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

Torchlira: remove secure mode

Browse source
This commit is contained in:
Akemi Izuko 2024-11-29 18:30:26 -07:00
parent fe7fcb2761
commit c7eee3cdc2
Signed by: akemi
GPG key ID: 8DE0764E1809E9FC
1 changed files with 1 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
lira-pytorch/train.py
View file

@ -115,7 +115,7 @@ def run():
optim = torch.optim.SGD(m.parameters(), lr=args.lr, momentum=0.9, weight_decay=5e-4) optim = torch.optim.SGD(m.parameters(), lr=args.lr, momentum=0.9, weight_decay=5e-4)
sched = torch.optim.lr_scheduler.CosineAnnealingLR(optim, T_max=args.epochs) sched = torch.optim.lr_scheduler.CosineAnnealingLR(optim, T_max=args.epochs)
privacy_engine = PrivacyEngine(accountant='rdp', secure_mod=True) privacy_engine = PrivacyEngine()
m, optim, train_dl = privacy_engine.make_private_with_epsilon( m, optim, train_dl = privacy_engine.make_private_with_epsilon(
module=m, module=m,
optimizer=optim, optimizer=optim,
@ -135,7 +135,6 @@ def run():
max_physical_batch_size=1000, max_physical_batch_size=1000,
optimizer=optim optimizer=optim
) as memory_safe_data_loader: ) as memory_safe_data_loader:
for i in tqdm(range(args.epochs)): for i in tqdm(range(args.epochs)):
m.train() m.train()
loss_total = 0 loss_total = 0