Skip to content

fix resnet50 loss print error #409

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

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

fix resnet50 loss print error #409

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
00ec158
fix resnet50 loss print error
cccddd77 May 28, 2024
afe49a3
fix loss error on eager mode for multi-devices.
cccddd77 May 29, 2024
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
1 change: 1 addition & 0 deletions Vision/classification/image/resnet50/examples/train_eager_consistent.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,3 +45,4 @@ python3 -m oneflow.distributed.launch \
--num-epochs $EPOCH \
--train-batch-size $TRAIN_BATCH_SIZE \
--val-batch-size $VAL_BATCH_SIZE \
--scale-grad
3 changes: 2 additions & 1 deletion Vision/classification/image/resnet50/train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -257,6 +257,7 @@ def train_one_epoch(self):
else:
top1_acc = 0

loss = loss * self.world_size
self.meter_train_iter(loss, top1_acc)

self.cur_batch += 1
Expand All @@ -269,7 +270,7 @@ def train_eager(self):
if loss.is_global and self.scale_grad:
# NOTE(zwx): scale init grad with world_size
# because global_tensor.mean() include dividor numel * world_size
loss = loss / self.world_size
# loss = loss / self.world_size
loss.backward()
for param_group in self.optimizer.param_groups:
for param in param_group.parameters:
Expand Down