chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Tiny fix for fp8 moe backend flashinfer_trtllm naming (#18243)

Browse Source
This commit is contained in:
Jincong Chen
2026-02-04 19:58:04 +08:00
committed by GitHub
parent ce02df8592
commit a72f4f839c
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
python/sglang/srt/layers/quantization/fp8_utils.py
View File

@@ -144,7 +144,7 @@ class Fp8GemmRunnerBackend(Enum):
def is_auto(self) -> bool:
return self == Fp8GemmRunnerBackend.AUTO
def is_flashinfer(self) -> bool:
def is_flashinfer_trtllm(self) -> bool:
return self == Fp8GemmRunnerBackend.FLASHINFER_TRTLLM
def is_flashinfer_deepgemm(self) -> bool:
@@ -199,7 +199,7 @@ def dispatch_w8a8_block_fp8_linear() -> Callable:
def _dispatch_explicit_backend(backend: Fp8GemmRunnerBackend) -> Callable:
"""Dispatch based on explicitly selected backend."""
if backend.is_flashinfer():
if backend.is_flashinfer_trtllm():
if not (is_blackwell_supported() and is_flashinfer_available()):
raise RuntimeError(
"FlashInfer FP8 GEMM requested via --fp8-gemm-backend=flashinfer_trtllm, "