Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Multiple GPU's and optimzer/lr_scheduler ordering #5

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Multiple GPU's and optimzer/lr_scheduler ordering #5

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 15 additions & 2 deletions train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -85,7 +85,6 @@ def train(self):
# Training loop
icount_loss = []
while self.icount < self.args.max_iteration:
model_lr_scheduler.step()
for step, (inputs_train, mask_train) in enumerate(dataset_train):
inputs_train = inputs_train.cuda()
mask_train = mask_train.cuda()
Expand All @@ -112,6 +111,9 @@ def train(self):
if self.args.icount_save > 0 and self.icount % self.args.icount_save == 0:
self.checkpoint()

# Call lr_schduler.step() after optimizer.step()
model_lr_scheduler.step()

f_loss.close()

def test(self):
Expand Down Expand Up @@ -160,7 +162,13 @@ def checkpoint(self):
filename = 'cscdnet-{0:08d}.pth'.format(self.icount)
else:
filename = 'cdnet-{0:08d}.pth'.format(self.icount)
torch.save(self.model.state_dict(), os.path.join(self.dn_save, filename))

# Enable generic saving of module if using multiple GPU's
if torch.cuda.device_count() > 1:
torch.save(self.model.module.state_dict(), os.path.join(self.dn_save, filename))
else:
torch.save(self.model.state_dict(), os.path.join(self.dn_save, filename))

print('save: {0} (iteration: {1})'.format(filename, self.icount))

def run(self):
Expand All @@ -172,6 +180,11 @@ def run(self):
print('Siamese Change Detection Network (Siamese CDResNet)')
self.model = cscdnet.Model(inc=6, outc=2, corr=False, pretrained=True)

# Run on muliple GPU's if available
if torch.cuda.device_count() > 1:
print("Training with", torch.cuda.device_count(), "GPU's")
model = torch.nn.DataParallel(model)

self.model = self.model.cuda()
self.train()

Expand Down