llama : remove unused vars (#4796)

diff --git a/llama.cpp b/llama.cpp
index 021e79a8f556dff3e6628ffb77229d1f451011ce..91aa3f8e79191d414c7215a35b42fc3cd9fe9a2a 100644 (file)
--- a/llama.cpp
+++ b/llama.cpp
@@ -4997,7 +4997,6 @@ struct llm_build_context {
         struct ggml_cgraph * gf = ggml_new_graph_custom(ctx0, LLAMA_MAX_NODES, false);
 
         const int64_t n_embd_head = hparams.n_embd_head_v;
-        const int64_t n_embd_gqa  = hparams.n_embd_v_gqa();
         GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
 
         const int64_t n_rot = n_embd_head_k / 2;
@@ -5210,7 +5209,6 @@ struct llm_build_context {
         struct ggml_cgraph * gf = ggml_new_graph_custom(ctx0, LLAMA_MAX_NODES, false);
 
         const int64_t n_embd_head = hparams.n_embd_head_v;
-        const int64_t n_embd_gqa  = hparams.n_embd_v_gqa();
         GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
 
         struct ggml_tensor * cur;