Skip to content

Fix bug in per tensor mul and add impls (#18481) #18481

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
DrJessop wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix bug in per tensor mul and add impls (#18481) #18481

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
18 changes: 4 additions & 14 deletions backends/cadence/aot/ref_implementations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -318,14 +318,8 @@ def quantized_add_per_tensor(
f"X and Y dtypes need to be in {supported_dtypes}. Got {dtype}"
)

if dtype == torch.uint8:
X = X.to(torch.int8)
Y = Y.to(torch.int8)

# TODO(agrebenisan): This should be done in fixed point arithmetic, but to match the quantized_add_out.cpp
# reference implementation, we'll do it in floating point.
dequant_X = X_scale * (X - X_zero_point)
dequant_Y = Y_scale * (Y - Y_zero_point)
dequant_X = X_scale * (X.float() - X_zero_point)
dequant_Y = Y_scale * (Y.float() - Y_zero_point)

# q_min/q_max are unused args
return quantize_per_tensor(
Expand Down Expand Up @@ -447,12 +441,8 @@ def quantized_mul_per_tensor(
f"X and Y dtypes need to be in {supported_dtypes}. Got {dtype}"
)

if dtype == torch.uint8:
X = X.to(torch.int8)
Y = Y.to(torch.int8)

dequant_X = X_scale * (X - X_zero_point)
dequant_Y = Y_scale * (Y - Y_zero_point)
dequant_X = X_scale * (X.float() - X_zero_point)
dequant_Y = Y_scale * (Y.float() - Y_zero_point)

return quantize_per_tensor(
dequant_X * dequant_Y,
Expand Down
Loading