626_privacy/mia_on_model_distillation
2
0
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Wres: epsilon in args

Browse source
This commit is contained in:
Akemi Izuko 2024-12-01 14:49:13 -07:00
parent aa190cd4f1
commit 7208c16efc
Signed by: akemi
GPG key ID: 8DE0764E1809E9FC
1 changed files with 10 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

19
wresnet-pytorch/src/train.py
View file

@ -12,6 +12,8 @@ from tqdm import tqdm
import opacus
from opacus.validators import ModuleValidator
from opacus.utils.batch_memory_manager import BatchMemoryManager
import warnings
warnings.filterwarnings("ignore")
def set_seed(seed=42):
@ -76,18 +78,15 @@ def train_no_cap(net, epochs, data_loader, device, optimizer, criterion, schedul
return best_test_set_accuracy
def _train_seed(net, loaders, device, dataset, log=False, logfile='', epochs=200, norm=1.0):
def _train_seed(net, loaders, device, dataset, log=False, logfile='', epochs=200, norm=1.0, dp_epsilon=None):
train_loader, test_loader = loaders
dp_epsilon = None
dp_delta = 1e-5
checkpointFile = 'wrn-{}-{}e-{}d-{}n-dict.pt'.format(int(time.time()), dp_epsilon, dp_delta, norm)
if dp_epsilon is not None:
print(f"DP epsilon = {dp_epsilon}, delta = {dp_delta}")
#net = ModuleValidator.fix(net, replace_bn_with_in=True)
net = ModuleValidator.fix(net)
ModuleValidator.validate(net, strict=True)
#net = ModuleValidator.fix(net, replace_bn_with_in=True)
net = ModuleValidator.fix(net)
ModuleValidator.validate(net, strict=True)
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=0.1, momentum=0.9, nesterov=True, weight_decay=5e-4)
@ -105,7 +104,8 @@ def _train_seed(net, loaders, device, dataset, log=False, logfile='', epochs=200
max_grad_norm=norm,
)
print(f"Using sigma={optimizer.noise_multiplier} and C={1.0}, norm = {norm}")
print(f"DP epsilon = {dp_epsilon}, delta = {dp_delta}")
print(f"Using sigma={optimizer.noise_multiplier} and C = norm = {norm}")
else:
print("Training without differential privacy")
@ -166,7 +166,7 @@ def train(args):
net = net.to(device)
epochs = training_configurations.epochs
best_test_set_accuracy = _train_seed(net, loaders, device, dataset, log, logfile, epochs, args.norm)
best_test_set_accuracy = _train_seed(net, loaders, device, dataset, log, logfile, epochs, args.norm, args.epsilon)
if log:
with open(logfile, 'a') as temp:
@ -192,6 +192,7 @@ if __name__ == '__main__':
parser.add_argument('-config', '--config', help='Training Configurations', required=True)
parser.add_argument('--norm', type=float, help='dpsgd norm clip factor', required=True)
parser.add_argument('--cuda', type=int, help='gpu index', required=False)
parser.add_argument('--epsilon', type=float, help='dp epsilon', required=False, default=None)
args = parser.parse_args()