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

Commits
  • [Bugfix][MiniMax-M3] Stabilize FP8 query CUDA graph allocation
    kyleliang-nv committed 93 days ago
  • Merge branch 'main' into agent/minimax-m3-query-fp8-cudagraph
    pavanimajety committed 91 days ago
Loading