def test(test_loader, network, C_out):
test_top1, test_top5 = AverageMeter(), AverageMeter()
network.eval()
for X, Y in test_loader:
test_inputs, test_targets = X.cuda(), Y.cuda()
output = network(test_inputs)
test_prec1, test_prec5 = obtain_accuracy(output.data,
test_targets.data,
topk=(1, min(5, C_out)))
test_top1.update(test_prec1.item(), test_inputs.size(0))
test_top5.update(test_prec5.item(), test_inputs.size(0))
print("***TEST result***" +
"accuracy@1 : {:.2f}%, accuracy@5 : {:.2f}%".format(
test_top1.avg, test_top5.avg))
def test(test_loader, network, C_out):
test_top1, test_top5 = AverageMeter(), AverageMeter()
network.eval()
with torch.no_grad():
for X, Y in test_loader:
test_inputs, test_targets = X.cuda(), Y.cuda()
output = network(test_inputs)
import numpy as np
# print(np.argmax(np.array(output.data.cpu()), axis=1))
# print(test_targets.data)
test_prec1, test_prec5 = obtain_accuracy(output.data,
test_targets.data,
topk=(1, min(5, C_out)))
test_top1.update(test_prec1.item(), test_inputs.size(0))
test_top5.update(test_prec5.item(), test_inputs.size(0))
return test_top1.avg, test_top5.avg
def train(data_loader, network, criterion, optimizer, config):
network.train()
losses, top1, top5 = AverageMeter(), AverageMeter(), AverageMeter()
for step, (X, Y, X1, Y1) in enumerate(data_loader):
inputs, targets = X.cuda(), Y.cuda()
optimizer.zero_grad()
output = network(inputs)
loss = criterion(output, targets)
loss.backward()
optimizer.step()
prec1, prec5 = obtain_accuracy(output.data,
targets.data,
topk=(1, min(5, config.C_out)))
losses.update(loss.item(), inputs.size(0))
top1.update(prec1.item(), inputs.size(0))
top5.update(prec5.item(), inputs.size(0))
return losses.avg, top1.avg, top5.avg
def train(xargs):
lib_dir = (Path(__file__).parent / '..' / '..' / 'lib').resolve()
if str(lib_dir) not in sys.path:
sys.path.insert(0, str(lib_dir))
assert (torch.cuda.is_available(), 'CUDA is not available.')
# start cudnn
cudnn.enabled = True
# make each conv is the same
cudnn.benchmark = False
# make sure the same seed has the same result
cudnn.deterministic = True
torch.set_num_threads(xargs.workers)
prepare_seed(xargs.rand_seed)
logger = prepare_logger(xargs)
# get original data(cifar10/cifar100/uci)
train_data, valid_data, xshape, class_num = get_dataset(
xargs.dataset, xargs.data_path, -1)
logger.log('{:}Train Config{:}'.format("-" * 50, "-" * 50))
opt_config = load_config(
xargs.opt_config, {
'class_num': class_num,
'xshape': xshape,
'batch_size': xargs.batch_size,
'epochs': xargs.epochs,
'LR': xargs.opt_learning_rate
}, logger)
search_loader, _, valid_loader = get_nas_search_loaders(
train_data, valid_data, xargs.dataset, 'config/',
opt_config.batch_size, xargs.workers)
logger.log('dataset: {:} Search-Loader-length={:}, batch size={:}'.format(
xargs.dataset, len(search_loader), opt_config.batch_size))
logger.log('{:}Arch Config{:}'.format("-" * 50, "-" * 50))
arch_config = load_config(
xargs.arch_config, {
'class_num': class_num,
'space': HAPT_SPACE,
'affine': False,
'track_running_stats': bool(xargs.track_running_stats)
}, logger)
if xargs.dataset == 'HAPT':
search_model = DNNModel(config=arch_config, logger=logger)
elif xargs.dataset in ('cifar10', 'cifar100'):
search_model = DNNModel(config=arch_config, logger=logger)
else:
raise NameError(
"dataset must be in \"HAPT\", \"cifar100\", \"cifar100\"")
if xargs.evaluate == 'test':
search_model.load_state_dict(
torch.load(logger.path('best'))['network'])
network = search_model.cuda()
test_loader = valid_loader
test(test_loader, network, arch_config.C_out)
return
# logger.log('search-model :\n{:}'.format(search_model))
logger.log('{:}model-config{:}\n{:}'.format("-" * 50, "-" * 50,
arch_config))
if opt_config.criterion == 'cross_entropy':
criterion = nn.CrossEntropyLoss()
else:
raise NameError('unknown loss function {:}'.format(
opt_config.criterion))
# criterion = nn.MSELoss()
w_optimizer = torch.optim.SGD(params=search_model.get_weights(),
lr=opt_config.LR,
weight_decay=opt_config.w_decay)
w_scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(
optimizer=w_optimizer,
T_max=opt_config.epochs,
eta_min=opt_config.eta_min)
a_optimizer = torch.optim.Adam(params=search_model.get_alphas(),
lr=xargs.arch_learning_rate,
betas=(0.5, 0.999),
weight_decay=opt_config.a_decay)
logger.log('{:}w-optimizer{:}\n{:}'.format("-" * 50, "-" * 50,
w_optimizer))
logger.log('{:}a-optimizer{:}\n{:}'.format("-" * 50, "-" * 50,
a_optimizer))
logger.log('{:}w-scheduler{:}\n{:}'.format("-" * 50, "-" * 50,
w_scheduler))
logger.log('{:}criterion{:}\n{:}'.format("-" * 50, "-" * 50, criterion))
flop, param = get_model_infos(search_model, xshape)
logger.log('FLOP = {:.6f} M, Params = {:.6f} MB'.format(flop, param))
logger.log('search-space [{:} ops] : {:}'.format(len(HAPT_SPACE),
HAPT_SPACE))
last_info, model_base_path, model_best_path = logger.path(
'info'), logger.path('model'), logger.path('best')
# network, criterion = torch.nn.DataParallel(search_model).cuda(), criterion.cuda()
network = search_model.cuda()
criterion = criterion.cuda()
if last_info.exists():
# automatically resume from previous checkpoint
logger.log("=> loading checkpoint of the last-info '{:}' start".format(
last_info))
last_info = torch.load(last_info)
start_epoch = last_info['epoch']
checkpoint = torch.load(last_info['last_checkpoint'])
genotypes = checkpoint['genotypes']
valid_accuracies = checkpoint['valid_accuracies']
network.load_state_dict(checkpoint['network'])
w_scheduler.load_state_dict(checkpoint['w_scheduler'])
w_optimizer.load_state_dict(checkpoint['w_optimizer'])
a_optimizer.load_state_dict(checkpoint['a_optimizer'])
logger.log(
"=> loading checkpoint of the last-info '{:}' start with {:}-th epoch."
.format(last_info, start_epoch))
else:
logger.log("=> do not find the last-info file : {:}".format(last_info))
start_epoch, valid_accuracies, genotypes = 0, {'best': -1}, \
{-1: network.genotype(xargs.save_dir + xargs.genotype_file)}
start_time, epoch_time = time.time(), AverageMeter()
total_epoch = opt_config.epochs
for epoch in range(start_epoch, total_epoch):
w_scheduler.step(epoch=epoch)
need_time = 'Time Left: {:}'.format(
convert_secs2time(epoch_time.avg * (total_epoch - epoch), True))
epoch_str = '{:03d}-{:03d}'.format(epoch, total_epoch)
network.set_tau(xargs.tau_max -
(xargs.tau_max - xargs.tau_min) * epoch /
(total_epoch - 1))
logger.log('\n[Search the {:}-th epoch] tau={:.2f} {:}'.format(
epoch_str, network.get_tau(), need_time))
epoch_start = time.time()
base_losses, base_top1, base_top5, arch_losses, arch_top1, arch_top5 = \
search(arch_config, search_loader, network, criterion, w_optimizer, a_optimizer,
xargs.print_frequency, epoch_str, logger)
epoch_time.update(time.time() - epoch_start)
logger.log(
'[{:}] searching : loss={:.2f}, accuracy@1={:.2f}%, accuracy@5={:.2f}%'
.format(epoch_str, base_losses, base_top1, base_top5))
logger.log(
'[{:}] evaluate : loss={:.2f}, accuracy@1={:.2f}%, accuracy@5={:.2f}%'
.format(epoch_str, arch_losses, arch_top1, arch_top5))
valid_accuracies[epoch] = arch_top1
if arch_top1 > valid_accuracies['best']:
valid_accuracies['best'] = arch_top1
genotypes['best'] = network.genotype(xargs.save_dir +
xargs.genotype_file)
find_best = True
else:
find_best = False
genotypes[epoch] = network.genotype(xargs.save_dir +
xargs.genotype_file)
logger.log('<<<--->>> The {:}-th epoch : {:}'.format(
epoch_str, genotypes[epoch]))
save_path = save_checkpoint(
{
'epoch': epoch + 1,
'args': deepcopy(xargs),
'network': network.state_dict(),
'w_optimizer': w_optimizer.state_dict(),
'a_optimizer': a_optimizer.state_dict(),
'w_scheduler': w_scheduler.state_dict(),
'genotypes': genotypes,
'valid_accuracies': valid_accuracies
}, model_base_path, logger)
last_info = save_checkpoint(
{
'epoch': epoch + 1,
'args': deepcopy(xargs),
'last_checkpoint': save_path,
}, logger.path('info'), logger)
if find_best:
logger.log(
'<<<--->>> The {:}-th epoch : find the highest validation accuracy : {:.2f}%.'
.format(epoch_str, arch_top1))
copy_checkpoint(model_base_path, model_best_path, logger)
with torch.no_grad():
logger.log('{:}'.format(network.show_alphas()))
logger.log('\n' + '-' * 100)
# check the performance from the architecture dataset
logger.log('GDAS : run {:} epochs, use time : {:}.'.format(
total_epoch, convert_secs2time(time.time() - start_time, True)))
logger.log('best geno is {:}.'.format(genotypes['best']))
def search(arch_config, data_loader, network, criterion, w_optimizer,
a_optimizer, print_frequency, epoch_str, logger):
base_losses, base_top1, base_top5 = AverageMeter(), AverageMeter(
), AverageMeter()
arch_losses, arch_top1, arch_top5 = AverageMeter(), AverageMeter(
), AverageMeter()
network.train()
for step, (X, Y, X1, Y1) in enumerate(data_loader):
base_inputs, base_targets = X.cuda(), Y.cuda()
arch_inputs, arch_targets = X1.cuda(), Y1.cuda()
# update the weights
w_optimizer.zero_grad()
output = network(base_inputs)
base_loss = criterion(output, base_targets)
base_loss.backward()
clip_grad_norm_(network.parameters(), 5)
w_optimizer.step()
# record
base_prec1, base_prec5 = obtain_accuracy(
output.data,
base_targets.data,
topk=(1, min(5, arch_config.C_out)))
base_losses.update(base_loss.item(), base_inputs.size(0))
base_top1.update(base_prec1.item(), base_inputs.size(0))
base_top5.update(base_prec5.item(), base_inputs.size(0))
# update the architecture-weight
a_optimizer.zero_grad()
output = network(arch_inputs)
arch_loss = criterion(output, arch_targets)
arch_loss.backward()
a_optimizer.step()
# record
arch_prec1, arch_prec5 = obtain_accuracy(
output.data,
arch_targets.data,
topk=(1, min(5, arch_config.C_out)))
arch_losses.update(arch_loss.item(), arch_inputs.size(0))
arch_top1.update(arch_prec1.item(), arch_inputs.size(0))
arch_top5.update(arch_prec5.item(), arch_inputs.size(0))
if step % print_frequency == 0 or step + 1 == len(data_loader):
str1 = "SEARCHING***" + "[{:}][{:}/{:}]".format(
epoch_str, step, len(data_loader))
str3 = "Base [Loss {loss.val:.3f} ({loss.avg:.3f}) Prec@1 {top1.val:.2f} ({top1.avg:.2f}) Prec@5 {top5.val:.2f} ({top5.avg:.2f})]".format(
loss=base_losses, top1=base_top1, top5=base_top5)
str4 = 'Arch [Loss {loss.val:.3f} ({loss.avg:.3f}) Prec@1 {top1.val:.2f} ({top1.avg:.2f}) Prec@5 {top5.val:.2f} ({top5.avg:.2f})]'.format(
loss=arch_losses, top1=arch_top1, top5=arch_top5)
logger.log(str1 + ' ' + str3 + ' ' + str4)
return base_losses.avg, base_top1.avg, base_top5.avg, arch_losses.avg, arch_top1.avg, arch_top5.avg
def train(xargs):
lib_dir = (Path(__file__).parent / '..' / '..' / 'lib').resolve()
if str(lib_dir) not in sys.path:
sys.path.insert(0, str(lib_dir))
assert (torch.cuda.is_available(), 'CUDA is not available.')
# start cudnn
cudnn.enabled = True
# make each conv is the same
cudnn.benchmark = False
# make sure the same seed has the same result
cudnn.deterministic = True
torch.set_num_threads(xargs.workers)
prepare_seed(xargs.rand_seed)
logger = prepare_logger(xargs)
# get original data
train_data, valid_data, xshape, class_num = get_dataset(
xargs.dataset, xargs.data_path, -1)
config = load_config(xargs.config_path, {
'class_num': class_num,
'xshape': xshape
}, logger)
search_loader, _, valid_loader = get_nas_search_loaders(
train_data, valid_data, xargs.dataset, 'config/', config.batch_size,
xargs.workers)
logger.log(
'||||||| {:10s} ||||||| Search-Loader-Num={:}, batch size={:}'.format(
xargs.dataset, len(search_loader), config.batch_size))
logger.log('||||||| {:10s} ||||||| Config={:}'.format(
xargs.dataset, config))
search_space = DARTS_SPACE
model_config = load_config(
xargs.model_config, {
'num_classes': class_num,
'space': search_space,
'affine': False,
'track_running_stats': bool(xargs.track_running_stats)
}, None)
search_model = get_cell_based_tiny_net(model_config)
# logger.log('search-model :\n{:}'.format(search_model))
# logger.log('model-config : {:}'.format(model_config))
w_optimizer, w_scheduler, criterion = get_optim_scheduler(
search_model.get_weights(), config)
a_optimizer = torch.optim.Adam(search_model.get_alphas(),
lr=xargs.arch_learning_rate,
betas=(0.5, 0.999),
weight_decay=xargs.arch_weight_decay)
# logger.log('w-optimizer : {:}'.format(w_optimizer))
# logger.log('a-optimizer : {:}'.format(a_optimizer))
# logger.log('w-scheduler : {:}'.format(w_scheduler))
# logger.log('criterion : {:}'.format(criterion))
# flop, param = get_model_infos(search_model, xshape)
# logger.log('FLOP = {:.2f} M, Params = {:.2f} MB'.format(flop, param))
logger.log('search-space [{:} ops] : {:}'.format(len(search_space),
search_space))
if xargs.arch_nas_dataset is None:
api = None
else:
pass
# api = API(xargs.arch_nas_dataset)
last_info, model_base_path, model_best_path = logger.path(
'info'), logger.path('model'), logger.path('best')
# network, criterion = torch.nn.DataParallel(search_model).cuda(), criterion.cuda()
network = search_model
if last_info.exists():
# automatically resume from previous checkpoint
logger.log("=> loading checkpoint of the last-info '{:}' start".format(
last_info))
last_info = torch.load(last_info)
start_epoch = last_info['epoch']
checkpoint = torch.load(last_info['last_checkpoint'])
genotypes = checkpoint['genotypes']
valid_accuracies = checkpoint['valid_accuracies']
search_model.load_state_dict(checkpoint['search_model'])
w_scheduler.load_state_dict(checkpoint['w_scheduler'])
w_optimizer.load_state_dict(checkpoint['w_optimizer'])
a_optimizer.load_state_dict(checkpoint['a_optimizer'])
logger.log(
"=> loading checkpoint of the last-info '{:}' start with {:}-th epoch."
.format(last_info, start_epoch))
else:
logger.log("=> do not find the last-info file : {:}".format(last_info))
start_epoch, valid_accuracies, genotypes = 0, {
'best': -1
}, {
-1: search_model.genotype()
}
start_time, search_time, epoch_time, total_epoch = time.time(
), AverageMeter(), AverageMeter(), config.epochs + config.warmup
for epoch in range(start_epoch, total_epoch):
w_scheduler.update(epoch, 0.0)
need_time = 'Time Left: {:}'.format(
convert_secs2time(epoch_time.val * (total_epoch - epoch), True))
epoch_str = '{:03d}-{:03d}'.format(epoch, total_epoch)
search_model.set_tau(xargs.tau_max -
(xargs.tau_max - xargs.tau_min) * epoch /
(total_epoch - 1))
logger.log('\n[Search the {:}-th epoch] {:}, tau={:}, LR={:}'.format(
epoch_str, need_time, search_model.get_tau(),
min(w_scheduler.get_lr())))
search_w_loss, search_w_top1, search_w_top5, valid_a_loss, valid_a_top1, valid_a_top5 \
= search_func(search_loader, network, criterion, w_scheduler, w_optimizer, a_optimizer, epoch_str,
xargs.print_freq, logger)
search_time.update(time.time() - start_time)
logger.log(
'[{:}] searching : loss={:.2f}, accuracy@1={:.2f}%, accuracy@5={:.2f}%, time-cost={:.1f} s'
.format(epoch_str, search_w_loss, search_w_top1, search_w_top5,
search_time.sum))
logger.log(
'[{:}] evaluate : loss={:.2f}, accuracy@1={:.2f}%, accuracy@5={:.2f}%'
.format(epoch_str, valid_a_loss, valid_a_top1, valid_a_top5))
# check the best accuracy
valid_accuracies[epoch] = valid_a_top1
if valid_a_top1 > valid_accuracies['best']:
valid_accuracies['best'] = valid_a_top1
genotypes['best'] = search_model.genotype()
find_best = True
else:
find_best = False
genotypes[epoch] = search_model.genotype()
logger.log('<<<--->>> The {:}-th epoch : {:}'.format(
epoch_str, genotypes[epoch]))
# save checkpoint
save_path = save_checkpoint(
{
'epoch': epoch + 1,
'args': deepcopy(xargs),
'search_model': search_model.state_dict(),
'w_optimizer': w_optimizer.state_dict(),
'a_optimizer': a_optimizer.state_dict(),
'w_scheduler': w_scheduler.state_dict(),
'genotypes': genotypes,
'valid_accuracies': valid_accuracies
}, model_base_path, logger)
last_info = save_checkpoint(
{
'epoch': epoch + 1,
'args': deepcopy(xargs),
'last_checkpoint': save_path,
}, logger.path('info'), logger)
if find_best:
logger.log(
'<<<--->>> The {:}-th epoch : find the highest validation accuracy : {:.2f}%.'
.format(epoch_str, valid_a_top1))
copy_checkpoint(model_base_path, model_best_path, logger)
with torch.no_grad():
logger.log('{:}'.format(search_model.show_alphas()))
# if api is not None: logger.log('{:}'.format(api.query_by_arch(genotypes[epoch])))
# measure elapsed time
epoch_time.update(time.time() - start_time)
start_time = time.time()
logger.log('\n' + '-' * 100)
# check the performance from the architecture dataset
logger.log(
'GDAS : run {:} epochs, cost {:.1f} s, last-geno is {:}.'.format(
total_epoch, search_time.sum, genotypes[total_epoch - 1]))
# if api is not None: logger.log('{:}'.format(api.query_by_arch(genotypes[total_epoch - 1])))
logger.close()
def search_func(xloader, network, criterion, scheduler, w_optimizer,
a_optimizer, epoch_str, print_freq, logger):
data_time, batch_time = AverageMeter(), AverageMeter()
base_losses, base_top1, base_top5 = AverageMeter(), AverageMeter(
), AverageMeter()
arch_losses, arch_top1, arch_top5 = AverageMeter(), AverageMeter(
), AverageMeter()
network.train()
end = time.time()
for step, (base_inputs, base_targets, arch_inputs,
arch_targets) in enumerate(xloader):
scheduler.update(None, 1.0 * step / len(xloader))
# base_targets = base_targets.cuda(non_blocking=True)
# arch_targets = arch_targets.cuda(non_blocking=True)
# measure data loading time
data_time.update(time.time() - end)
# update the weights
w_optimizer.zero_grad()
_, logits = network(base_inputs)
base_loss = criterion(logits, base_targets)
base_loss.backward()
torch.nn.utils.clip_grad_norm_(network.parameters(), 5)
w_optimizer.step()
# record
base_prec1, base_prec5 = obtain_accuracy(logits.data,
base_targets.data,
topk=(1, 5))
base_losses.update(base_loss.item(), base_inputs.size(0))
base_top1.update(base_prec1.item(), base_inputs.size(0))
base_top5.update(base_prec5.item(), base_inputs.size(0))
# update the architecture-weight
a_optimizer.zero_grad()
_, logits = network(arch_inputs)
arch_loss = criterion(logits, arch_targets)
arch_loss.backward()
a_optimizer.step()
# record
arch_prec1, arch_prec5 = obtain_accuracy(logits.data,
arch_targets.data,
topk=(1, 5))
arch_losses.update(arch_loss.item(), arch_inputs.size(0))
arch_top1.update(arch_prec1.item(), arch_inputs.size(0))
arch_top5.update(arch_prec5.item(), arch_inputs.size(0))
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if step % print_freq == 0 or step + 1 == len(xloader):
Sstr = '*SEARCH* ' + time_string(
) + ' [{:}][{:03d}/{:03d}]'.format(epoch_str, step, len(xloader))
Tstr = 'Time {batch_time.val:.2f} ({batch_time.avg:.2f}) Data {data_time.val:.2f} ({data_time.avg:.2f})'.format(
batch_time=batch_time, data_time=data_time)
Wstr = 'Base [Loss {loss.val:.3f} ({loss.avg:.3f}) Prec@1 {top1.val:.2f} ({top1.avg:.2f}) Prec@5 {top5.val:.2f} ({top5.avg:.2f})]'.format(
loss=base_losses, top1=base_top1, top5=base_top5)
Astr = 'Arch [Loss {loss.val:.3f} ({loss.avg:.3f}) Prec@1 {top1.val:.2f} ({top1.avg:.2f}) Prec@5 {top5.val:.2f} ({top5.avg:.2f})]'.format(
loss=arch_losses, top1=arch_top1, top5=arch_top5)
logger.log(Sstr + ' ' + Tstr + ' ' + Wstr + ' ' + Astr)
return base_losses.avg, base_top1.avg, base_top5.avg, arch_losses.avg, arch_top1.avg, arch_top5.avg