onnxruntime
[WebGPU] Add per-head Q/K normalization to PagedAttention
#32808
Merged

[WebGPU] Add per-head Q/K normalization to PagedAttention #32808

aciddelgado merged 22 commits into main from aciddelgado/paged_qknorm_webgpu
aciddelgado
aciddelgado paged attention noncausal webgpu
7b8d9ffd
aciddelgado Update onnxruntime/test/python/transformers/test_paged_attention.py
cdcbd14e
aciddelgado fix
5f79b82a
aciddelgado Merge branch 'aciddelgado/qwen38_webgpu' of https://github.com/micros…
afc54ec6
impl
f39e9383
prathikr copilot
93b42744
prathikr lint
0c6f7b7c
docs
42faf915
aciddelgado Apply suggestion from @github-actions[bot]
e7ba8319
aciddelgado Enforce is_causal attribute values in PagedAttention
3d281cba
docs
217287d7
Merge remote-tracking branch 'origin' into prathikrao/gateddeltanet-s…
6d78be3d
Copilot Merge prathikrao/gateddeltanet-support-stateupdatecapacity into acidd…
01a9611b
aciddelgado noncausal local
3b808a63
aciddelgado webgpu paged attention qknorm
00cdaa24
aciddelgado aciddelgado requested a review from copilot-pull-request-reviewer copilot-pull-request-reviewer 23 days ago
aciddelgado aciddelgado changed the title Aciddelgado/paged qknorm webgpu [WebGPU] Add per-head Q/K normalization to PagedAttention 23 days ago
copilot-pull-request-reviewer
copilot-pull-request-reviewer commented on 2026-09-25
aciddelgado Merge main into aciddelgado/paged_qknorm_webgpu
052bf660
aciddelgado aciddelgado requested a review from copilot-pull-request-reviewer copilot-pull-request-reviewer 19 days ago
copilot-pull-request-reviewer
copilot-pull-request-reviewer commented on 2026-09-29
aciddelgado
Copilot Merge remote-tracking branch 'origin/main' into aciddelgado/paged_qkn…
5f5ccf53
copilot-swe-agent
aciddelgado moe routing record errror fix
1f57e92a
baijumeswani
baijumeswani commented on 2026-09-29
aciddelgado add test case
886d4dd0
aciddelgado Remove CUDA QMoE repair from WebGPU Q/K-norm PR
2b047241
aciddelgado aciddelgado enabled auto-merge (squash) 19 days ago
aciddelgado Merge main into aciddelgado/paged_qknorm_webgpu
6539bc50
baijumeswani
baijumeswani commented on 2026-09-30
aciddelgado cpu reference test
bd73a0c0
baijumeswani
baijumeswani approved these changes on 2026-10-02
aciddelgado aciddelgado merged c971ca26 into main 17 days ago
aciddelgado aciddelgado deleted the aciddelgado/paged_qknorm_webgpu branch 17 days ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone