vllm
[Bugfix][MiniMax-M3] Keep FP8 query allocation stable across CUDA graph replay
#51203
Merged

[Bugfix][MiniMax-M3] Keep FP8 query allocation stable across CUDA graph replay #51203

kyleliang-nv
github-actions
mergify mergify added nvidia
mergify mergify added bug
xinli-sw
xinli-sw commented on 2026-08-05
kyleliang-nv kyleliang-nv marked this pull request as ready for review 103 days ago
kyleliang-nv kyleliang-nv requested a review from mgoin mgoin 103 days ago
kyleliang-nv kyleliang-nv requested a review from tlrmchlsmth tlrmchlsmth 103 days ago
kyleliang-nv kyleliang-nv requested a review from WoosukKwon WoosukKwon 103 days ago
kyleliang-nv kyleliang-nv requested a review from yewentao256 yewentao256 103 days ago
kyleliang-nv kyleliang-nv requested a review from zyongye zyongye 103 days ago
kyleliang-nv kyleliang-nv requested a review from AndreasKaratzas AndreasKaratzas 103 days ago
claude
claude commented on 2026-08-06
kyleliang-nv [Bugfix][MiniMax-M3] Stabilize FP8 query CUDA graph allocation
bc9471ed
kyleliang-nv kyleliang-nv force pushed to bc9471ed 92 days ago
mergify mergify added minimax
pavanimajety pavanimajety added verified
pavanimajety Merge branch 'main' into agent/minimax-m3-query-fp8-cudagraph
ac244c56
pavanimajety
github-actions
xinli-sw
pavanimajety
pavanimajety approved these changes on 2026-08-20
pavanimajety pavanimajety merged 7c8b68b9 into main 89 days ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone