Fix CoreML iOS26 numerics in static attention #16144
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16144
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (1 Unrelated Failure)As of commit 83c8d50 with merge base c9f6df1 ( UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
|
@metascroy has exported this pull request. If you are a Meta employee, you can view the originating Diff in D88705980. |
This PR needs a
|
Summary: This diff decomposes SDPA to fix iOS26 numerics in Core ML. It also removes repeat interleave to further optimize performance on Core ML by about 10-15%, depending on the hardware. Differential Revision: D88705980
metascroy
force-pushed
the
export-D88705980
branch
from
18009fd to
d7dd45d
Compare
metascroy
changed the title
Summary: This diff decomposes SDPA to fix iOS26 numerics in Core ML. It also removes repeat interleave to further optimize performance on Core ML by about 10-15%, depending on the hardware. Reviewed By: billmguo Differential Revision: D88705980
metascroy
force-pushed
the
export-D88705980
branch
from
d7dd45d to
d237c0e
Compare
Summary: This diff decomposes SDPA to fix iOS26 numerics in Core ML. It also removes repeat interleave to further optimize performance on Core ML by about 10-15%, depending on the hardware. Reviewed By: billmguo Differential Revision: D88705980
metascroy
force-pushed
the
export-D88705980
branch
from
d237c0e to
83c8d50
Compare
2 participants
Summary:
This diff decomposes SDPA to fix iOS26 numerics in Core ML.
It also removes repeat interleave to further optimize performance on Core ML by about 10-15%, depending on the hardware.
Differential Revision: D88705980