From f32b4ecd26ffc5c69a8a5a9d4b20aa3287efabcc Mon Sep 17 00:00:00 2001 From: Mohammad Miadh Angkad <176301910+mmangkad@users.noreply.github.com> Date: Tue, 7 Jul 2026 16:44:01 +0800 Subject: [PATCH] [Docs] Use trtllm_mha for Qwen3.6 B300 (#29964) --- docs_new/src/snippets/autoregressive/qwen36-deployment.jsx | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/docs_new/src/snippets/autoregressive/qwen36-deployment.jsx b/docs_new/src/snippets/autoregressive/qwen36-deployment.jsx index fd8e093de..d825dcca3 100644 --- a/docs_new/src/snippets/autoregressive/qwen36-deployment.jsx +++ b/docs_new/src/snippets/autoregressive/qwen36-deployment.jsx @@ -222,12 +222,9 @@ export const Qwen36Deployment = () => { } } - if (hardware === 'b200') { + if (hardware === 'b200' || hardware === 'b300') { cmd += ` \\\n --attention-backend trtllm_mha`; } - if (hardware === 'b300') { - cmd += ` \\\n --attention-backend flashinfer`; - } if (hwConfig.mem !== undefined) { cmd += ` \\\n --mem-fraction-static ${hwConfig.mem}`; }