Skip to content

fix(datasets): build the DP all_reduce tensor on the mesh device instead of assuming CUDA #2127

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

fix(datasets): build the DP all_reduce tensor on the mesh device instead of assuming CUDA #2127

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
12f3d93
fix(datasets): build the DP all_reduce tensor on the mesh device
li-lizhe Sep 25, 2026
6728cb1
Merge branch 'main' into fix/dp-allreduce-device
li-lizhe Oct 31, 2026
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
9 changes: 5 additions & 4 deletions xtuner/v1/datasets/dataloader.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,10 +28,11 @@ def reduce_sum_across_dp_group(dp_mesh: DeviceMesh | None, local_value: int) ->
return int(local_value)
if not dist.is_available() or not dist.is_initialized():
return int(local_value)
if torch.cuda.is_available():
device = torch.device(f"cuda:{torch.cuda.current_device()}")
else:
device = torch.device("cpu")
# ``dp_mesh`` is guaranteed to be set here (the early return above handles
# ``None``), and it already carries the device type of the mesh the
# collective runs on (cuda / npu / xpu / ...), so use it instead of
# assuming CUDA.
device = torch.device(dp_mesh.device_type)
tensor = torch.tensor([local_value], dtype=torch.int64, device=device)
dist.all_reduce(tensor, op=dist.ReduceOp.SUM, group=dp_mesh.get_group())
return int(tensor.item())
Expand Down