def train(self, model, optimizer, supernet=False):
objs = utils.AvgrageMeter()
top1 = utils.AvgrageMeter()
top5 = utils.AvgrageMeter()
model.train()
iterative_indices = np.random.permutation(self.iterative_indices)
train_loader = self.train_loader_super if supernet else self.train_loader_sub
for step, (input, target) in enumerate(train_loader):
if self.mode == 'uniform':
mask = self.subnet_masks[iterative_indices[
step]] if not supernet else self.supernet_mask
else:
mask = random.choice(self.subnet_masks)
input = Variable(input).cuda()
target = Variable(target).cuda(non_blocking=True)
optimizer.zero_grad()
logits = model(input, mask=mask)
loss = self.criterion(logits, target)
loss.backward()
nn.utils.clip_grad_norm_(model.parameters(), self.grad_clip)
optimizer.step()
prec1, prec5 = utils.accuracy(logits, target, topk=(1, 5))
n = input.size(0)
objs.update(loss.data.item(), n)
top1.update(prec1.data.item(), n)
top5.update(prec5.data.item(), n)
if step % self.report_freq == 0:
logging.info('supernet train %03d %e %f %f', step, objs.avg,
top1.avg / 100, top5.avg / 100)
copy_log_dir()
def train(train_queue, model, criterion, optimizer):
objs = utils.AvgrageMeter()
top1 = utils.AvgrageMeter()
top5 = utils.AvgrageMeter()
model.train()
for step, (input, target) in enumerate(train_queue):
target = target.cuda(async=True)
input = input.cuda()
input = Variable(input)
target = Variable(target)
optimizer.zero_grad()
logits, logits_aux = model(input)
loss = criterion(logits, target)
if args.auxiliary:
loss_aux = criterion(logits_aux, target)
loss += args.auxiliary_weight * loss_aux
loss.backward()
nn.utils.clip_grad_norm(model.parameters(), args.grad_clip)
optimizer.step()
prec1, prec5 = utils.accuracy(logits, target, topk=(1, 5))
n = input.size(0)
objs.update(loss.item(), n)
top1.update(prec1.item(), n)
top5.update(prec5.item(), n)
if step % args.report_freq == 0:
logging.info('train %03d %e %f %f', step, objs.avg, top1.avg, top5.avg)
return top1.avg, objs.avg
def evaluate(self, model, mask):
objs = utils.AvgrageMeter()
top1 = utils.AvgrageMeter()
top5 = utils.AvgrageMeter()
model.eval()
with torch.no_grad():
for step, (input, target) in enumerate(self.valid_loader):
input = Variable(input).cuda()
target = Variable(target).cuda(non_blocking=True)
logits = model(input, mask=mask)
loss = self.criterion(logits, target)
prec1, prec5 = utils.accuracy(logits, target, topk=(1, 5))
n = input.size(0)
objs.update(loss.data.item(), n)
top1.update(prec1.data.item(), n)
top5.update(prec5.data.item(), n)
return objs.avg, top1.avg / 100, top5.avg / 100
def infer(valid_queue, model, criterion):
objs = utils.AvgrageMeter()
top1 = utils.AvgrageMeter()
top5 = utils.AvgrageMeter()
model.eval()
for step, (input, target) in enumerate(valid_queue):
input = Variable(input, volatile=True).cuda()
target = Variable(target, volatile=True).cuda(async=True)
logits, _ = model(input)
loss = criterion(logits, target)
prec1, prec5 = utils.accuracy(logits, target, topk=(1, 5))
n = input.size(0)
objs.update(loss.item(), n)
top1.update(prec1.item(), n)
top5.update(prec5.item(), n)
if step % args.report_freq == 0:
logging.info('valid %03d %e %f %f', step, objs.avg, top1.avg, top5.avg)
return top1.avg, top5.avg, objs.avg