Skip to content

[FIX] Move loss and n_items to logits device in fast_cross_entropy_loss loss for multi-GPU support #4063

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
danielhanchen merged 2 commits into from
Feb 15, 2026
Merged

[FIX] Move loss and n_items to logits device in fast_cross_entropy_loss loss for multi-GPU support #4063

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
74a349e
bug fix for multi-GPU
nole69 Feb 15, 2026
5f68761
Apply suggestion from @gemini-code-assist[bot]
nole69 Feb 15, 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
3 changes: 3 additions & 0 deletions unsloth/kernels/cross_entropy_loss.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -438,6 +438,7 @@ def fast_cross_entropy_loss(
batch, seq_len, d = logits.shape
assert labels.shape == (batch, seq_len)

device = logits.device
loss = Fast_CrossEntropyLoss.apply(
logits.view(batch * seq_len, d),
labels.view(-1),
Expand All @@ -446,6 +447,8 @@ def fast_cross_entropy_loss(
)
if n_items is None:
n_items = torch.count_nonzero(labels != -100)
if torch.is_tensor(n_items):
n_items = n_items.to(device)
return loss.sum() / n_items


Expand Down