onnxruntime
[CUDA] Support attention_bias in GroupQueryAttention via the unfused path
#29525

Merged

[CUDA] Support attention_bias in GroupQueryAttention via the unfused path #29525

tianleiwu merged 5 commits into microsoft:main from jiangzhuo:gqa-attention-bias-cuda

[CUDA] Support attention_bias in GroupQueryAttention via the unfused …

5479bb8b

jiangzhuo requested a review from

copilot-pull-request-reviewer 9 days ago

copilot-pull-request-reviewer commented on 2026-07-03

Address review: consistent symbolic dim name, wrap long line

47e935c7

tianleiwu requested a review from

copilot-pull-request-reviewer 8 days ago

copilot-pull-request-reviewer commented on 2026-07-03

tianleiwu commented on 2026-07-03

Address review: bias rank guard, CUDA skip gate, broadcast-flag test …

254da447

github-advanced-security commented on 2026-07-04

Lint: RUF005 tuple unpacking, ruff-format line wrap

6436b286

tianleiwu dismissed these changes on 2026-07-05

Add C++ attention_bias parity test (CUDA vs CPU)

b3d4a795

jiangzhuo dismissed their stale review via b3d4a795 5 days ago

jiangzhuo requested a review from

tianleiwu 4 days ago

tianleiwu approved these changes on 2026-07-08

tianleiwu enabled auto-merge (squash) 4 days ago

tianleiwu merged 2cd57057 into main 4 days ago

Reviewers

tianleiwu

Assignees

No one assigned

Labels

None yet

Milestone

No milestone