onnxruntime
Support boolean attention mask in Attention(23) CUDA - MHA case
#27428

Merged

Support boolean attention mask in Attention(23) CUDA - MHA case #27428

justinchuby merged 7 commits into main from titaiwang/suuport_book_attn_mask_mha

draft

8783e562

tests

5283ea05

titaiwangms requested a review from

tianleiwu 17 days ago

titaiwangms requested a review from

copilot-pull-request-reviewer 17 days ago

titaiwangms added ep:CUDA

copilot-pull-request-reviewer commented on 2026-02-24

fix half overflow

6fb6f45a

titaiwangms requested a review from

xadupre 17 days ago

fix half again

3e2bd312

Merge branch 'main' into titaiwang/suuport_book_attn_mask_mha

054c015b

update mha tests with bool attn_mask

dd2ff165

justinchuby requested a review from

justinchuby 14 days ago

address reviews - 1

6e2f9322

justinchuby approved these changes on 2026-02-27

justinchuby merged bf71213a into main 14 days ago

justinchuby deleted the titaiwang/suuport_book_attn_mask_mha branch 14 days ago

Reviewers

justinchuby

tianleiwu

xadupre

Assignees

No one assigned

Labels

ep:CUDA

Milestone

No milestone