Skip to content

Attach empty grad to its param to ensure it's copied after reduction #316

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

Merged
jeffra merged 1 commit into from
Aug 13, 2020
Merged

Attach empty grad to its param to ensure it's copied after reduction #316

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
8 changes: 4 additions & 4 deletions deepspeed/pt/deepspeed_light.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1013,10 +1013,10 @@ def buffered_allreduce_fallback(self, grads=None, elements_per_buffer=500000000)
# rank is reducing the same size. In some cases it may make
# sense in the future to support the ability to average not
# w.r.t. world size but with a different value.
grads.append(
torch.zeros(param.size(),
dtype=param.dtype,
device=param.device))
param.grad = torch.zeros(param.size(),
dtype=param.dtype,
device=param.device)
grads.append(param.grad.data)
else:
grad_data = param.grad.data
if self.sparse_gradients_enabled(
Expand Down