Average pooling clamped divisor should be done on all conditions where the kernel can go out of bounds#4144

Open

ivangarcia44 wants to merge 12 commits intollvm:mainfrom

ivangarcia44:avgPoolClampDivisorRuleUpdate

Commits on Apr 17, 2025

Average pooling clamped divisor should be done on all conditions where the kernel can go out of bounds.
Ivan Garcia
committed

Commits on Apr 23, 2025

Updated divisor algorithm after finding SWA in existing logic (prior any change of mine).
Ivan Garcia
committed

Commits on Apr 24, 2025

Commits on Apr 29, 2025

Merging with Vivek's change.
Ivan Garcia
committed
Merging with Vivek's change.
Ivan Garcia
committed

Commits on May 2, 2025

Addressing round 2 of Vivek's feedback.
Ivan Garcia
committed

Commits on May 3, 2025

Bring back PyTorch based average pooling divisor computation after a couple of corrections.
Ivan Garcia
committed

Commits on May 12, 2025

Addressing saha3's feedback. Typo correction and removing redundant test.
Ivan Garcia
committed

Commits on May 19, 2025

Update lib/Conversion/TorchToLinalg/Pooling.cpp

ivangarcia44
and
vivekkhandelwal1
authored

Commits on May 21, 2025

Addressing Vivek's feedback: removing unnecessary parameter reversal by correcting order of spatial dimensions and adding comment suggested by Vivek.
ivangarcia44
committed