Skip to content

fix(python): avoid CPU bf16 compile flake #7587

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
wjones127 merged 1 commit into from
Jul 6, 2026
Merged

fix(python): avoid CPU bf16 compile flake #7587

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
5 changes: 3 additions & 2 deletions python/python/lance/torch/distance.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,8 +150,9 @@ def pairwise_l2(
if x.dtype != y.dtype or (y2 is not None and x.dtype != y2.dtype):
raise ValueError("pairwise_l2 data types do not match")
origin_dtype = x.dtype
if x.device == torch.device("cpu") and x.dtype == torch.float16:
# Pytorch does not support `x @ y.T` for float16 on CPU
if x.device == torch.device("cpu") and x.dtype in (torch.float16, torch.bfloat16):
# Pytorch does not support `x @ y.T` for float16 on CPU, and bfloat16
# can hit backend-specific compiler paths. Accumulate in float32.
x = x.type(torch.float32)
y = y.type(torch.float32)
if y2 is not None:
Expand Down
Loading