From d87aa806b507bd69959f5d6e317ed312cf834bb6 Mon Sep 17 00:00:00 2001 From: Nico Bosshard Date: Wed, 16 Oct 2024 14:20:20 +0200 Subject: [PATCH] llama : bump max layers from 512 to 1024 --- include/llama.h | 2 +- src/llama.cpp | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/include/llama.h b/include/llama.h index 02bc7f087..dd71f7b0d 100644 --- a/include/llama.h +++ b/include/llama.h @@ -41,7 +41,7 @@ #define LLAMA_FILE_MAGIC_GGSQ 0x67677371u // 'ggsq' #define LLAMA_SESSION_MAGIC LLAMA_FILE_MAGIC_GGSN -#define LLAMA_SESSION_VERSION 9 +#define LLAMA_SESSION_VERSION 10 #define LLAMA_STATE_SEQ_MAGIC LLAMA_FILE_MAGIC_GGSQ #define LLAMA_STATE_SEQ_VERSION 2 diff --git a/src/llama.cpp b/src/llama.cpp index c51b49c56..862a13d87 100644 --- a/src/llama.cpp +++ b/src/llama.cpp @@ -86,7 +86,7 @@ #endif // bump if necessary -#define LLAMA_MAX_LAYERS 512 +#define LLAMA_MAX_LAYERS 1024 #define LLAMA_MAX_EXPERTS 160 // DeepSeekV2 //