[V1] [Hybrid] Enable piecewise CUDA Graph for mamba layers (#21194)

Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
This commit is contained in:
Thomas Parnell
2025-07-19 19:27:21 +00:00
committed by GitHub
parent 9f414a12ad
commit 881e3cbe3b
10 changed files with 100 additions and 31 deletions
+1
View File
@@ -4312,6 +4312,7 @@ class CompilationConfig:
self.splitting_ops = [] if self.full_cuda_graph else [
"vllm.unified_attention",
"vllm.unified_attention_with_output",
"vllm.mamba_mixer2",
]