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

Jump to bottom

[FIX] cpu loss in actions #308

Merged
merged 2 commits into from
Feb 1, 2020
Merged

[FIX] cpu loss in actions #308

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
852f528
fix cpu loss
tkornuta-nvidia Jan 29, 2020
1575143
removed tensor.to(tensor) from final_loss
tkornuta-nvidia Jan 31, 2020
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 examples/start_here/simplest_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@
import nemo

nf = nemo.core.NeuralModuleFactory()

# To use CPU-only do:
# from nemo.core import DeviceType
# nf = nemo.core.NeuralModuleFactory(placement=DeviceType.CPU)
Expand Down
6 changes: 5 additions & 1 deletion nemo/backends/pytorch/actions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1334,7 +1334,11 @@ def train(
final_loss.backward(bps_scale.to(final_loss.get_device()))
# single device (CPU or GPU)
else:
final_loss.backward(bps_scale.to(final_loss.get_device()))
# Fix (workaround?) enabling to backpropagate gradiens on CPUs.
if final_loss.get_device() < 0:
final_loss.backward(bps_scale.to(final_loss))
tkornuta-nvidia marked this conversation as resolved.
Show resolved Hide resolved
else:
final_loss.backward(bps_scale.to(final_loss.get_device()))

batch_counter += 1

Expand Down