From 3714c3ee1a62ed64ac328ec7d699410ad1219150 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sigbj=C3=B8rn=20Skj=C3=A6ret?= <sigbjorn.skjaeret@scala.com>
Date: Fri, 28 Mar 2025 22:13:02 +0100
Subject: [PATCH] llama : fix incorrect Qwen2Moe ffn_moe_out graph callback
 (#12631)

---
 src/llama-model.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index a442abeb..a4f06112 100644
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -6323,7 +6323,7 @@ struct llm_build_qwen2moe : public llm_graph_context {
                         false, 0.0,
                         LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX,
                         il);
-            cb(cur, "ffn_moe_out", il);
+            cb(moe_out, "ffn_moe_out", il);
 
             // FFN shared expert
             {