jeans/ollama
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Quiet down llama.cpp logging by default

Browse Source 
By default builds will now produce non-debug and non-verbose binaries.
To enable verbose logs in llama.cpp and debug symbols in the
native code, set `CGO_CFLAGS=-g`
This commit is contained in:
Daniel Hiltgen
2023-12-22 08:47:18 -08:00
parent 96fb441abd
commit e5202eb687
5 changed files with 25 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
llm/llama.cpp/patches/0001-Expose-callable-API-for-server.patch
View File

@@ -1,22 +1,22 @@
From 4c72576c5f6c2217b1ecf7fd8523616acc5526ae Mon Sep 17 00:00:00 2001
From 90c332fe2ef61149b38561d02836e66715df214d Mon Sep 17 00:00:00 2001
From: Daniel Hiltgen <daniel@ollama.com>
Date: Mon, 13 Nov 2023 12:25:58 -0800
Subject: [PATCH] Expose callable API for server
This adds an extern "C" interface within the example server
---
examples/server/CMakeLists.txt | 24 +++
examples/server/server.cpp | 279 +++++++++++++++++++++++++++++++++
examples/server/CMakeLists.txt | 27 ++++
examples/server/server.cpp | 280 +++++++++++++++++++++++++++++++++
examples/server/server.h | 89 +++++++++++
ggml-cuda.cu | 1 +
4 files changed, 393 insertions(+)
4 files changed, 397 insertions(+)
create mode 100644 examples/server/server.h
diff --git a/examples/server/CMakeLists.txt b/examples/server/CMakeLists.txt
index 859cd12..4ea47a7 100644
index 859cd12..da2b9bf 100644
--- a/examples/server/CMakeLists.txt
+++ b/examples/server/CMakeLists.txt
@@ -11,3 +11,27 @@ if (WIN32)
@@ -11,3 +11,30 @@ if (WIN32)
TARGET_LINK_LIBRARIES(${TARGET} PRIVATE ws2_32)
endif()
target_compile_features(${TARGET} PRIVATE cxx_std_11)
@@ -29,6 +29,9 @@ index 859cd12..4ea47a7 100644
+target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_definitions(${TARGET} PUBLIC LLAMA_SERVER_LIBRARY=1)
+target_link_libraries(${TARGET} PRIVATE common llama llava ${CMAKE_THREAD_LIBS_INIT})
+target_compile_definitions(${TARGET} PRIVATE
+ SERVER_VERBOSE=$<BOOL:${LLAMA_SERVER_VERBOSE}>
+)
+
+if (BUILD_SHARED_LIBS)
+ set_target_properties(ext_server PROPERTIES POSITION_INDEPENDENT_CODE ON)
@@ -46,7 +49,7 @@ index 859cd12..4ea47a7 100644
+endif()
\ No newline at end of file
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index 0403853..5e78e4d 100644
index 0403853..07fb05c 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -5,6 +5,9 @@
@@ -67,7 +70,7 @@ index 0403853..5e78e4d 100644
int main(int argc, char **argv)
{
#if SERVER_VERBOSE != 1
@@ -3123,3 +3127,278 @@ int main(int argc, char **argv)
@@ -3123,3 +3127,279 @@ int main(int argc, char **argv)
llama_backend_free();
return 0;
}
@@ -81,6 +84,7 @@ index 0403853..5e78e4d 100644
+void llama_server_init(ext_server_params *sparams, ext_server_resp_t *err)
+{
+#if SERVER_VERBOSE != 1
+ LOG_TEE("disabling verbose llm logging\n");
+ log_disable();
+#endif
+ assert(err != NULL && sparams != NULL);