from network.wideresnet import WideResNet as WRN
import utils
import warnings
from cutmix.cutmix import CutMix
from cutmix.utils import CutMixCrossEntropyLoss
from autoaug.archive import fa_reduced_cifar10, fa_reduced_imagenet, autoaug_paper_cifar10, autoaug_policy
from autoaug.augmentations import Augmentation
warnings.filterwarnings("ignore")
model_names = sorted(name for name in models.__dict__
if name.islower() and not name.startswith("__")
and callable(models.__dict__[name]))
parser = ConfigArgumentParser(conflict_handler='resolve')
parser.add_argument('-j',
'--workers',
default=16,
type=int,
metavar='N',
help='number of data loading workers (default: 4)')
parser.add_argument('--expname',
default='TEST',
type=str,
help='name of experiment')
parser.add_argument('--cifarpath',
default='/data/private/pretrainedmodels/',
type=str)
parser.add_argument('--imagenetpath',
default='/data/private/pretrainedmodels/imagenet/',
print('loaded')
def infer(root_path, top_k=1):
if C.get()['mode'] != 'test':
return _infer(model, root_path)
if C.get()['infer_mode'] == 'ret':
return _infer_ret(model, root_path)
else:
return _infer(model, root_path)
nsml.bind(save=save, load=load, infer=infer)
if __name__ == '__main__':
# mode argument
args = ConfigArgumentParser(conflict_handler='resolve')
args.add_argument("--cv", type=int, default=0)
args.add_argument("--ratio", type=float, default=0.1)
# reserved for nsml
args.add_argument("--cuda", type=bool, default=True)
args.add_argument("--mode", type=str, default="train")
args.add_argument("--iteration", type=str, default='0')
args.add_argument("--pause", type=int, default=0)
args.add_argument("--transfer", type=bool, default=False)
config = args.parse_args()
logger.info(str(C.get().conf))
from theconf import Config as C
from commons import tta_num, get_logger, decode, \
print_log, encoded_tta_default, mirror_expansion
from conf import dataroot
from imagenet import ImageNet
from metrics import accuracy, Accumulator
from networks import get_model
from profile import Profiler
logger = get_logger('learn2test')
logger.setLevel(logging.DEBUG)
if __name__ == '__main__':
parser = ConfigArgumentParser(conflict_handler='resolve')
parser.add_argument('--test-batch', type=int, default=32)
parser.add_argument('--tta', type=str, default='center')
parser.add_argument('--deform', type=str, default='')
parser.add_argument('--corrupt', type=str, default='')
args = parser.parse_args()
assert args.dataset == 'imagenet'
model_target = get_model(args.target_network, gpus=[0], num_classes=args.num_classes, train_aug=args.target_aug).eval()
profiler = Profiler(model_target)
print('target network, FLOPs=', profiler.flops(torch.zeros((1, 3, C.get()['target_size'], C.get()['target_size'])).cuda(), ))
scaled_size = int(math.floor(args.target_size / 0.875))
if args.deform != '':
testset1 = ImageNet(root=imagenet_path,
split='val',
transform=transform_train)
testset2 = ImageNet(root=imagenet_path,
split='val',
transform=transform_test)
trainset.num_class = testset1.num_class = testset2.num_class = 1000
trainset.targets = [lb for _, lb in trainset.samples]
else:
raise ValueError(dataset)
return trainset, testset1, testset2
if __name__ == '__main__':
parser = ConfigArgumentParser(conflict_handler='resolve')
parser.add_argument('--aug-p', default=0.2, type=float)
parser.add_argument('--port', default=1958, type=int)
args = parser.parse_args()
total_trainset, testset_t, testset = get_dataset(args.dataset)
num_class = total_trainset.num_class
sss = StratifiedShuffleSplit(n_splits=5, test_size=0.2, random_state=0)
sss = sss.split(list(range(len(total_trainset))), total_trainset.targets)
train_idx, valid_idx = next(sss)
valid_size = 25600
small_train_size = 2048
test_size = len(testset)
validset = Subset(
testset_t, valid_idx[:valid_size]) # TODO : max images for validation
from architectures.efficientnet_pytorch.rmsproptf import RMSpropTF
from metrics import Accumulator, prediction_correlation
from networks import get_model
from profile import Profiler
from remote_dataloader.loader import RemoteDataLoader
from theconf import Config as C
from imagenet_c import corruption_dict
from conf import sodeep_model
from sodeep.sodeep import SpearmanLoss, load_sorter
if __name__ == '__main__':
logger = get_logger('learn2test-train')
logger.setLevel(logging.DEBUG)
parser = ConfigArgumentParser(conflict_handler='resolve')
parser.add_argument('--tag', default='dev', type=str)
parser.add_argument('--port', default=1958, type=int)
parser.add_argument('--cv', default=0, type=int)
parser.add_argument('--regularization', default=2, type=int)
parser.add_argument('--ema-momentum', default=0.999, type=float)
parser.add_argument('--data-cache', default=0, type=int)
parser.add_argument('--cutout', default=-1, type=int)
parser.add_argument('--aug-corrupt', default=1, type=int)
parser.add_argument('--aug-p', default=0.4, type=float)
args = parser.parse_args()
tv = 'v8' # 12 tta, bugfix
y = np.exp(y)
# take the sum along the specified axis
ax_sum = np.expand_dims(np.sum(y, axis=axis), axis)
# finally: divide elementwise
p = y / ax_sum
# flatten if X was 1D
if len(X.shape) == 1: p = p.flatten()
return p
if __name__ == '__main__':
parser = ConfigArgumentParser(conflict_handler='resolve')
args = parser.parse_args()
# L2T Dataset
trainset, testset = get_dataset(args.dataset)
# Legacy Dataset
transform_test = transforms.Compose([
transforms.Resize(args.target_size + 32, interpolation=Image.BICUBIC),
transforms.CenterCrop(args.target_size),
])
testset_legacy = ImageNet(root='/data/public/rw/datasets/imagenet-pytorch',
split='val',
transform=transform_test)