vllm
7c8b68b9 - [Bugfix][MiniMax-M3] Keep FP8 query allocation stable across CUDA graph replay (#51203)

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>
Author
Parents
Loading