llama : fix incorrect Qwen2Moe ffn_moe_out graph callback (#12631)

diff --git a/src/llama-model.cpp b/src/llama-model.cpp
index a442abeb85392f58db38386012e4e323e8cef75a..a4f06112d2842547afc5219a99cb43baeaa5535b 100644 (file)
--- 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
             {