique(batch.batch))
losses.update(loss.item(), batch_size)
cm.add(gt.cpu().data.numpy(), outputs.cpu().data.numpy())
loader.set_postfix({"loss": loss.item()})
torch.cuda.empty_cache()
return losses.avg, cm
def train(model,
loader,
loss_criteron,
optimizer,
label_names,
batch_parts=0,
cuda=True):
model.train()
numIt = len(loader)
loader = tqdm(loader, ncols=100)
acc = metrics.AverageMeter()
losses = metrics.AverageMeter()
cm = metrics.ConfusionMatrixMeter(label_names, cmap='Oranges')
prev = -1
optimizer.zero_grad()
for i, batch in enumerate(loader, start=0):
if cuda: batch = batch.to('cuda:0')
outputs = model(batch)
out_device = outputs.device
gt = batch.y.to(out_device)
loss = loss_criterion(outputs, gt)
loss.backward()
batch_size = len(torch.unique(batch.batch))
losses.update(loss.item(), batch_size)
cm.add(gt.cpu().data.numpy(), outputs.cpu().data.numpy())
if (i + 1) % batch_parts == 0 or (i + 1) == numIt:
if batch_parts > 1:
accum = i - prev
prev = i
for p in model.parameters():
p.grad.div_(accum)
optimizer.step()
optimizer.zero_grad()
torch.cuda.empty_cache()
return losses.avg, cm
def train(train_loader,
model,
criterion,
optimizer,
epoch,
print_freq,
save_history=True,
ngpu=1):
device = torch.device("cuda:0" if (
torch.cuda.is_available() and ngpu > 0) else "cpu")
batch_time = metrics.AverageMeter()
data_time = metrics.AverageMeter()
losses = metrics.AverageMeter()
top1 = metrics.AverageMeter()
top5 = metrics.AverageMeter()
history = {'epoch': [], 'loss': [], 'acc_topk1': [], 'acc_topk5': []}
# switch to train mode
model.train()
end = time.time()
for i, (input, target) in enumerate(train_loader):
# measure data loading time
data_time.update(time.time() - end)
input = input.to(device)
target = target.to(device)
optimizer.zero_grad()
# compute output
output = model(input)
loss = criterion(output, target)
# measure metrics.accuracy and record loss
acc1, acc5 = metrics.accuracy(output, target, topk=(1, 5))
losses.update(loss.item(), input.size(0))
top1.update(acc1[0], input.size(0))
top5.update(acc5[0], input.size(0))
# compute gradient and do SGD step
loss.backward()
optimizer.step()
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if i % print_freq == 0:
history['epoch'].append(epoch)
history['loss'].append(losses.avg)
history['acc_topk1'].append(top1.avg)
history['acc_topk5'].append(top5.avg)
print('Epoch: [{0}][{1}/{2}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Data {data_time.val:.3f} ({data_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
'Acc@1 {top1.val:.3f} ({top1.avg:.3f})\t'
'Acc@5 {top5.val:.3f} ({top5.avg:.3f})'.format(
epoch,
i,
len(train_loader),
batch_time=batch_time,
data_time=data_time,
loss=losses,
top1=top1,
top5=top5))
return history