Skip to content

test: loosening absolute tolerance for layer norm tests #1581

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
laserkelvin merged 2 commits into from
Apr 28, 2026
Merged

test: loosening absolute tolerance for layer norm tests #1581

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
719fe84
test: loosening absolute tolerance for layer norm tests
laserkelvin Apr 22, 2026
12c5b30
Merge branch 'main' into layernorm-test-precisions
laserkelvin Apr 28, 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
2 changes: 1 addition & 1 deletion test/experimental/nn/symmetry/conftest.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,7 +110,7 @@ def get_rtol_atol(dtype: torch.dtype, scale: float = 1.0) -> tuple[float, float]
elif dtype == torch.float32:
# float32 needs looser tolerances for equivariance tests with
# accumulated numerical operations (matrix multiplications, rotations)
rtol, atol = 1e-4, 1e-4
rtol, atol = 1e-4, 5e-3
Comment thread
laserkelvin marked this conversation as resolved.
elif dtype == torch.float64:
# float64 needs slightly looser tolerances for accumulated numerical
# operations in Wigner D-matrix and rotation calculations
Expand Down
Loading