From 333e27b31fba031fd011dea83e968e8227fcbc1b Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 25 Aug 2023 11:54:57 +0300
Subject: [PATCH] falcon : do not offload the embeddings layer

---
 llama.cpp | 2 --
 1 file changed, 2 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index dfe7c281a..a40ed034d 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -2756,8 +2756,6 @@ static struct ggml_cgraph * llm_build_falcon(
         cur = ggml_add(ctx0,
                 ggml_mul(ctx0, cur, model.output_norm),
                 model.output_norm_b);
-        offload_func_nr(cur->src[0]);
-        offload_func_nr(cur);
         ggml_set_name(cur, "result_norm");
     }