Skip to content

Update Executorch ops registration for rms_norm #9920

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
merged 2 commits into from
Apr 7, 2025
Merged

Update Executorch ops registration for rms_norm #9920

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e3ef2c7
Update Executorch ops registration for rms_norm
Vysarat Apr 4, 2025
185f7b0
Merge branch 'main' into export-D72485973
Vysarat Apr 7, 2025
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
18 changes: 9 additions & 9 deletions backends/cadence/aot/ops_registrations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -293,6 +293,15 @@
"attention_mask.out(Tensor input, Tensor start, Tensor stop, *, Tensor(a!) out) -> Tensor(a!)"
)

# Custom ops in aten namespace. RMSNorm is usually decomposed, so having
# an out-variant is non-standard

lib_aten = Library("aten", "FRAGMENT")

lib_aten.define(
"rms_norm.out(Tensor input, SymInt[] normalized_shape, Tensor? weight=None, float? eps=None, *, Tensor(a!) out) -> Tensor(a!)"
)


@register_fake("cadence::quantize_per_tensor")
def quantize_per_tensor_meta(
Expand Down Expand Up @@ -619,15 +628,6 @@ def linalg_vector_norm_meta(
return X.new_empty([], dtype=X.dtype)


@register_fake("cadence::rms_norm")
def rms_norm_meta(
X: torch.Tensor,
eps: float,
weight: torch.Tensor,
) -> torch.Tensor:
return X.new_empty(X.shape, dtype=X.dtype)


@register_fake("cadence::requantize")
def requantize_meta(
input: torch.Tensor,
Expand Down
Loading