merge latest upstream

.github/workflows/build.yml

@@ -23,6 +23,9 @@ env:
   BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
   GGML_NLOOP: 3
   GGML_N_THREADS: 1
+  LLAMA_LOG_COLORS: 1
+  LLAMA_LOG_PREFIX: 1
+  LLAMA_LOG_TIMESTAMPS: 1
 
 jobs:
   macOS-latest-cmake-arm64:
@@ -375,7 +378,7 @@ jobs:
     steps:
       - name: Clone
         id: checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Dependencies
         id: depends
@@ -401,7 +404,7 @@ jobs:
     continue-on-error: true
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
 
       - name: add oneAPI to apt
         shell: bash
@@ -442,7 +445,7 @@ jobs:
     continue-on-error: true
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
 
       - name: add oneAPI to apt
         shell: bash
@@ -546,7 +549,7 @@ jobs:
     steps:
       - name: Clone
         id: checkout
-        uses: actions/checkout@v1
+        uses: actions/checkout@v4
 
       - name: Dependencies
         id: depends
@@ -576,7 +579,7 @@ jobs:
     steps:
       - name: Clone
         id: checkout
-        uses: actions/checkout@v1
+        uses: actions/checkout@v4
 
       - name: Dependencies
         id: depends
@@ -610,7 +613,7 @@ jobs:
     steps:
       - name: Clone
         id: checkout
-        uses: actions/checkout@v1
+        uses: actions/checkout@v4
 
       - name: Dependencies
         id: depends
@@ -969,14 +972,14 @@ jobs:
     steps:
       - name: Clone
         id: checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Install
         id: depends
         run: |
           $ErrorActionPreference = "Stop"
           write-host "Downloading AMD HIP SDK Installer"
-          Invoke-WebRequest -Uri "https://download.amd.com/developer/eula/rocm-hub/AMD-Software-PRO-Edition-23.Q4-WinSvr2022-For-HIP.exe" -OutFile "${env:RUNNER_TEMP}\rocm-install.exe"
+          Invoke-WebRequest -Uri "https://download.amd.com/developer/eula/rocm-hub/AMD-Software-PRO-Edition-24.Q3-WinSvr2022-For-HIP.exe" -OutFile "${env:RUNNER_TEMP}\rocm-install.exe"
           write-host "Installing AMD HIP SDK"
           Start-Process "${env:RUNNER_TEMP}\rocm-install.exe" -ArgumentList '-install' -NoNewWindow -Wait
           write-host "Completed AMD HIP SDK installation"

.github/workflows/server.yml

@@ -20,6 +20,12 @@ on:
     types: [opened, synchronize, reopened]
     paths: ['.github/workflows/server.yml', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.swift', '**/*.m', 'examples/server/**.*']
 
+env:
+  LLAMA_LOG_COLORS: 1
+  LLAMA_LOG_PREFIX: 1
+  LLAMA_LOG_TIMESTAMPS: 1
+  LLAMA_LOG_VERBOSITY: 10
+
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.run_id }}
   cancel-in-progress: true
@@ -173,6 +179,7 @@ jobs:
         if: ${{ !matrix.disabled_on_pr || !github.event.pull_request }}
         run: |
           cd examples/server/tests
+          $env:PYTHONIOENCODING = ":replace"
           behave.exe --summary --stop --no-capture --exclude 'issues|wrong_usages|passkey' --tags llama.cpp
 
       - name: Slow tests

CMakeLists.txt

@@ -82,11 +82,11 @@ set(GGML_FATAL_WARNINGS     ${LLAMA_FATAL_WARNINGS})
 
 # change the default for these ggml options
 if (NOT DEFINED GGML_LLAMAFILE)
-    set(GGML_LLAMAFILE ON)
+    set(GGML_LLAMAFILE_DEFAULT ON)
 endif()
 
-if (NOT DEFINED GGML_CUDA_USE_GRAPHS)
-    set(GGML_CUDA_USE_GRAPHS ON)
+if (NOT DEFINED GGML_CUDA_GRAPHS)
+    set(GGML_CUDA_GRAPHS_DEFAULT ON)
 endif()
 
 # transition helpers
@@ -139,10 +139,16 @@ set(LLAMA_BIN_INSTALL_DIR     ${CMAKE_INSTALL_BINDIR}     CACHE PATH "Location o
 # determining _precisely_ which defines are necessary for the llama-config
 # package.
 #
+set(GGML_TRANSIENT_DEFINES)
 get_target_property(GGML_DIRECTORY ggml SOURCE_DIR)
 get_directory_property(GGML_DIR_DEFINES DIRECTORY ${GGML_DIRECTORY} COMPILE_DEFINITIONS)
+if (GGML_DIR_DEFINES)
+    list(APPEND GGML_TRANSIENT_DEFINES ${GGML_DIR_DEFINES})
+endif()
 get_target_property(GGML_TARGET_DEFINES ggml COMPILE_DEFINITIONS)
-set(GGML_TRANSIENT_DEFINES ${GGML_TARGET_DEFINES} ${GGML_DIR_DEFINES})
+if (GGML_TARGET_DEFINES)
+    list(APPEND GGML_TRANSIENT_DEFINES ${GGML_TARGET_DEFINES})
+endif()
 get_target_property(GGML_LINK_LIBRARIES ggml LINK_LIBRARIES)
 
 set_target_properties(llama PROPERTIES PUBLIC_HEADER ${CMAKE_CURRENT_SOURCE_DIR}/include/llama.h)

Makefile

@@ -54,6 +54,7 @@ TEST_TARGETS = \
 	tests/test-grammar-parser \
 	tests/test-json-schema-to-grammar \
 	tests/test-llama-grammar \
+	tests/test-log \
 	tests/test-model-load-cancel \
 	tests/test-opt \
 	tests/test-quantize-fns \
@@ -148,6 +149,14 @@ GGML_NO_METAL := 1
 DEPRECATE_WARNING := 1
 endif
 
+ifdef LLAMA_DISABLE_LOGS
+REMOVE_WARNING := 1
+endif
+
+ifdef LLAMA_SERVER_VERBOSE
+REMOVE_WARNING := 1
+endif
+
 ifndef UNAME_S
 UNAME_S := $(shell uname -s)
 endif
@@ -351,19 +360,11 @@ ifdef LLAMA_SANITIZE_UNDEFINED
 	MK_LDFLAGS  += -fsanitize=undefined -g
 endif
 
-ifdef LLAMA_SERVER_VERBOSE
-	MK_CPPFLAGS += -DSERVER_VERBOSE=$(LLAMA_SERVER_VERBOSE)
-endif
-
 ifdef LLAMA_SERVER_SSL
 	MK_CPPFLAGS += -DCPPHTTPLIB_OPENSSL_SUPPORT
 	MK_LDFLAGS += -lssl -lcrypto
 endif
 
-ifdef LLAMA_DISABLE_LOGS
-	MK_CPPFLAGS += -DLOG_DISABLE_LOGS
-endif # LLAMA_DISABLE_LOGS
-
 # warnings
 WARN_FLAGS = \
 	-Wall \
@@ -434,7 +435,7 @@ endif
 # TODO: probably these flags need to be tweaked on some architectures
 #       feel free to update the Makefile for your architecture and send a pull request or issue
 
-ifndef RISCV
+ifndef RISCV_CROSS_COMPILE
 
 ifeq ($(UNAME_M),$(filter $(UNAME_M),x86_64 i686 amd64))
 	# Use all CPU extensions that are available:
@@ -514,7 +515,12 @@ ifneq ($(filter loongarch64%,$(UNAME_M)),)
 	MK_CXXFLAGS += -mlasx
 endif
 
-else
+ifneq ($(filter riscv64%,$(UNAME_M)),)
+	MK_CFLAGS   += -march=rv64gcv -mabi=lp64d
+	MK_CXXFLAGS += -march=rv64gcv -mabi=lp64d
+endif
+
+else # RISC-V CROSS COMPILATION
 	MK_CFLAGS   += -march=rv64gcv -mabi=lp64d
 	MK_CXXFLAGS += -march=rv64gcv -mabi=lp64d
 endif
@@ -613,7 +619,7 @@ ifdef GGML_CUDA
 			CUDA_PATH ?= /usr/local/cuda
 		endif
 
-		MK_CPPFLAGS  += -DGGML_USE_CUDA -I$(CUDA_PATH)/include -I$(CUDA_PATH)/targets/$(UNAME_M)-linux/include -DGGML_CUDA_USE_GRAPHS
+		MK_CPPFLAGS  += -DGGML_USE_CUDA -DGGML_CUDA_USE_GRAPHS -I$(CUDA_PATH)/include -I$(CUDA_PATH)/targets/$(UNAME_M)-linux/include
 		MK_LDFLAGS   += -lcuda -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L$(CUDA_PATH)/lib64 -L/usr/lib64 -L$(CUDA_PATH)/targets/$(UNAME_M)-linux/lib -L$(CUDA_PATH)/lib64/stubs -L/usr/lib/wsl/lib
 		MK_NVCCFLAGS += -use_fast_math
 	endif # GGML_MUSA
@@ -925,6 +931,8 @@ OBJ_LLAMA = \
 
 OBJ_COMMON = \
 	common/common.o \
+	common/arg.o \
+	common/log.o \
 	common/console.o \
 	common/ngram-cache.o \
 	common/sampling.o \
@@ -1021,6 +1029,14 @@ $(info   - LLAMA_NO_CCACHE)
 $(info )
 endif
 
+ifdef REMOVE_WARNING
+$(info !!! REMOVAL WARNING !!!)
+$(info The following LLAMA_ options have been removed and are no longer supported)
+$(info   - LLAMA_DISABLE_LOGS   (https://github.com/ggerganov/llama.cpp/pull/9418))
+$(info   - LLAMA_SERVER_VERBOSE (https://github.com/ggerganov/llama.cpp/pull/9418))
+$(info )
+endif
+
 #
 # Build libraries
 #
@@ -1157,6 +1173,16 @@ common/common.o: \
 	include/llama.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
+common/arg.o: \
+	common/arg.cpp \
+	common/arg.h
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
+common/log.o: \
+	common/log.cpp \
+	common/log.h
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
 common/sampling.o: \
 	common/sampling.cpp \
 	common/sampling.h \
@@ -1335,7 +1361,7 @@ llama-cvector-generator: examples/cvector-generator/cvector-generator.cpp \
 	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
 
 llama-convert-llama2c-to-ggml: examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp \
-	$(OBJ_GGML) $(OBJ_LLAMA)
+	$(OBJ_ALL)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
 
@@ -1429,6 +1455,7 @@ llama-server: \
 	examples/server/system-prompts.js.hpp \
 	examples/server/prompt-formats.js.hpp \
 	examples/server/json-schema-to-grammar.mjs.hpp \
+	examples/server/loading.html.hpp \
 	common/json.hpp \
 	common/stb_image.h \
 	$(OBJ_ALL)
@@ -1448,7 +1475,6 @@ llama-gen-docs: examples/gen-docs/gen-docs.cpp \
 	$(OBJ_ALL)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
-	./llama-gen-docs
 
 libllava.a: examples/llava/llava.cpp \
 	examples/llava/llava.h \
@@ -1517,6 +1543,11 @@ tests/test-llama-grammar: tests/test-llama-grammar.cpp \
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
 
+tests/test-log: tests/test-log.cpp \
+	$(OBJ_ALL)
+	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
+	$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
+
 tests/test-grammar-parser: tests/test-grammar-parser.cpp \
 	$(OBJ_ALL)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)

README.md

@@ -17,7 +17,7 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others)
 
 ## Hot topics
 
-- *add hot topics here*
+- Huggingface GGUF editor: [discussion](https://github.com/ggerganov/llama.cpp/discussions/9268) | [tool](https://huggingface.co/spaces/CISCai/gguf-editor)
 
 ----
 
@@ -77,6 +77,7 @@ Typically finetunes of the base models below are supported as well.
 - [x] [SEA-LION](https://huggingface.co/models?search=sea-lion)
 - [x] [GritLM-7B](https://huggingface.co/GritLM/GritLM-7B) + [GritLM-8x7B](https://huggingface.co/GritLM/GritLM-8x7B)
 - [x] [OLMo](https://allenai.org/olmo)
+- [x] [OLMoE](https://huggingface.co/allenai/OLMoE-1B-7B-0924)
 - [x] [Granite models](https://huggingface.co/collections/ibm-granite/granite-code-models-6624c5cec322e4c148c8b330)
 - [x] [GPT-NeoX](https://github.com/EleutherAI/gpt-neox) + [Pythia](https://github.com/EleutherAI/pythia)
 - [x] [Snowflake-Arctic MoE](https://huggingface.co/collections/Snowflake/arctic-66290090abe542894a5ac520)
@@ -89,6 +90,7 @@ Typically finetunes of the base models below are supported as well.
 - [x] [SmolLM](https://huggingface.co/collections/HuggingFaceTB/smollm-6695016cad7167254ce15966)
 - [x] [EXAONE-3.0-7.8B-Instruct](https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct)
 - [x] [FalconMamba Models](https://huggingface.co/collections/tiiuae/falconmamba-7b-66b9a580324dd1598b0f6d4a)
+- [x] [Jais](https://huggingface.co/inceptionai/jais-13b-chat)
 
 (instructions for supporting more models: [HOWTO-add-model.md](./docs/development/HOWTO-add-model.md))
 
@@ -163,6 +165,7 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 - [AI Sublime Text plugin](https://github.com/yaroslavyaroslav/OpenAI-sublime-text) (MIT)
 - [AIKit](https://github.com/sozercan/aikit) (MIT)
 - [LARS - The LLM & Advanced Referencing Solution](https://github.com/abgulati/LARS) (AGPL)
+- [LLMUnity](https://github.com/undreamai/LLMUnity) (MIT)
 
 *(to have a project listed here, it should clearly state that it depends on `llama.cpp`)*
 
@@ -171,6 +174,7 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 - [akx/ggify](https://github.com/akx/ggify) – download PyTorch models from HuggingFace Hub and convert them to GGML
 - [crashr/gppm](https://github.com/crashr/gppm) – launch llama.cpp instances utilizing NVIDIA Tesla P40 or P100 GPUs with reduced idle power consumption
 - [gpustack/gguf-parser](https://github.com/gpustack/gguf-parser-go/tree/main/cmd/gguf-parser) - review/check the GGUF file and estimate the memory usage
+- [Styled Lines](https://marketplace.unity.com/packages/tools/generative-ai/styled-lines-llama-cpp-model-292902) (proprietary licensed, async wrapper of inference part for game development in Unity3d with prebuild Mobile and Web platform wrappers and a model example)
 
 **Infrastructure:**
 

ci/run.sh

@@ -737,6 +737,9 @@ function gg_sum_embd_bge_small {
 
 ## main
 
+export LLAMA_LOG_PREFIX=1
+export LLAMA_LOG_TIMESTAMPS=1
+
 if [ -z ${GG_BUILD_LOW_PERF} ]; then
     # Create symlink: ./llama.cpp/models-mnt -> $MNT/models/models-mnt
     rm -rf ${SRC}/models-mnt

common/CMakeLists.txt

@@ -51,19 +51,23 @@ endif()
 set(TARGET common)
 
 add_library(${TARGET} STATIC
+    arg.cpp
+    arg.h
     base64.hpp
-    common.h
     common.cpp
-    sampling.h
-    sampling.cpp
-    console.h
+    common.h
     console.cpp
-    json.hpp
+    console.h
     json-schema-to-grammar.cpp
-    train.h
-    train.cpp
-    ngram-cache.h
+    json.hpp
+    log.cpp
+    log.h
     ngram-cache.cpp
+    ngram-cache.h
+    sampling.cpp
+    sampling.h
+    train.cpp
+    train.h
     )
 
 if (BUILD_SHARED_LIBS)