vllm
7c8b68b9
- [Bugfix][MiniMax-M3] Keep FP8 query allocation stable across CUDA graph replay (#51203)
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
89 days ago
[Bugfix][MiniMax-M3] Keep FP8 query allocation stable across CUDA graph replay (#51203) Signed-off-by: Kyle Liang <kylliang@nvidia.com>
References
#51203 - [Bugfix][MiniMax-M3] Keep FP8 query allocation stable across CUDA graph replay
Author
kyleliang-nv
Parents
00f7f258
Loading