From bdb3b45e02c2c50fa5924210ebdd3ec4247eff38 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Luka=20Govedi=C4=8D?= Date: Wed, 29 Apr 2026 16:03:22 -0400 Subject: [PATCH] Apply suggestion from @ProExpertProg MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Luka Govedič --- vllm/config/kernel.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/vllm/config/kernel.py b/vllm/config/kernel.py index 56c29ee182e..0e8f13b1937 100644 --- a/vllm/config/kernel.py +++ b/vllm/config/kernel.py @@ -145,8 +145,8 @@ class KernelConfig: - "cutlass": Use vLLM CUTLASS kernels - "flashinfer_trtllm": Use FlashInfer with TRTLLM-GEN kernels (BF16 input) - "flashinfer_trtllm_afp8": Use FlashInfer with TRTLLM-GEN kernels (FP8 input) - - "flashinfer_cutlass": Use FlashInfer with CUTLASS kernels (BF16 output) - - "flashinfer_cutlass_afp8": Use FlashInfer with CUTLASS kernels (FP8 output) + - "flashinfer_cutlass": Use FlashInfer with CUTLASS kernels (BF16 input) + - "flashinfer_cutlass_afp8": Use FlashInfer with CUTLASS kernels (FP8 input) - "flashinfer_cutedsl": Use FlashInfer with CuteDSL kernels (FP4 only) - "marlin": Use Marlin kernels (weight-only quantization) - "aiter": Use AMD AITer kernels (ROCm only)