def main(args):
config = Config(args)
logger = logging.getLogger(__name__)
def setup_logger(level=logging.DEBUG, filename=None):
logger.setLevel(level)
stream_handler = logging.StreamHandler()
stream_handler.setLevel(logging.DEBUG)
formatter = logging.Formatter('[%(asctime)s] %(message)s')
stream_handler.setFormatter(formatter)
logger.addHandler(stream_handler)
if filename is not None:
file_handler = logging.FileHandler(filename=filename, mode='a')
file_handler.setLevel(logging.INFO)
file_handler.setFormatter(formatter)
logger.addHandler(file_handler)
logger.debug("logger '%s' has been setup.", __name__)
setup_logger(level=logging.DEBUG, filename=config.log_path)
if config.dataset == 'fm':
trainset = dataset.FashionMnist(train=True)
testset = dataset.FashionMnist(train=False)
elif config.dataset == 'c10':
trainset = dataset.Cifar10(train=True)
testset = dataset.Cifar10(train=False)
elif config.dataset == 'svhn':
trainset = dataset.SVHN(train=True)
testset = dataset.SVHN(train=False)
elif config.dataset == 'cinic':
trainset = dataset.Cinic(train=True)
testset = dataset.Cinic(train=False)
else:
raise ValueError("Dataset '{}' not found.".format(config.dataset))
logger.info(
"****************************************************************")
logger.info("%s", config)
logger.info("trainset size: %d", len(trainset))
logger.info("testset size: %d", len(testset))
start_time = time.time()
if config.type == 'stream':
stream(config=config, trainset=trainset, streamset=testset)
logger.info(
"-------------------------------- %.3fs --------------------------------",
time.time() - start_time)
config = json.loads(args.config.replace("'", '"'))
if args.dataset == 'fmnist':
d = dataset.fMNIST()
elif args.dataset == 'binary_fmnist':
d = dataset.fMNIST(binary=True)
elif args.dataset == 'binary_mnist':
d = dataset.MNIST(binary=True)
elif args.dataset == 'mnist':
d = dataset.MNIST()
elif args.dataset in ['omniglot', 'omni']:
d = dataset.Omniglot()
elif args.dataset == 'dsprites':
d = dataset.DSprites()
elif args.dataset == "cifar10" or args.dataset == 'cifar':
d = dataset.Cifar10()
if args.per_label is not None:
d.shrink_supervised(args.per_label)
m = model.NoiseModel(d,
config=config,
filename=args.filename,
verbose=args.verbose)
m.lagrangian_fit = args.fit_tf
if args.noise is not None:
if args.noise == 'multiplicative':
m.layers[0]['layer_kwargs']['multiplicative'] = True
if args.noise == 'additive':
m.layers[0]['layer_kwargs']['multiplicative'] = False