xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-04 00:49:10 +08:00
Code Issues Packages Projects Releases Wiki Activity

[LoRA] Set default MXFP4 LoRA backend to Marlin (#30598)

Browse Source 
Signed-off-by: Xin Yang <xyangx@amazon.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
This commit is contained in:
Xin Yang 2025-12-18 08:42:22 -08:00 committed by GitHub
parent 326e7c3105
commit 9a5e96523b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
vllm/model_executor/layers/quantization/mxfp4.py
View File

@ -95,12 +95,12 @@ def get_mxfp4_backend_with_lora() -> Mxfp4Backend:
# SM120 needs this fix: https://github.com/triton-lang/triton/pull/8498 # SM120 needs this fix: https://github.com/triton-lang/triton/pull/8498
and (9, 0) <= current_platform.get_device_capability() < (11, 0) and (9, 0) <= current_platform.get_device_capability() < (11, 0)
) )
if envs.VLLM_MXFP4_USE_MARLIN or not triton_kernels_supported: if envs.VLLM_MXFP4_USE_MARLIN is False and triton_kernels_supported:
logger.info_once("[get_mxfp4_backend_with_lora] Using Marlin backend") logger.info_once("[get_mxfp4_backend_with_lora] Using Triton backend")
return Mxfp4Backend.MARLIN return Mxfp4Backend.TRITON
logger.info_once("[get_mxfp4_backend_with_lora] Using Triton backend") logger.info_once("[get_mxfp4_backend_with_lora] Using Marlin backend")
return Mxfp4Backend.TRITON return Mxfp4Backend.MARLIN
def get_mxfp4_backend(with_lora_support: bool) -> Mxfp4Backend: def get_mxfp4_backend(with_lora_support: bool) -> Mxfp4Backend: