diff --git a/llama.cpp b/llama.cpp
index 45da226b1..650c043ef 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -8160,7 +8160,6 @@ struct llm_build_context {
                 inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
             }
             struct ggml_tensor * attn_out = cur;
-            // only used for non-parallel residual
             struct ggml_tensor * ffn_inp  = ggml_add(ctx0, attn_out, inpL);
             cb(cur, "ffn_inp", il);