PR #27383 optimized qmoe code path for 1 token

optimized qmoe code path for 1 token #27383

guschmue merged 7 commits into main from gs/wgpu-qmoe-opt

add weight_index_indirect to nbitmm

acabc516

optimized qmoe code pass for 1 token

41701c8d

missing file

7e4b875b

add missing file

aa177a12

Merge branch 'main' into gs/wgpu-qmoe-opt

327763ee

guschmue added ep:WebGPU

guschmue requested a review from

copilot-pull-request-reviewer 147 days ago

copilot-pull-request-reviewer commented on 2026-02-18

Update onnxruntime/contrib_ops/webgpu/moe/gate_1token.wgsl.template

89c648a5

Update onnxruntime/contrib_ops/webgpu/moe/final_mix.wgsl.template

957e65fc

guschmue marked this pull request as ready for review 146 days ago

fs-eire approved these changes on 2026-02-26

guschmue merged 5f087c41 into main 140 days ago

guschmue deleted the gs/wgpu-qmoe-opt branch 140 days ago

Reviewers

fs-eire

Assignees

No one assigned

Labels

ep:WebGPU

Milestone

No milestone