From 3b87a9e8ae87ee998b98954b0813348ce6f34a78 Mon Sep 17 00:00:00 2001
From: Yuan Luo <yuan.luo@hotmail.com>
Date: Mon, 4 Aug 2025 11:05:02 +0800
Subject: [PATCH] Fix bug of refactoring TopKOutput in w4afp8 (#8745)

Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
---
 python/sglang/srt/layers/quantization/w4afp8.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/sglang/srt/layers/quantization/w4afp8.py b/python/sglang/srt/layers/quantization/w4afp8.py
index 8619c042b..ba11a4b6e 100644
--- a/python/sglang/srt/layers/quantization/w4afp8.py
+++ b/python/sglang/srt/layers/quantization/w4afp8.py
@@ -282,7 +282,7 @@ class W4AFp8MoEMethod(FusedMoEMethodBase):
         # TODO(ch-wan): move it out of this class
         from sglang.srt.layers.moe.cutlass_w4a8_moe import cutlass_w4a8_moe
 
-        topk_ids, topk_weights, _ = topk_output
+        topk_weights, topk_ids, _ = topk_output
         local_topk_ids = topk_ids
         if layer.expert_map is not None:
             "Translate info from expert_map to topk_ids"