chore: fix llama.cpp patch

mybigday · Nov 2, 2024 · 05b13a4 · 05b13a4
1 parent a61a857
commit 05b13a4
Showing 1 changed file with 3 additions and 3 deletions.
diff --git a/scripts/llama.cpp.patch b/scripts/llama.cpp.patch
@@ -1,5 +1,5 @@
---- llama.cpp.orig	2024-11-02 11:13:58
-+++ llama.cpp	2024-11-02 11:19:21
+--- llama.cpp.orig	2024-11-02 12:42:13
++++ llama.cpp	2024-11-02 13:00:37
 @@ -80,6 +80,17 @@
  #define LLAMA_MAX_LAYERS  512
  #define LLAMA_MAX_EXPERTS 160  // DeepSeekV2
@@ -41,7 +41,7 @@
          }
 @@ -19086,7 +19097,9 @@
 
- #ifdef GGML_USE_METAL
+ #ifdef LM_GGML_USE_METAL
      // note: we usually have plenty of VRAM, so by default offload all layers to the GPU
 -    result.n_gpu_layers = 999;
 +    if (result.n_gpu_layers > 0) {