From a72f4f839c4dd0a7cab88f563c8e47dec01a2cf2 Mon Sep 17 00:00:00 2001 From: Jincong Chen Date: Wed, 4 Feb 2026 19:58:04 +0800 Subject: [PATCH] Tiny fix for fp8 moe backend flashinfer_trtllm naming (#18243) --- python/sglang/srt/layers/quantization/fp8_utils.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/python/sglang/srt/layers/quantization/fp8_utils.py b/python/sglang/srt/layers/quantization/fp8_utils.py index a03900549..3b9616e27 100644 --- a/python/sglang/srt/layers/quantization/fp8_utils.py +++ b/python/sglang/srt/layers/quantization/fp8_utils.py @@ -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, "