From 375de5b1f8c07b5bfdef7f00b738eb176f8431ba Mon Sep 17 00:00:00 2001
From: Francis Couture-Harpin <git@compilade.net>
Date: Sun, 1 Sep 2024 21:59:24 -0400
Subject: [PATCH] llama : use unused n_embd_k_gqa in k_shift

This also slightly reduces the diff from the master branch
---
 src/llama.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/llama.cpp b/src/llama.cpp
index cf7dccb384f2b..043f3d7ec7853 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -10806,7 +10806,7 @@ struct llm_build_context {
                         ggml_view_3d(ctx0, kv_self.k_l[il],
                             n_embd_head_k, n_head_kv, n_ctx,
                             ggml_row_size(kv_self.k_l[il]->type, n_embd_head_k),
-                            ggml_row_size(kv_self.k_l[il]->type, hparams.n_embd_k_gqa(il)),
+                            ggml_row_size(kv_self.k_l[il]->type, n_embd_k_gqa),
                             0),
                         lctx.inp_K_shift, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
                         ext_factor, attn_factor, beta_fast, beta_slow);