PR #17947 Enable Past and Present Tensors for QAttention

Enable Past and Present Tensors for QAttention #17947

AnaghaRaoAMD merged 2 commits into DmlPrototype from user/anagrao/QAttention

AnaghaRaoAMD force pushed from 5955269c to e894e957 2 years ago

AnaghaRaoAMD force pushed from e894e957 to ad9be949 2 years ago

AnaghaRaoAMD force pushed from ad9be949 to 44f1c636 2 years ago

AnaghaRaoAMD force pushed from 44f1c636 to dc1faac6 2 years ago

AnaghaRaoAMD force pushed from dc1faac6 to 32e9744b 2 years ago

AnaghaRaoAMD force pushed from 32e9744b to 81aeea72 2 years ago

AnaghaRaoAMD marked this pull request as ready for review 2 years ago

AnaghaRaoAMD requested a review from

PatriceVignola 2 years ago

AnaghaRaoAMD requested a review from

sumitsays 2 years ago

AnaghaRaoAMD requested a review from

zhangxiang1993 2 years ago

zhangxiang1993 approved these changes on 2023-10-19

sumitsays commented on 2023-10-19

PatriceVignola commented on 2023-10-19

[REview comments] Remove qkvhiddensize, disable support for past_pres…

1bf58756

AnaghaRaoAMD force pushed from 81aeea72 to 1bf58756 2 years ago

sumitsays commented on 2023-10-24

[Review Comments] Update typo and clena up

83a50694

sumitsays approved these changes on 2023-10-24

AnaghaRaoAMD merged fcb48ae2 into DmlPrototype 2 years ago

AnaghaRaoAMD deleted the user/anagrao/QAttention branch 2 years ago

Reviewers

sumitsays

zhangxiang1993

PatriceVignola

Assignees

No one assigned

Labels

None yet

Milestone

No milestone