onnxruntime
[CUDA] Fix packed-INT QMoE expert-counting capture and consume
#32942

Merged

[CUDA] Fix packed-INT QMoE expert-counting capture and consume #32942

aciddelgado merged 3 commits into main from aciddelgado/qmoe_packed_int_expert_counting

moe routing record errror fix

0833f3e5

Test packed-INT CUDA QMoE expert counting

0247939f

aciddelgado requested a review from

copilot-pull-request-reviewer 38 days ago

copilot-pull-request-reviewer commented on 2026-09-29

tianleiwu dismissed these changes on 2026-09-29

aciddelgado enabled auto-merge (squash) 38 days ago

Merge remote-tracking branch 'origin/main' into aciddelgado/qmoe_pack…

d2b730f6

copilot-swe-agent dismissed their stale review via d2b730f6 38 days ago

prathikr approved these changes on 2026-09-29

baijumeswani approved these changes on 2026-09-29

aciddelgado merged 74810e56 into main 38 days ago

aciddelgado deleted the aciddelgado/qmoe_packed_int_expert_counting branch 38 days ago

Reviewers

baijumeswani

prathikr

tianleiwu

Assignees

No one assigned

Labels

None yet

Milestone

No milestone