xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-11 05:42:15 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] Move flashinfer kernel check into ``__init__` function of `FusedMoE`` (#29018)

Browse Source 
Signed-off-by: Max Hu <hyoung2991@gmail.com>
This commit is contained in:
Max Hu 2025-11-19 16:54:15 -05:00 committed by GitHub
parent 8f4f77a727
commit cb0a7b4bea
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/model_executor/layers/fused_moe/layer.py
View File

@ -574,6 +574,9 @@ class FusedMoE(CustomOp):
is_act_and_mul=is_act_and_mul, is_act_and_mul=is_act_and_mul,
is_lora_enabled=vllm_config.lora_config is not None, is_lora_enabled=vllm_config.lora_config is not None,
) )
self.moe_config_use_flashinfer_cutlass_kernels = (
self.moe_config.use_flashinfer_cutlass_kernels
)
self.quant_config = quant_config self.quant_config = quant_config
@ -728,7 +731,7 @@ class FusedMoE(CustomOp):
return ( return (
self.moe_quant_config is not None self.moe_quant_config is not None
and self.moe_quant_config.quant_dtype == "nvfp4" and self.moe_quant_config.quant_dtype == "nvfp4"
and self.moe_config.use_flashinfer_cutlass_kernels and self.moe_config_use_flashinfer_cutlass_kernels
) )
@property @property