rm get_work_group_size() by local cache for performance (#8286)

Co-authored-by: arthw <14088817+arthw@users.noreply.github.com>

.devops/cloud-v-pipeline

@@ -15,7 +15,7 @@ node('x86_runner1'){            // Running on x86 runner containing latest vecto
     stage('Running llama.cpp'){
         sh'''#!/bin/bash
             module load gnu-bin2/0.1            # loading latest versions of vector qemu and vector gcc
-            qemu-riscv64 -L /softwares/gnu-bin2/sysroot  -cpu rv64,v=true,vlen=256,elen=64,vext_spec=v1.0 ./main -m /home/alitariq/codellama-7b.Q4_K_M.gguf -p "Anything" -n 9 > llama_log.txt            # Running llama.cpp on vector qemu-riscv64
+            qemu-riscv64 -L /softwares/gnu-bin2/sysroot  -cpu rv64,v=true,vlen=256,elen=64,vext_spec=v1.0 ./llama-cli -m /home/alitariq/codellama-7b.Q4_K_M.gguf -p "Anything" -n 9 > llama_log.txt            # Running llama.cpp on vector qemu-riscv64
             cat llama_log.txt                   # Printing results
         '''
     }

.devops/full-cuda.Dockerfile

@@ -12,7 +12,7 @@ FROM ${BASE_CUDA_DEV_CONTAINER} as build
 ARG CUDA_DOCKER_ARCH=all
 
 RUN apt-get update && \
-    apt-get install -y build-essential python3 python3-pip git libcurl4-openssl-dev
+    apt-get install -y build-essential python3 python3-pip git libcurl4-openssl-dev libgomp1
 
 COPY requirements.txt   requirements.txt
 COPY requirements       requirements
@@ -27,10 +27,10 @@ COPY . .
 # Set nvcc architecture
 ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
 # Enable CUDA
-ENV LLAMA_CUDA=1
+ENV GGML_CUDA=1
 # Enable cURL
 ENV LLAMA_CURL=1
 
-RUN make
+RUN make -j$(nproc)
 
 ENTRYPOINT ["/app/.devops/tools.sh"]

.devops/full-rocm.Dockerfile

@@ -36,7 +36,7 @@ COPY . .
 # Set nvcc architecture
 ENV GPU_TARGETS=${ROCM_DOCKER_ARCH}
 # Enable ROCm
-ENV LLAMA_HIPBLAS=1
+ENV GGML_HIPBLAS=1
 ENV CC=/opt/rocm/llvm/bin/clang
 ENV CXX=/opt/rocm/llvm/bin/clang++
 
@@ -45,6 +45,6 @@ ENV LLAMA_CURL=1
 RUN apt-get update && \
     apt-get install -y libcurl4-openssl-dev
 
-RUN make
+RUN make -j$(nproc)
 
 ENTRYPOINT ["/app/.devops/tools.sh"]

.devops/full.Dockerfile

@@ -3,7 +3,7 @@ ARG UBUNTU_VERSION=22.04
 FROM ubuntu:$UBUNTU_VERSION as build
 
 RUN apt-get update && \
-    apt-get install -y build-essential python3 python3-pip git libcurl4-openssl-dev
+    apt-get install -y build-essential python3 python3-pip git libcurl4-openssl-dev libgomp1
 
 COPY requirements.txt   requirements.txt
 COPY requirements       requirements
@@ -18,7 +18,7 @@ COPY . .
 ENV LLAMA_CURL=1
 
 
-RUN make
+RUN make -j$(nproc)
 
 ENV LC_ALL=C.utf8
 

.devops/llama-cli-cuda.Dockerfile

@@ -21,12 +21,15 @@ COPY . .
 # Set nvcc architecture
 ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
 # Enable CUDA
-ENV LLAMA_CUDA=1
+ENV GGML_CUDA=1
 
-RUN make
+RUN make -j$(nproc) llama-cli
 
 FROM ${BASE_CUDA_RUN_CONTAINER} as runtime
 
-COPY --from=build /app/main /main
+RUN apt-get update && \
+    apt-get install -y libgomp1
 
-ENTRYPOINT [ "/main" ]
+COPY --from=build /app/llama-cli /llama-cli
+
+ENTRYPOINT [ "/llama-cli" ]

.devops/llama-cli-intel.Dockerfile

@@ -0,0 +1,26 @@
+ARG ONEAPI_VERSION=2024.1.1-devel-ubuntu22.04
+
+FROM intel/oneapi-basekit:$ONEAPI_VERSION as build
+
+ARG GGML_SYCL_F16=OFF
+RUN apt-get update && \
+    apt-get install -y git
+
+WORKDIR /app
+
+COPY . .
+
+RUN if [ "${GGML_SYCL_F16}" = "ON" ]; then \
+        echo "GGML_SYCL_F16 is set" && \
+        export OPT_SYCL_F16="-DGGML_SYCL_F16=ON"; \
+    fi && \
+    cmake -B build -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ${OPT_SYCL_F16} && \
+    cmake --build build --config Release --target llama-cli
+
+FROM intel/oneapi-basekit:$ONEAPI_VERSION as runtime
+
+COPY --from=build /app/build/bin/llama-cli /llama-cli
+
+ENV LC_ALL=C.utf8
+
+ENTRYPOINT [ "/llama-cli" ]

.devops/llama-cli-rocm.Dockerfile

@@ -36,10 +36,10 @@ COPY . .
 # Set nvcc architecture
 ENV GPU_TARGETS=${ROCM_DOCKER_ARCH}
 # Enable ROCm
-ENV LLAMA_HIPBLAS=1
+ENV GGML_HIPBLAS=1
 ENV CC=/opt/rocm/llvm/bin/clang
 ENV CXX=/opt/rocm/llvm/bin/clang++
 
-RUN make
+RUN make -j$(nproc) llama-cli
 
-ENTRYPOINT [ "/app/main" ]
+ENTRYPOINT [ "/app/llama-cli" ]

.devops/llama-cli-vulkan.Dockerfile

@@ -3,7 +3,7 @@ ARG UBUNTU_VERSION=jammy
 FROM ubuntu:$UBUNTU_VERSION as build
 
 # Install build tools
-RUN apt update && apt install -y git build-essential cmake wget
+RUN apt update && apt install -y git build-essential cmake wget libgomp1
 
 # Install Vulkan SDK
 RUN wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | apt-key add - && \
@@ -14,14 +14,14 @@ RUN wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | apt-key
 # Build it
 WORKDIR /app
 COPY . .
-RUN cmake -B build -DLLAMA_VULKAN=1 && \
-    cmake --build build --config Release --target main
+RUN cmake -B build -DGGML_VULKAN=1 && \
+    cmake --build build --config Release --target llama-cli
 
 # Clean up
 WORKDIR /
-RUN cp /app/build/bin/main /main && \
+RUN cp /app/build/bin/llama-cli /llama-cli && \
     rm -rf /app
 
 ENV LC_ALL=C.utf8
 
-ENTRYPOINT [ "/main" ]
+ENTRYPOINT [ "/llama-cli" ]

.devops/llama-cli.Dockerfile

@@ -9,12 +9,15 @@ WORKDIR /app
 
 COPY . .
 
-RUN make
+RUN make -j$(nproc) llama-cli
 
 FROM ubuntu:$UBUNTU_VERSION as runtime
 
-COPY --from=build /app/main /main
+RUN apt-get update && \
+    apt-get install -y libgomp1
+
+COPY --from=build /app/llama-cli /llama-cli
 
 ENV LC_ALL=C.utf8
 
-ENTRYPOINT [ "/main" ]
+ENTRYPOINT [ "/llama-cli" ]

.devops/llama-cpp-clblast.srpm.spec

@@ -1,84 +0,0 @@
-# SRPM for building from source and packaging an RPM for RPM-based distros.
-# https://docs.fedoraproject.org/en-US/quick-docs/creating-rpm-packages
-# Built and maintained by John Boero - boeroboy@gmail.com
-# In honor of Seth Vidal https://www.redhat.com/it/blog/thank-you-seth-vidal
-
-# Notes for llama.cpp:
-# 1. Tags are currently based on hash - which will not sort asciibetically.
-#    We need to declare standard versioning if people want to sort latest releases.
-# 2. Builds for CUDA/OpenCL support are separate, with different depenedencies.
-# 3. NVidia's developer repo must be enabled with nvcc, cublas, clblas, etc installed.
-#    Example: https://developer.download.nvidia.com/compute/cuda/repos/fedora37/x86_64/cuda-fedora37.repo
-# 4. OpenCL/CLBLAST support simply requires the ICD loader and basic opencl libraries.
-#    It is up to the user to install the correct vendor-specific support.
-
-Name:           llama.cpp-clblast
-Version:        %( date "+%%Y%%m%%d" )
-Release:        1%{?dist}
-Summary:        OpenCL Inference of LLaMA model in C/C++
-License:        MIT
-Source0:        https://github.com/ggerganov/llama.cpp/archive/refs/heads/master.tar.gz
-BuildRequires:  coreutils make gcc-c++ git mesa-libOpenCL-devel clblast-devel
-Requires:       clblast
-URL:            https://github.com/ggerganov/llama.cpp
-
-%define debug_package %{nil}
-%define source_date_epoch_from_changelog 0
-
-%description
-CPU inference for Meta's Lllama2 models using default options.
-
-%prep
-%setup -n llama.cpp-master
-
-%build
-make -j LLAMA_CLBLAST=1
-
-%install
-mkdir -p %{buildroot}%{_bindir}/
-cp -p main %{buildroot}%{_bindir}/llamaclblast
-cp -p server %{buildroot}%{_bindir}/llamaclblastserver
-cp -p simple %{buildroot}%{_bindir}/llamaclblastsimple
-
-mkdir -p %{buildroot}/usr/lib/systemd/system
-%{__cat} <<EOF  > %{buildroot}/usr/lib/systemd/system/llamaclblast.service
-[Unit]
-Description=Llama.cpp server, CPU only (no GPU support in this build).
-After=syslog.target network.target local-fs.target remote-fs.target nss-lookup.target
-
-[Service]
-Type=simple
-EnvironmentFile=/etc/sysconfig/llama
-ExecStart=/usr/bin/llamaclblastserver $LLAMA_ARGS
-ExecReload=/bin/kill -s HUP $MAINPID
-Restart=never
-
-[Install]
-WantedBy=default.target
-EOF
-
-mkdir -p %{buildroot}/etc/sysconfig
-%{__cat} <<EOF  > %{buildroot}/etc/sysconfig/llama
-LLAMA_ARGS="-m /opt/llama2/ggml-model-f32.bin"
-EOF
-
-%clean
-rm -rf %{buildroot}
-rm -rf %{_builddir}/*
-
-%files
-%{_bindir}/llamaclblast
-%{_bindir}/llamaclblastserver
-%{_bindir}/llamaclblastsimple
-/usr/lib/systemd/system/llamaclblast.service
-%config /etc/sysconfig/llama
-
-
-%pre
-
-%post
-
-%preun
-%postun
-
-%changelog

.devops/llama-cpp-cuda.srpm.spec

@@ -32,13 +32,13 @@ CPU inference for Meta's Lllama2 models using default options.
 %setup -n llama.cpp-master
 
 %build
-make -j LLAMA_CUDA=1
+make -j GGML_CUDA=1
 
 %install
 mkdir -p %{buildroot}%{_bindir}/
-cp -p main %{buildroot}%{_bindir}/llamacppcuda
-cp -p server %{buildroot}%{_bindir}/llamacppcudaserver
-cp -p simple %{buildroot}%{_bindir}/llamacppcudasimple
+cp -p llama-cli %{buildroot}%{_bindir}/llama-cuda-cli
+cp -p llama-server %{buildroot}%{_bindir}/llama-cuda-server
+cp -p llama-simple %{buildroot}%{_bindir}/llama-cuda-simple
 
 mkdir -p %{buildroot}/usr/lib/systemd/system
 %{__cat} <<EOF  > %{buildroot}/usr/lib/systemd/system/llamacuda.service
@@ -49,7 +49,7 @@ After=syslog.target network.target local-fs.target remote-fs.target nss-lookup.t
 [Service]
 Type=simple
 EnvironmentFile=/etc/sysconfig/llama
-ExecStart=/usr/bin/llamacppcudaserver $LLAMA_ARGS
+ExecStart=/usr/bin/llama-cuda-server $LLAMA_ARGS
 ExecReload=/bin/kill -s HUP $MAINPID
 Restart=never
 
@@ -67,9 +67,9 @@ rm -rf %{buildroot}
 rm -rf %{_builddir}/*
 
 %files
-%{_bindir}/llamacppcuda
-%{_bindir}/llamacppcudaserver
-%{_bindir}/llamacppcudasimple
+%{_bindir}/llama-cuda-cli
+%{_bindir}/llama-cuda-server
+%{_bindir}/llama-cuda-simple
 /usr/lib/systemd/system/llamacuda.service
 %config /etc/sysconfig/llama
 

.devops/llama-cpp.srpm.spec

@@ -38,9 +38,9 @@ make -j
 
 %install
 mkdir -p %{buildroot}%{_bindir}/
-cp -p main %{buildroot}%{_bindir}/llama
-cp -p server %{buildroot}%{_bindir}/llamaserver
-cp -p simple %{buildroot}%{_bindir}/llamasimple
+cp -p llama-cli %{buildroot}%{_bindir}/llama-cli
+cp -p llama-server %{buildroot}%{_bindir}/llama-server
+cp -p llama-simple %{buildroot}%{_bindir}/llama-simple
 
 mkdir -p %{buildroot}/usr/lib/systemd/system
 %{__cat} <<EOF  > %{buildroot}/usr/lib/systemd/system/llama.service
@@ -51,7 +51,7 @@ After=syslog.target network.target local-fs.target remote-fs.target nss-lookup.t
 [Service]
 Type=simple
 EnvironmentFile=/etc/sysconfig/llama
-ExecStart=/usr/bin/llamaserver $LLAMA_ARGS
+ExecStart=/usr/bin/llama-server $LLAMA_ARGS
 ExecReload=/bin/kill -s HUP $MAINPID
 Restart=never
 
@@ -69,9 +69,9 @@ rm -rf %{buildroot}
 rm -rf %{_builddir}/*
 
 %files
-%{_bindir}/llama
-%{_bindir}/llamaserver
-%{_bindir}/llamasimple
+%{_bindir}/llama-cli
+%{_bindir}/llama-server
+%{_bindir}/llama-simple
 /usr/lib/systemd/system/llama.service
 %config /etc/sysconfig/llama
 

.devops/llama-server-cuda.Dockerfile

@@ -21,17 +21,19 @@ COPY . .
 # Set nvcc architecture
 ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
 # Enable CUDA
-ENV LLAMA_CUDA=1
+ENV GGML_CUDA=1
 # Enable cURL
 ENV LLAMA_CURL=1
 
-RUN make
+RUN make -j$(nproc) llama-server
 
 FROM ${BASE_CUDA_RUN_CONTAINER} as runtime
 
 RUN apt-get update && \
-    apt-get install -y libcurl4-openssl-dev
+    apt-get install -y libcurl4-openssl-dev libgomp1 curl
 
-COPY --from=build /app/server /server
+COPY --from=build /app/llama-server /llama-server
 
-ENTRYPOINT [ "/server" ]
+HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]
+
+ENTRYPOINT [ "/llama-server" ]

.devops/llama-server-intel.Dockerfile

@@ -0,0 +1,31 @@
+ARG ONEAPI_VERSION=2024.1.1-devel-ubuntu22.04
+
+FROM intel/oneapi-basekit:$ONEAPI_VERSION as build
+
+ARG GGML_SYCL_F16=OFF
+RUN apt-get update && \
+    apt-get install -y git libcurl4-openssl-dev
+
+WORKDIR /app
+
+COPY . .
+
+RUN if [ "${GGML_SYCL_F16}" = "ON" ]; then \
+        echo "GGML_SYCL_F16 is set" && \
+        export OPT_SYCL_F16="-DGGML_SYCL_F16=ON"; \
+    fi && \
+    cmake -B build -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_CURL=ON ${OPT_SYCL_F16} && \
+    cmake --build build --config Release --target llama-server
+
+FROM intel/oneapi-basekit:$ONEAPI_VERSION as runtime
+
+RUN apt-get update && \
+    apt-get install -y libcurl4-openssl-dev curl
+
+COPY --from=build /app/build/bin/llama-server /llama-server
+
+ENV LC_ALL=C.utf8
+
+HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]
+
+ENTRYPOINT [ "/llama-server" ]

.devops/llama-server-rocm.Dockerfile

@@ -36,15 +36,17 @@ COPY . .
 # Set nvcc architecture
 ENV GPU_TARGETS=${ROCM_DOCKER_ARCH}
 # Enable ROCm
-ENV LLAMA_HIPBLAS=1
+ENV GGML_HIPBLAS=1
 ENV CC=/opt/rocm/llvm/bin/clang
 ENV CXX=/opt/rocm/llvm/bin/clang++
 
 # Enable cURL
 ENV LLAMA_CURL=1
 RUN apt-get update && \
-    apt-get install -y libcurl4-openssl-dev
+    apt-get install -y libcurl4-openssl-dev curl
 
-RUN make
+RUN make -j$(nproc) llama-server
 
-ENTRYPOINT [ "/app/server" ]
+HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]
+
+ENTRYPOINT [ "/app/llama-server" ]

.devops/llama-server-vulkan.Dockerfile

@@ -5,27 +5,25 @@ FROM ubuntu:$UBUNTU_VERSION as build
 # Install build tools
 RUN apt update && apt install -y git build-essential cmake wget
 
-# Install Vulkan SDK
+# Install Vulkan SDK and cURL
 RUN wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | apt-key add - && \
     wget -qO /etc/apt/sources.list.d/lunarg-vulkan-jammy.list https://packages.lunarg.com/vulkan/lunarg-vulkan-jammy.list && \
     apt update -y && \
-    apt-get install -y vulkan-sdk
-
-# Install cURL
-RUN apt-get update && \
-    apt-get install -y libcurl4-openssl-dev
+    apt-get install -y vulkan-sdk libcurl4-openssl-dev curl
 
 # Build it
 WORKDIR /app
 COPY . .
-RUN cmake -B build -DLLAMA_VULKAN=1 -DLLAMA_CURL=1 && \
-    cmake --build build --config Release --target server
+RUN cmake -B build -DGGML_VULKAN=1 -DLLAMA_CURL=1 && \
+    cmake --build build --config Release --target llama-server
 
 # Clean up
 WORKDIR /
-RUN cp /app/build/bin/server /server && \
+RUN cp /app/build/bin/llama-server /llama-server && \
     rm -rf /app
 
 ENV LC_ALL=C.utf8
 
-ENTRYPOINT [ "/server" ]
+HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]
+
+ENTRYPOINT [ "/llama-server" ]

.devops/llama-server.Dockerfile

@@ -0,0 +1,27 @@
+ARG UBUNTU_VERSION=22.04
+
+FROM ubuntu:$UBUNTU_VERSION as build
+
+RUN apt-get update && \
+    apt-get install -y build-essential git libcurl4-openssl-dev curl
+
+WORKDIR /app
+
+COPY . .
+
+ENV LLAMA_CURL=1
+
+RUN make -j$(nproc) llama-server
+
+FROM ubuntu:$UBUNTU_VERSION as runtime
+
+RUN apt-get update && \
+    apt-get install -y libcurl4-openssl-dev libgomp1
+
+COPY --from=build /app/llama-server /llama-server
+
+ENV LC_ALL=C.utf8
+
+HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]
+
+ENTRYPOINT [ "/llama-server" ]

.devops/main-intel.Dockerfile

@@ -1,26 +0,0 @@
-ARG ONEAPI_VERSION=2024.0.1-devel-ubuntu22.04
-
-FROM intel/oneapi-basekit:$ONEAPI_VERSION as build
-
-ARG LLAMA_SYCL_F16=OFF
-RUN apt-get update && \
-    apt-get install -y git
-
-WORKDIR /app
-
-COPY . .
-
-RUN if [ "${LLAMA_SYCL_F16}" = "ON" ]; then \
-        echo "LLAMA_SYCL_F16 is set" && \
-        export OPT_SYCL_F16="-DLLAMA_SYCL_F16=ON"; \
-    fi && \
-    cmake -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ${OPT_SYCL_F16} && \
-    cmake --build build --config Release --target main
-
-FROM intel/oneapi-basekit:$ONEAPI_VERSION as runtime
-
-COPY --from=build /app/build/bin/main /main
-
-ENV LC_ALL=C.utf8
-
-ENTRYPOINT [ "/main" ]

.devops/nix/apps.nix

@@ -6,11 +6,11 @@
         let
           inherit (config.packages) default;
           binaries = [
-            "llama"
+            "llama-cli"
             "llama-embedding"
             "llama-server"
-            "quantize"
-            "train-text-from-scratch"
+            "llama-quantize"
+            "llama-train-text-from-scratch"
           ];
           mkApp = name: {
             type = "app";

.devops/nix/package.nix

@@ -17,19 +17,18 @@
   rocmPackages,
   vulkan-headers,
   vulkan-loader,
-  clblast,
+  curl,
   useBlas ? builtins.all (x: !x) [
     useCuda
     useMetalKit
-    useOpenCL
     useRocm
     useVulkan
   ] && blas.meta.available,
   useCuda ? config.cudaSupport,
-  useMetalKit ? stdenv.isAarch64 && stdenv.isDarwin && !useOpenCL,
+  useMetalKit ? stdenv.isAarch64 && stdenv.isDarwin,
   useMpi ? false, # Increases the runtime closure size by ~700M
-  useOpenCL ? false,
   useRocm ? config.rocmSupport,
+  enableCurl ? true,
   useVulkan ? false,
   llamaVersion ? "0.0.0", # Arbitrary version, substituted by the flake
 
@@ -56,7 +55,6 @@ let
     ++ lib.optionals useCuda [ "CUDA" ]
     ++ lib.optionals useMetalKit [ "MetalKit" ]
     ++ lib.optionals useMpi [ "MPI" ]
-    ++ lib.optionals useOpenCL [ "OpenCL" ]
     ++ lib.optionals useRocm [ "ROCm" ]
     ++ lib.optionals useVulkan [ "Vulkan" ];
 
@@ -160,9 +158,9 @@ effectiveStdenv.mkDerivation (
     };
 
     postPatch = ''
-      substituteInPlace ./ggml-metal.m \
+      substituteInPlace ./ggml/src/ggml-metal.m \
         --replace '[bundle pathForResource:@"ggml-metal" ofType:@"metal"];' "@\"$out/bin/ggml-metal.metal\";"
-      substituteInPlace ./ggml-metal.m \
+      substituteInPlace ./ggml/src/ggml-metal.m \
         --replace '[bundle pathForResource:@"default" ofType:@"metallib"];' "@\"$out/bin/default.metallib\";"
     '';
 
@@ -198,24 +196,24 @@ effectiveStdenv.mkDerivation (
       optionals effectiveStdenv.isDarwin darwinBuildInputs
       ++ optionals useCuda cudaBuildInputs
       ++ optionals useMpi [ mpi ]
-      ++ optionals useOpenCL [ clblast ]
       ++ optionals useRocm rocmBuildInputs
       ++ optionals useBlas [ blas ]
-      ++ optionals useVulkan vulkanBuildInputs;
+      ++ optionals useVulkan vulkanBuildInputs
+      ++ optionals enableCurl [ curl ];
 
     cmakeFlags =
       [
-        (cmakeBool "LLAMA_NATIVE" false)
         (cmakeBool "LLAMA_BUILD_SERVER" true)
         (cmakeBool "BUILD_SHARED_LIBS" (!enableStatic))
         (cmakeBool "CMAKE_SKIP_BUILD_RPATH" true)
-        (cmakeBool "LLAMA_BLAS" useBlas)
-        (cmakeBool "LLAMA_CLBLAST" useOpenCL)
-        (cmakeBool "LLAMA_CUDA" useCuda)
-        (cmakeBool "LLAMA_HIPBLAS" useRocm)
-        (cmakeBool "LLAMA_METAL" useMetalKit)
-        (cmakeBool "LLAMA_VULKAN" useVulkan)
-        (cmakeBool "LLAMA_STATIC" enableStatic)
+        (cmakeBool "LLAMA_CURL" enableCurl)
+        (cmakeBool "GGML_NATIVE" false)
+        (cmakeBool "GGML_BLAS" useBlas)
+        (cmakeBool "GGML_CUDA" useCuda)
+        (cmakeBool "GGML_HIPBLAS" useRocm)
+        (cmakeBool "GGML_METAL" useMetalKit)
+        (cmakeBool "GGML_VULKAN" useVulkan)
+        (cmakeBool "GGML_STATIC" enableStatic)
       ]
       ++ optionals useCuda [
         (
@@ -231,7 +229,7 @@ effectiveStdenv.mkDerivation (
       ]
       ++ optionals useMetalKit [
         (lib.cmakeFeature "CMAKE_C_FLAGS" "-D__ARM_FEATURE_DOTPROD=1")
-        (cmakeBool "LLAMA_METAL_EMBED_LIBRARY" (!precompileMetalShaders))
+        (cmakeBool "GGML_METAL_EMBED_LIBRARY" (!precompileMetalShaders))
       ];
 
     # Environment variables needed for ROCm
@@ -243,10 +241,8 @@ effectiveStdenv.mkDerivation (
     # TODO(SomeoneSerge): It's better to add proper install targets at the CMake level,
     # if they haven't been added yet.
     postInstall = ''
-      mv $out/bin/main${executableSuffix} $out/bin/llama${executableSuffix}
-      mv $out/bin/server${executableSuffix} $out/bin/llama-server${executableSuffix}
       mkdir -p $out/include
-      cp $src/llama.h $out/include/
+      cp $src/include/llama.h $out/include/
     '';
 
     # Define the shells here, but don't add in the inputsFrom to avoid recursion.
@@ -256,7 +252,6 @@ effectiveStdenv.mkDerivation (
         useCuda
         useMetalKit
         useMpi
-        useOpenCL
         useRocm
         useVulkan
         ;
@@ -283,7 +278,7 @@ effectiveStdenv.mkDerivation (
       # Configurations we don't want even the CI to evaluate. Results in the
       # "unsupported platform" messages. This is mostly a no-op, because
       # cudaPackages would've refused to evaluate anyway.
-      badPlatforms = optionals (useCuda || useOpenCL) lib.platforms.darwin;
+      badPlatforms = optionals useCuda lib.platforms.darwin;
 
       # Configurations that are known to result in build failures. Can be
       # overridden by importing Nixpkgs with `allowBroken = true`.
@@ -294,7 +289,7 @@ effectiveStdenv.mkDerivation (
       license = lib.licenses.mit;
 
       # Accommodates `nix run` and `lib.getExe`
-      mainProgram = "llama";
+      mainProgram = "llama-cli";
 
       # These people might respond, on the best effort basis, if you ping them
       # in case of Nix-specific regressions or for reviewing Nix-specific PRs.

.devops/server-intel.Dockerfile

@@ -1,29 +0,0 @@
-ARG ONEAPI_VERSION=2024.0.1-devel-ubuntu22.04
-
-FROM intel/oneapi-basekit:$ONEAPI_VERSION as build
-
-ARG LLAMA_SYCL_F16=OFF
-RUN apt-get update && \
-    apt-get install -y git libcurl4-openssl-dev
-
-WORKDIR /app
-
-COPY . .
-
-RUN if [ "${LLAMA_SYCL_F16}" = "ON" ]; then \
-        echo "LLAMA_SYCL_F16 is set" && \
-        export OPT_SYCL_F16="-DLLAMA_SYCL_F16=ON"; \
-    fi && \
-    cmake -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_CURL=ON ${OPT_SYCL_F16} && \
-    cmake --build build --config Release --target server
-
-FROM intel/oneapi-basekit:$ONEAPI_VERSION as runtime
-
-RUN apt-get update && \
-    apt-get install -y libcurl4-openssl-dev
-
-COPY --from=build /app/build/bin/server /server
-
-ENV LC_ALL=C.utf8
-
-ENTRYPOINT [ "/server" ]

.devops/server.Dockerfile

@@ -1,25 +0,0 @@
-ARG UBUNTU_VERSION=22.04
-
-FROM ubuntu:$UBUNTU_VERSION as build
-
-RUN apt-get update && \
-    apt-get install -y build-essential git libcurl4-openssl-dev
-
-WORKDIR /app
-
-COPY . .
-
-ENV LLAMA_CURL=1
-
-RUN make
-
-FROM ubuntu:$UBUNTU_VERSION as runtime
-
-RUN apt-get update && \
-    apt-get install -y libcurl4-openssl-dev
-
-COPY --from=build /app/server /server
-
-ENV LC_ALL=C.utf8
-
-ENTRYPOINT [ "/server" ]

.devops/tools.sh

@@ -8,13 +8,13 @@ arg1="$1"
 shift
 
 if [[ "$arg1" == '--convert' || "$arg1" == '-c' ]]; then
-    python3 ./convert.py "$@"
+    python3 ./convert-hf-to-gguf.py "$@"
 elif [[ "$arg1" == '--quantize' || "$arg1" == '-q' ]]; then
-    ./quantize "$@"
+    ./llama-quantize "$@"
 elif [[ "$arg1" == '--run' || "$arg1" == '-r' ]]; then
-    ./main "$@"
+    ./llama-cli "$@"
 elif [[ "$arg1" == '--finetune' || "$arg1" == '-f' ]]; then
-    ./finetune "$@"
+    ./llama-finetune "$@"
 elif [[ "$arg1" == '--all-in-one' || "$arg1" == '-a' ]]; then
     echo "Converting PTH to GGML..."
     for i in `ls $1/$2/ggml-model-f16.bin*`; do
@@ -22,11 +22,11 @@ elif [[ "$arg1" == '--all-in-one' || "$arg1" == '-a' ]]; then
             echo "Skip model quantization, it already exists: ${i/f16/q4_0}"
         else
             echo "Converting PTH to GGML: $i into ${i/f16/q4_0}..."
-            ./quantize "$i" "${i/f16/q4_0}" q4_0
+            ./llama-quantize "$i" "${i/f16/q4_0}" q4_0
         fi
     done
 elif [[ "$arg1" == '--server' || "$arg1" == '-s' ]]; then
-    ./server "$@"
+    ./llama-server "$@"
 else
     echo "Unknown command: $arg1"
     echo "Available commands: "

.dockerignore

@@ -12,8 +12,8 @@ build*/
 
 models/*
 
-/main
-/quantize
+/llama-cli
+/llama-quantize
 
 arm_neon.h
 compile_commands.json

.editorconfig

@@ -26,3 +26,7 @@ indent_size = 2
 
 [examples/llama.swiftui/llama.swiftui.xcodeproj/*]
 indent_style = tab
+
+[examples/cvector-generator/*.txt]
+trim_trailing_whitespace = unset
+insert_final_newline = unset

.github/ISSUE_TEMPLATE/01-bug-low.yml

@@ -0,0 +1,50 @@
+name: Low Severity Bugs
+description: Used to report low severity bugs in llama.cpp (e.g. cosmetic issues, non critical UI glitches)
+title: "Bug: "
+labels: ["bug-unconfirmed", "low severity"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to fill out this bug report!
+        Please include information about your system, the steps to reproduce the bug,
+        and the version of llama.cpp that you are using.
+        If possible, please provide a minimal code example that reproduces the bug.
+  - type: textarea
+    id: what-happened
+    attributes:
+      label: What happened?
+      description: Also tell us, what did you expect to happen?
+      placeholder: Tell us what you see!
+    validations:
+      required: true
+  - type: textarea
+    id: version
+    attributes:
+      label: Name and Version
+      description: Which executable and which version of our software are you running? (use `--version` to get a version string)
+      placeholder: |
+        $./llama-cli --version
+        version: 2999 (42b4109e)
+        built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
+    validations:
+      required: true
+  - type: dropdown
+    id: operating-system
+    attributes:
+      label: What operating system are you seeing the problem on?
+      multiple: true
+      options:
+        - Linux
+        - Mac
+        - Windows
+        - BSD
+        - Other? (Please let us know in description)
+    validations:
+      required: false
+  - type: textarea
+    id: logs
+    attributes:
+      label: Relevant log output
+      description: Please copy and paste any relevant log output. This will be automatically formatted into code, so no need for backticks.
+      render: shell

.github/ISSUE_TEMPLATE/02-bug-medium.yml

@@ -0,0 +1,50 @@
+name: Medium Severity Bug
+description: Used to report medium severity bugs in llama.cpp (e.g. Malfunctioning Features but generally still useable)
+title: "Bug: "
+labels: ["bug-unconfirmed", "medium severity"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to fill out this bug report!
+        Please include information about your system, the steps to reproduce the bug,
+        and the version of llama.cpp that you are using.
+        If possible, please provide a minimal code example that reproduces the bug.
+  - type: textarea
+    id: what-happened
+    attributes:
+      label: What happened?
+      description: Also tell us, what did you expect to happen?
+      placeholder: Tell us what you see!
+    validations:
+      required: true
+  - type: textarea
+    id: version
+    attributes:
+      label: Name and Version
+      description: Which executable and which version of our software are you running? (use `--version` to get a version string)
+      placeholder: |
+        $./llama-cli --version
+        version: 2999 (42b4109e)
+        built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
+    validations:
+      required: true
+  - type: dropdown
+    id: operating-system
+    attributes:
+      label: What operating system are you seeing the problem on?
+      multiple: true
+      options:
+        - Linux
+        - Mac
+        - Windows
+        - BSD
+        - Other? (Please let us know in description)
+    validations:
+      required: false
+  - type: textarea
+    id: logs
+    attributes:
+      label: Relevant log output
+      description: Please copy and paste any relevant log output. This will be automatically formatted into code, so no need for backticks.
+      render: shell

.github/ISSUE_TEMPLATE/03-bug-high.yml

@@ -0,0 +1,50 @@
+name: High Severity Bug
+description: Used to report high severity bugs in llama.cpp (e.g. Malfunctioning features hindering important common workflow)
+title: "Bug: "
+labels: ["bug-unconfirmed", "high severity"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to fill out this bug report!
+        Please include information about your system, the steps to reproduce the bug,
+        and the version of llama.cpp that you are using.
+        If possible, please provide a minimal code example that reproduces the bug.
+  - type: textarea
+    id: what-happened
+    attributes:
+      label: What happened?
+      description: Also tell us, what did you expect to happen?
+      placeholder: Tell us what you see!
+    validations:
+      required: true
+  - type: textarea
+    id: version
+    attributes:
+      label: Name and Version
+      description: Which executable and which version of our software are you running? (use `--version` to get a version string)
+      placeholder: |
+        $./llama-cli --version
+        version: 2999 (42b4109e)
+        built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
+    validations:
+      required: true
+  - type: dropdown
+    id: operating-system
+    attributes:
+      label: What operating system are you seeing the problem on?
+      multiple: true
+      options:
+        - Linux
+        - Mac
+        - Windows
+        - BSD
+        - Other? (Please let us know in description)
+    validations:
+      required: false
+  - type: textarea
+    id: logs
+    attributes:
+      label: Relevant log output
+      description: Please copy and paste any relevant log output. This will be automatically formatted into code, so no need for backticks.
+      render: shell

.github/ISSUE_TEMPLATE/04-bug-critical.yml

@@ -0,0 +1,50 @@
+name: Critical Severity Bug
+description: Used to report critical severity bugs in llama.cpp (e.g. Crashing, Corrupted, Dataloss)
+title: "Bug: "
+labels: ["bug-unconfirmed", "critical severity"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to fill out this bug report!
+        Please include information about your system, the steps to reproduce the bug,
+        and the version of llama.cpp that you are using.
+        If possible, please provide a minimal code example that reproduces the bug.
+  - type: textarea
+    id: what-happened
+    attributes:
+      label: What happened?
+      description: Also tell us, what did you expect to happen?
+      placeholder: Tell us what you see!
+    validations:
+      required: true
+  - type: textarea
+    id: version
+    attributes:
+      label: Name and Version
+      description: Which executable and which version of our software are you running? (use `--version` to get a version string)
+      placeholder: |
+        $./llama-cli --version
+        version: 2999 (42b4109e)
+        built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
+    validations:
+      required: true
+  - type: dropdown
+    id: operating-system
+    attributes:
+      label: What operating system are you seeing the problem on?
+      multiple: true
+      options:
+        - Linux
+        - Mac
+        - Windows
+        - BSD
+        - Other? (Please let us know in description)
+    validations:
+      required: false
+  - type: textarea
+    id: logs
+    attributes:
+      label: Relevant log output
+      description: Please copy and paste any relevant log output. This will be automatically formatted into code, so no need for backticks.
+      render: shell

.github/ISSUE_TEMPLATE/05-enhancement.yml

@@ -0,0 +1,51 @@
+name: Enhancement
+description: Used to request enhancements for llama.cpp
+title: "Feature Request: "
+labels: ["enhancement"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        [Please post your idea first in Discussion if there is not yet a consensus for this enhancement request. This will help to keep this issue tracker focused on enhancements that the community has agreed needs to be implemented.](https://github.com/ggerganov/llama.cpp/discussions/categories/ideas)
+
+  - type: checkboxes
+    id: prerequisites
+    attributes:
+      label: Prerequisites
+      description: Please confirm the following before submitting your enhancement request.
+      options:
+        - label: I am running the latest code. Mention the version if possible as well.
+          required: true
+        - label: I carefully followed the [README.md](https://github.com/ggerganov/llama.cpp/blob/master/README.md).
+          required: true
+        - label: I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
+          required: true
+        - label: I reviewed the [Discussions](https://github.com/ggerganov/llama.cpp/discussions), and have a new and useful enhancement to share.
+          required: true
+
+  - type: textarea
+    id: feature-description
+    attributes:
+      label: Feature Description
+      description: Please provide a detailed written description of what you were trying to do, and what you expected `llama.cpp` to do as an enhancement.
+      placeholder: Detailed description of the enhancement
+    validations:
+      required: true
+
+  - type: textarea
+    id: motivation
+    attributes:
+      label: Motivation
+      description: Please provide a detailed written description of reasons why this feature is necessary and how it is useful to `llama.cpp` users.
+      placeholder: Explanation of why this feature is needed and its benefits
+    validations:
+      required: true
+
+  - type: textarea
+    id: possible-implementation
+    attributes:
+      label: Possible Implementation
+      description: If you have an idea as to how it can be implemented, please write a detailed description. Feel free to give links to external sources or share visuals that might be helpful to understand the details better.
+      placeholder: Detailed description of potential implementation
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/06-research.yml

@@ -0,0 +1,52 @@
+name: Research
+description: Track new technical research area
+title: "Research: "
+labels: ["research 🔬"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Don't forget to check for any [duplicate research issue tickets](https://github.com/ggerganov/llama.cpp/issues?q=is%3Aopen+is%3Aissue+label%3A%22research+%F0%9F%94%AC%22)
+
+  - type: checkboxes
+    id: research-stage
+    attributes:
+      label: Research Stage
+      description: Track general state of this research ticket
+      options:
+        - label: Background Research (Let's try to avoid reinventing the wheel)
+        - label: Hypothesis Formed (How do you think this will work and it's effect?)
+        - label: Strategy / Implementation Forming
+        - label: Analysis of results
+        - label: Debrief / Documentation (So people in the future can learn from us)
+
+  - type: textarea
+    id: background
+    attributes:
+      label: Previous existing literature and research
+      description: Whats the current state of the art and whats the motivation for this research?
+
+  - type: textarea
+    id: hypothesis
+    attributes:
+      label: Hypothesis
+      description: How do you think this will work and it's effect?
+
+  - type: textarea
+    id: implementation
+    attributes:
+      label: Implementation
+      description: Got an approach? e.g. a PR ready to go?
+
+  - type: textarea
+    id: analysis
+    attributes:
+      label: Analysis
+      description: How does the proposed implementation behave?
+
+  - type: textarea
+    id: logs
+    attributes:
+      label: Relevant log output
+      description: Please copy and paste any relevant log output. This will be automatically formatted into code, so no need for backticks.
+      render: shell

.github/ISSUE_TEMPLATE/07-refactor.yml

@@ -0,0 +1,28 @@
+name: Refactor (Maintainers)
+description: Used to track refactoring opportunities
+title: "Refactor: "
+labels: ["refactor"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Don't forget to [check for existing refactor issue tickets](https://github.com/ggerganov/llama.cpp/issues?q=is%3Aopen+is%3Aissue+label%3Arefactoring) in case it's already covered.
+        Also you may want to check [Pull request refactor label as well](https://github.com/ggerganov/llama.cpp/pulls?q=is%3Aopen+is%3Apr+label%3Arefactoring) for duplicates too.
+
+  - type: textarea
+    id: background-description
+    attributes:
+      label: Background Description
+      description: Please provide a detailed written description of the pain points you are trying to solve.
+      placeholder: Detailed description behind your motivation to request refactor
+    validations:
+      required: true
+
+  - type: textarea
+    id: possible-approaches
+    attributes:
+      label: Possible Refactor Approaches
+      description: If you have some idea of possible approaches to solve this problem. You may want to make it a todo list.
+      placeholder: Your idea of possible refactoring opportunity/approaches
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/bug.md

@@ -1,11 +0,0 @@
----
-name: Bug template
-about: Used to report bugs in llama.cpp
-labels: ["bug-unconfirmed"]
-assignees: ''
-
----
-
-Please include information about your system, the steps to reproduce the bug, and the version of llama.cpp that you are using. If possible, please provide a minimal code example that reproduces the bug.
-
-If the bug concerns the server, please try to reproduce it first using the [server test scenario framework](https://github.com/ggerganov/llama.cpp/tree/master/examples/server/tests).

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,11 @@
+blank_issues_enabled: true
+contact_links:
+  - name: Got an idea?
+    url: https://github.com/ggerganov/llama.cpp/discussions/categories/ideas
+    about: Pop it there. It may then become an enhancement ticket.
+  - name: Got a question?
+    url: https://github.com/ggerganov/llama.cpp/discussions/categories/q-a
+    about: Ask a question there!
+  - name: Want to contribute?
+    url: https://github.com/ggerganov/llama.cpp/wiki/contribute
+    about: Head to the contribution guide page of the wiki for areas you can help with

.github/ISSUE_TEMPLATE/enhancement.md

@@ -1,28 +0,0 @@
----
-name: Enhancement template
-about: Used to request enhancements for llama.cpp
-labels: ["enhancement"]
-assignees: ''
-
----
-
-# Prerequisites
-
-Please answer the following questions for yourself before submitting an issue.
-
-- [ ] I am running the latest code. Development is very rapid so there are no tagged versions as of now.
-- [ ] I carefully followed the [README.md](https://github.com/ggerganov/llama.cpp/blob/master/README.md).
-- [ ] I [searched using keywords relevant to my issue](https://docs.github.com/en/issues/tracking-your-work-with-issues/filtering-and-searching-issues-and-pull-requests) to make sure that I am creating a new issue that is not already open (or closed).
-- [ ] I reviewed the [Discussions](https://github.com/ggerganov/llama.cpp/discussions), and have a new bug or useful enhancement to share.
-
-# Feature Description
-
-Please provide a detailed written description of what you were trying to do, and what you expected `llama.cpp` to do as an enhancement.
-
-# Motivation
-
-Please provide a detailed written description of reasons why this feature is necessary and how it is useful to `llama.cpp` users.
-
-# Possible Implementation
-
-If you have an idea as to how it can be implemented, please write a detailed description. Feel free to give links to external sources or share visuals that might be helpful to understand the details better.

.github/labeler.yml

@@ -2,31 +2,31 @@
 Kompute:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml-kompute.h
-            - ggml-kompute.cpp
+            - ggml/include/ggml-kompute.h
+            - ggml/src/ggml-kompute.cpp
             - README-kompute.md
 Apple Metal:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml-metal.h
-            - ggml-metal.cpp
+            - ggml/include/ggml-metal.h
+            - ggml/src/ggml-metal.cpp
             - README-metal.md
 SYCL:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml-sycl.h
-            - ggml-sycl.cpp
+            - ggml/include/ggml-sycl.h
+            - ggml/src/ggml-sycl.cpp
             - README-sycl.md
 Nvidia GPU:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml-cuda.h
-            - ggml-cuda/**
+            - ggml/include/ggml-cuda.h
+            - ggml/src/ggml-cuda/**
 Vulkan:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml_vk_generate_shaders.py
-            - ggml-vulkan*
+            - ggml/ggml_vk_generate_shaders.py
+            - ggml/src/ggml-vulkan*
 documentation:
     - changed-files:
         - any-glob-to-any-file:
@@ -42,7 +42,6 @@ build:
             - cmake/**
             - CMakeLists.txt
             - CMakePresets.json
-            - codecov.yml
 examples:
     - changed-files:
         - any-glob-to-any-file: examples/**
@@ -74,10 +73,10 @@ server:
 ggml:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml.c
-            - ggml.h
-            - ggml-*.c
-            - ggml-*.h
+            - ggml/include/ggml*.h
+            - ggml/src/ggml*.c
+            - ggml/src/ggml*.cpp
+            - ggml/src/ggml*.h
             - ggml-cuda/**
 nix:
     - changed-files:

.github/pull_request_template.md

@@ -0,0 +1,7 @@
+
+
+- [x] I have read the [contributing guidelines](https://github.com/ggerganov/llama.cpp/blob/master/CONTRIBUTING.md)
+- Self-reported review complexity:
+  - [ ] Low
+  - [ ] Medium
+  - [ ] High

.github/workflows/bench.yml

@@ -109,7 +109,7 @@ jobs:
         run: |
           set -eux
           cmake -B build \
-              -DLLAMA_NATIVE=OFF \
+              -DGGML_NATIVE=OFF \
               -DLLAMA_BUILD_SERVER=ON \
               -DLLAMA_CURL=ON \
               -DLLAMA_CUBLAS=ON \
@@ -119,7 +119,7 @@ jobs:
               -DLLAMA_FATAL_WARNINGS=OFF \
               -DLLAMA_ALL_WARNINGS=OFF \
               -DCMAKE_BUILD_TYPE=Release;
-          cmake --build build --config Release -j $(nproc) --target server
+          cmake --build build --config Release -j $(nproc) --target llama-server
 
       - name: Download the dataset
         id: download_dataset

.github/workflows/build.yml

@@ -10,10 +10,10 @@ on:
   push:
     branches:
       - master
-    paths: ['.github/workflows/**', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.swift', '**/*.m']
+    paths: ['.github/workflows/build.yml', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.cuh', '**/*.swift', '**/*.m', '**/*.metal']
   pull_request:
     types: [opened, synchronize, reopened]
-    paths: ['**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.swift', '**/*.m']
+    paths: ['.github/workflows/build.yml', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.cuh', '**/*.swift', '**/*.m', '**/*.metal']
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
@@ -47,7 +47,7 @@ jobs:
           sysctl -a
           mkdir build
           cd build
-          cmake -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_METAL_EMBED_LIBRARY=ON -DLLAMA_CURL=ON ..
+          cmake -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL_EMBED_LIBRARY=ON -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF ..
           cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
 
       - name: Test
@@ -84,7 +84,7 @@ jobs:
           name: llama-bin-macos-arm64.zip
 
   macOS-latest-cmake-x64:
-    runs-on: macos-latest
+    runs-on: macos-12
 
     steps:
       - name: Clone
@@ -103,12 +103,10 @@ jobs:
         id: cmake_build
         run: |
           sysctl -a
-          mkdir build
-          cd build
           # Metal is disabled due to intermittent failures with Github runners not having a GPU:
           # https://github.com/ggerganov/llama.cpp/actions/runs/8635935781/job/23674807267#step:5:2313
-          cmake -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_METAL=OFF -DLLAMA_CURL=ON ..
-          cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
+          cmake -B build -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL=OFF -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF
+          cmake --build build --config Release -j $(sysctl -n hw.logicalcpu)
 
       - name: Test
         id: cmake_test
@@ -224,7 +222,7 @@ jobs:
         run: |
           mkdir build
           cd build
-          cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_CURL=ON
+          cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF
           cmake --build . --config Release -j $(nproc)
 
       - name: Test
@@ -241,8 +239,8 @@ jobs:
           wget https://huggingface.co/karpathy/tinyllamas/resolve/main/stories260K/tok512.bin
           echo "Fetch llama2c model"
           wget https://huggingface.co/karpathy/tinyllamas/resolve/main/stories260K/stories260K.bin
-          ./bin/convert-llama2c-to-ggml --copy-vocab-from-model ./tok512.bin --llama2c-model stories260K.bin --llama2c-output-model stories260K.gguf
-          ./bin/main -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256
+          ./bin/llama-convert-llama2c-to-ggml --copy-vocab-from-model ./tok512.bin --llama2c-model stories260K.bin --llama2c-output-model stories260K.gguf
+          ./bin/llama-cli -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256
 
       - name: Determine tag name
         id: tag
@@ -294,12 +292,22 @@ jobs:
 
       - name: Build
         id: cmake_build
+        if: ${{ matrix.sanitizer != 'THREAD' }}
         run: |
           mkdir build
           cd build
           cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
           cmake --build . --config ${{ matrix.build_type }} -j $(nproc)
 
+      - name: Build (no OpenMP)
+        id: cmake_build_no_openmp
+        if: ${{ matrix.sanitizer == 'THREAD' }}
+        run: |
+          mkdir build
+          cd build
+          cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} -DGGML_OPENMP=OFF
+          cmake --build . --config ${{ matrix.build_type }} -j $(nproc)
+
       - name: Test
         id: cmake_test
         run: |
@@ -327,7 +335,7 @@ jobs:
         run: |
           mkdir build
           cd build
-          cmake -DLLAMA_RPC=ON ..
+          cmake -DGGML_RPC=ON ..
           cmake --build . --config Release -j $(nproc)
 
       - name: Test
@@ -355,7 +363,7 @@ jobs:
         run: |
           mkdir build
           cd build
-          cmake -DLLAMA_VULKAN=ON ..
+          cmake -DGGML_VULKAN=ON ..
           cmake --build . --config Release -j $(nproc)
 
   ubuntu-22-cmake-hip:
@@ -376,13 +384,13 @@ jobs:
       - name: Build with native CMake HIP support
         id: cmake_build
         run: |
-          cmake -B build -S . -DCMAKE_HIP_COMPILER="$(hipconfig -l)/clang" -DLLAMA_HIPBLAS=ON
+          cmake -B build -S . -DCMAKE_HIP_COMPILER="$(hipconfig -l)/clang" -DGGML_HIPBLAS=ON
           cmake --build build --config Release -j $(nproc)
 
       - name: Build with legacy HIP support
         id: cmake_build_legacy_hip
         run: |
-          cmake -B build2 -S . -DCMAKE_C_COMPILER=hipcc -DCMAKE_CXX_COMPILER=hipcc -DLLAMA_HIPBLAS=ON
+          cmake -B build2 -S . -DCMAKE_C_COMPILER=hipcc -DCMAKE_CXX_COMPILER=hipcc -DGGML_HIPBLAS=ON
           cmake --build build2 --config Release -j $(nproc)
 
   ubuntu-22-cmake-sycl:
@@ -423,7 +431,7 @@ jobs:
           source /opt/intel/oneapi/setvars.sh
           mkdir build
           cd build
-          cmake -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ..
+          cmake -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ..
           cmake --build . --config Release -j $(nproc)
 
   ubuntu-22-cmake-sycl-fp16:
@@ -464,10 +472,10 @@ jobs:
           source /opt/intel/oneapi/setvars.sh
           mkdir build
           cd build
-          cmake -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON ..
+          cmake -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON ..
           cmake --build . --config Release -j $(nproc)
 
-  # TODO: build with LLAMA_NO_METAL because test-backend-ops fail on "Apple Paravirtual device" and I don't know
+  # TODO: build with GGML_NO_METAL because test-backend-ops fail on "Apple Paravirtual device" and I don't know
   #       how to debug it.
   #       ref: https://github.com/ggerganov/llama.cpp/actions/runs/7131777249/job/19420981052#step:5:1124
   macOS-latest-make:
@@ -489,15 +497,15 @@ jobs:
         env:
             LLAMA_FATAL_WARNINGS: 1
         run: |
-          LLAMA_NO_METAL=1 make -j $(sysctl -n hw.logicalcpu)
+          GGML_NO_METAL=1 make -j $(sysctl -n hw.logicalcpu)
 
       - name: Test
         id: make_test
         run: |
-          LLAMA_NO_METAL=1 make tests -j $(sysctl -n hw.logicalcpu)
-          LLAMA_NO_METAL=1 make test  -j $(sysctl -n hw.logicalcpu)
+          GGML_NO_METAL=1 make tests -j $(sysctl -n hw.logicalcpu)
+          GGML_NO_METAL=1 make test  -j $(sysctl -n hw.logicalcpu)
 
-  # TODO: build with LLAMA_METAL=OFF because test-backend-ops fail on "Apple Paravirtual device" and I don't know
+  # TODO: build with GGML_METAL=OFF because test-backend-ops fail on "Apple Paravirtual device" and I don't know
   #       how to debug it.
   #       ref: https://github.com/ggerganov/llama.cpp/actions/runs/7132125951/job/19422043567?pr=4359#step:5:6584
   #       would be great if we fix these
@@ -521,7 +529,7 @@ jobs:
           sysctl -a
           mkdir build
           cd build
-          cmake -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_METAL=OFF ..
+          cmake -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL=OFF ..
           cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
 
       - name: Test
@@ -551,13 +559,14 @@ jobs:
           mkdir build
           cd build
           cmake -G Xcode .. \
-            -DLLAMA_METAL_EMBED_LIBRARY=ON \
+            -DGGML_METAL_EMBED_LIBRARY=ON \
             -DLLAMA_BUILD_EXAMPLES=OFF \
             -DLLAMA_BUILD_TESTS=OFF \
             -DLLAMA_BUILD_SERVER=OFF \
             -DCMAKE_SYSTEM_NAME=iOS \
-            -DCMAKE_OSX_DEPLOYMENT_TARGET=14.0
-          cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
+            -DCMAKE_OSX_DEPLOYMENT_TARGET=14.0 \
+            -DCMAKE_XCODE_ATTRIBUTE_DEVELOPMENT_TEAM=ggml
+          cmake --build . --config Release -j $(sysctl -n hw.logicalcpu) -- CODE_SIGNING_ALLOWED=NO
 
   macOS-latest-cmake-tvos:
     runs-on: macos-latest
@@ -580,13 +589,14 @@ jobs:
           mkdir build
           cd build
           cmake -G Xcode .. \
-            -DLLAMA_METAL_EMBED_LIBRARY=ON \
+            -DGGML_METAL_EMBED_LIBRARY=ON \
             -DLLAMA_BUILD_EXAMPLES=OFF \
             -DLLAMA_BUILD_TESTS=OFF \
             -DLLAMA_BUILD_SERVER=OFF \
             -DCMAKE_SYSTEM_NAME=tvOS \
-            -DCMAKE_OSX_DEPLOYMENT_TARGET=14.0
-          cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
+            -DCMAKE_OSX_DEPLOYMENT_TARGET=14.0 \
+            -DCMAKE_XCODE_ATTRIBUTE_DEVELOPMENT_TEAM=ggml
+          cmake --build . --config Release -j $(sysctl -n hw.logicalcpu) -- CODE_SIGNING_ALLOWED=NO
 
   macOS-latest-swift:
     runs-on: macos-latest
@@ -654,7 +664,7 @@ jobs:
       - name: Build using make w/ OpenBLAS
         shell: msys2 {0}
         run: |
-            make LLAMA_OPENBLAS=1 -j $(nproc)
+            make GGML_OPENBLAS=1 -j $(nproc)
 
       - name: Build using CMake
         shell: msys2 {0}
@@ -670,16 +680,14 @@ jobs:
       - name: Build using CMake w/ OpenBLAS
         shell: msys2 {0}
         run: |
-            cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
+            cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS
             cmake --build build --config ${{ matrix.build }} -j $(nproc)
 
   windows-latest-cmake:
-    runs-on: windows-latest
+    runs-on: windows-2019
 
     env:
       OPENBLAS_VERSION: 0.3.23
-      OPENCL_VERSION: 2023.04.17
-      CLBLAST_VERSION: 1.6.0
       SDE_VERSION: 9.33.0-2024-01-07
       VULKAN_VERSION: 1.3.261.1
 
@@ -687,27 +695,25 @@ jobs:
       matrix:
         include:
           - build: 'rpc-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_RPC=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'noavx-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DBUILD_SHARED_LIBS=ON'
           - build: 'avx2-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'avx-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX2=OFF -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX2=OFF -DBUILD_SHARED_LIBS=ON'
           - build: 'avx512-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX512=ON -DBUILD_SHARED_LIBS=ON'
-          - build: 'clblast-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_CLBLAST=ON -DBUILD_SHARED_LIBS=ON -DCMAKE_PREFIX_PATH="$env:RUNNER_TEMP/clblast"'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX512=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'openblas-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DBUILD_SHARED_LIBS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_BLAS=ON -DBUILD_SHARED_LIBS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
           - build: 'kompute-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_KOMPUTE=ON -DKOMPUTE_OPT_DISABLE_VULKAN_VERSION_CHECK=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_KOMPUTE=ON -DKOMPUTE_OPT_DISABLE_VULKAN_VERSION_CHECK=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'vulkan-x64'
-            defines: '-DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_VULKAN=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_VULKAN=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'llvm-arm64'
-            defines: '-G "Ninja Multi-Config" -D CMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-llvm.cmake -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-G "Ninja Multi-Config" -D CMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-llvm.cmake -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
           - build: 'msvc-arm64'
-            defines: '-G "Ninja Multi-Config" -D CMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-msvc.cmake -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
+            defines: '-G "Ninja Multi-Config" -D CMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-msvc.cmake -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
 
     steps:
       - name: Clone
@@ -720,28 +726,7 @@ jobs:
         id: clone_kompute
         if: ${{ matrix.build == 'kompute-x64' }}
         run: |
-          git submodule update --init kompute
-
-      - name: Download OpenCL SDK
-        id: get_opencl
-        if: ${{ matrix.build == 'clblast-x64' }}
-        run: |
-          curl.exe -o $env:RUNNER_TEMP/opencl.zip -L "https://github.com/KhronosGroup/OpenCL-SDK/releases/download/v${env:OPENCL_VERSION}/OpenCL-SDK-v${env:OPENCL_VERSION}-Win-x64.zip"
-          mkdir $env:RUNNER_TEMP/opencl
-          tar.exe -xvf $env:RUNNER_TEMP/opencl.zip --strip-components=1 -C $env:RUNNER_TEMP/opencl
-
-      - name: Download CLBlast
-        id: get_clblast
-        if: ${{ matrix.build == 'clblast-x64' }}
-        run: |
-          curl.exe -o $env:RUNNER_TEMP/clblast.7z -L "https://github.com/CNugteren/CLBlast/releases/download/${env:CLBLAST_VERSION}/CLBlast-${env:CLBLAST_VERSION}-windows-x64.7z"
-          curl.exe -o $env:RUNNER_TEMP/CLBlast.LICENSE.txt -L "https://github.com/CNugteren/CLBlast/raw/${env:CLBLAST_VERSION}/LICENSE"
-          7z x "-o${env:RUNNER_TEMP}" $env:RUNNER_TEMP/clblast.7z
-          rename-item $env:RUNNER_TEMP/CLBlast-${env:CLBLAST_VERSION}-windows-x64 clblast
-          foreach ($f in (gci -Recurse -Path "$env:RUNNER_TEMP/clblast" -Filter '*.cmake')) {
-            $txt = Get-Content -Path $f -Raw
-            $txt.Replace('C:/vcpkg/packages/opencl_x64-windows/', "$($env:RUNNER_TEMP.Replace('\','/'))/opencl/") | Set-Content -Path $f -Encoding UTF8
-          }
+          git submodule update --init ggml/src/kompute
 
       - name: Download OpenBLAS
         id: get_openblas
@@ -776,13 +761,6 @@ jobs:
           cmake -S . -B build ${{ matrix.defines }}
           cmake --build build --config Release -j ${env:NUMBER_OF_PROCESSORS}
 
-      - name: Add clblast.dll
-        id: add_clblast_dll
-        if: ${{ matrix.build == 'clblast-x64' }}
-        run: |
-          cp $env:RUNNER_TEMP/clblast/lib/clblast.dll ./build/bin/Release
-          cp $env:RUNNER_TEMP/CLBlast.LICENSE.txt ./build/bin/Release/CLBlast-${env:CLBLAST_VERSION}.txt
-
       - name: Add libopenblas.dll
         id: add_libopenblas_dll
         if: ${{ matrix.build == 'openblas-x64' }}
@@ -806,7 +784,7 @@ jobs:
       - name: Test
         id: cmake_test
         # not all machines have native AVX-512
-        if: ${{ matrix.build != 'msvc-arm64' && matrix.build != 'llvm-arm64' && matrix.build != 'clblast-x64' && matrix.build != 'kompute-x64' && matrix.build != 'vulkan-x64' && (matrix.build != 'avx512-x64' || env.HAS_AVX512F == '1') }}
+        if: ${{ matrix.build != 'msvc-arm64' && matrix.build != 'llvm-arm64' && matrix.build != 'kompute-x64' && matrix.build != 'vulkan-x64' && (matrix.build != 'avx512-x64' || env.HAS_AVX512F == '1') }}
         run: |
           cd build
           ctest -L main -C Release --verbose --timeout 900
@@ -821,6 +799,7 @@ jobs:
           7z x "-o${env:RUNNER_TEMP}" $env:RUNNER_TEMP/sde.tar
           $sde = $(join-path $env:RUNNER_TEMP sde-external-${env:SDE_VERSION}-win/sde.exe)
           cd build
+          $env:LLAMA_SKIP_TESTS_SLOW_ON_EMULATOR = 1
           & $sde -future -- ctest -L main -C Release --verbose --timeout 900
 
       - name: Determine tag name
@@ -851,7 +830,7 @@ jobs:
           name: llama-bin-win-${{ matrix.build }}.zip
 
   windows-latest-cmake-cuda:
-    runs-on: windows-latest
+    runs-on: windows-2019
 
     strategy:
       matrix:
@@ -865,8 +844,9 @@ jobs:
         with:
           fetch-depth: 0
 
-      - uses: Jimver/cuda-toolkit@v0.2.11
+      - name: Install CUDA toolkit
         id: cuda-toolkit
+        uses: Jimver/cuda-toolkit@v0.2.15
         with:
           cuda: ${{ matrix.cuda }}
           method: 'network'
@@ -877,7 +857,7 @@ jobs:
         run: |
           mkdir build
           cd build
-          cmake .. -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=ON
+          cmake .. -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_CUDA=ON -DBUILD_SHARED_LIBS=ON
           cmake --build . --config Release -j ${env:NUMBER_OF_PROCESSORS}
 
       - name: Determine tag name
@@ -1010,7 +990,7 @@ jobs:
         run: |
           $env:HIP_PATH=$(Resolve-Path 'C:\Program Files\AMD\ROCm\*\bin\clang.exe' | split-path | split-path)
           $env:CMAKE_PREFIX_PATH="${env:HIP_PATH}"
-          cmake -G "Unix Makefiles" -B build -S . -DCMAKE_C_COMPILER="${env:HIP_PATH}\bin\clang.exe" -DCMAKE_CXX_COMPILER="${env:HIP_PATH}\bin\clang++.exe" -DLLAMA_HIPBLAS=ON
+          cmake -G "Unix Makefiles" -B build -S . -DCMAKE_C_COMPILER="${env:HIP_PATH}\bin\clang.exe" -DCMAKE_CXX_COMPILER="${env:HIP_PATH}\bin\clang++.exe" -DGGML_HIPBLAS=ON
           cmake --build build --config Release
 
   ios-xcode-build:
@@ -1061,7 +1041,7 @@ jobs:
 #        hypervisor: 'qemu'
 #        run: |
 #            sudo pkg update
-#            sudo pkg install -y gmake automake autoconf pkgconf llvm15 clinfo clover opencl clblast openblas
+#            sudo pkg install -y gmake automake autoconf pkgconf llvm15 openblas
 #            gmake CC=/usr/local/bin/clang15 CXX=/usr/local/bin/clang++15 -j `sysctl -n hw.ncpu`
 
   release:

.github/workflows/code-coverage.yml

@@ -1,40 +0,0 @@
-name: Code Coverage
-on: [push, pull_request]
-
-env:
-  GGML_NLOOP: 3
-  GGML_N_THREADS: 1
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
-  cancel-in-progress: true
-
-jobs:
-  run:
-    runs-on: ubuntu-20.04
-    steps:
-      - name: Checkout
-        uses: actions/checkout@v4
-
-      - name: Dependencies
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential gcc-8 lcov
-
-      - name: Build
-        run: CC=gcc-8 make -j LLAMA_CODE_COVERAGE=1 tests
-
-      - name: Run tests
-        run: CC=gcc-8 make test
-
-      - name: Generate coverage report
-        run: |
-          make coverage
-          make lcov-report
-
-      - name: Upload coverage to Codecov
-        uses: codecov/codecov-action@v3
-        env:
-           CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
-        with:
-          files: lcov-report/coverage.info

.github/workflows/docker.yml

@@ -10,10 +10,11 @@
 name: Publish Docker image
 
 on:
-  pull_request:
+  #pull_request:
   push:
     branches:
       - master
+    paths: ['.github/workflows/docker.yml', '.devops/*.Dockerfile', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.cuh', '**/*.swift', '**/*.m', '**/*.metal']
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
@@ -22,7 +23,7 @@ concurrency:
 jobs:
   push_to_registry:
     name: Push Docker image to Docker Hub
-    if: github.event.pull_request.draft == false
+    #if: github.event.pull_request.draft == false
 
     runs-on: ubuntu-latest
     env:
@@ -30,21 +31,18 @@ jobs:
     strategy:
       matrix:
         config:
-          - { tag: "light", dockerfile: ".devops/main.Dockerfile", platforms: "linux/amd64,linux/arm64" }
+          - { tag: "light", dockerfile: ".devops/llama-cli.Dockerfile", platforms: "linux/amd64,linux/arm64" }
+          - { tag: "server", dockerfile: ".devops/llama-server.Dockerfile", platforms: "linux/amd64,linux/arm64" }
           - { tag: "full", dockerfile: ".devops/full.Dockerfile", platforms: "linux/amd64,linux/arm64" }
-          - { tag: "server", dockerfile: ".devops/server.Dockerfile", platforms: "linux/amd64,linux/arm64" }
-          # NOTE(canardletter): The CUDA builds on arm64 are very slow, so I
-          #                     have disabled them for now until the reason why
-          #                     is understood.
-          - { tag: "light-cuda", dockerfile: ".devops/main-cuda.Dockerfile", platforms: "linux/amd64" }
+          - { tag: "light-cuda", dockerfile: ".devops/llama-cli-cuda.Dockerfile", platforms: "linux/amd64" }
+          - { tag: "server-cuda", dockerfile: ".devops/llama-server-cuda.Dockerfile", platforms: "linux/amd64" }
           - { tag: "full-cuda", dockerfile: ".devops/full-cuda.Dockerfile", platforms: "linux/amd64" }
-          - { tag: "server-cuda", dockerfile: ".devops/server-cuda.Dockerfile", platforms: "linux/amd64" }
-          - { tag: "light-rocm", dockerfile: ".devops/main-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
-          - { tag: "full-rocm", dockerfile: ".devops/full-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
-          - { tag: "server-rocm", dockerfile: ".devops/server-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
-          # TODO: Disabled due to build issues https://github.com/ggerganov/llama.cpp/issues/7507
-          #- { tag: "light-intel", dockerfile: ".devops/main-intel.Dockerfile", platforms: "linux/amd64" }
-          #- { tag: "server-intel", dockerfile: ".devops/server-intel.Dockerfile", platforms: "linux/amd64" }
+          - { tag: "light-rocm", dockerfile: ".devops/llama-cli-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
+          - { tag: "server-rocm", dockerfile: ".devops/llama-server-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
+          # Note: the full-rocm image is failing due to a "no space left on device" error. It is disabled for now to allow the workflow to complete.
+          #- { tag: "full-rocm", dockerfile: ".devops/full-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
+          - { tag: "light-intel", dockerfile: ".devops/llama-cli-intel.Dockerfile", platforms: "linux/amd64" }
+          - { tag: "server-intel", dockerfile: ".devops/llama-server-intel.Dockerfile", platforms: "linux/amd64" }
     steps:
       - name: Check out the repo
         uses: actions/checkout@v4

.github/workflows/server.yml

@@ -16,11 +16,9 @@ on:
     branches:
       - master
     paths: ['.github/workflows/server.yml', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.swift', '**/*.m', 'examples/server/**.*']
-  pull_request_target:
+  pull_request:
     types: [opened, synchronize, reopened]
     paths: ['.github/workflows/server.yml', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu', '**/*.swift', '**/*.m', 'examples/server/**.*']
-  schedule:
-    -  cron: '2 4 * * *'
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.run_id }}
@@ -32,7 +30,7 @@ jobs:
 
     strategy:
       matrix:
-        sanitizer: [ADDRESS, THREAD, UNDEFINED]
+        sanitizer: [ADDRESS, UNDEFINED] # THREAD is broken
         build_type: [RelWithDebInfo]
         include:
           - build_type: Release
@@ -89,16 +87,30 @@ jobs:
             exit 1
           fi
 
-      - name: Build
-        id: cmake_build
+      - name: Build (no OpenMP)
+        id: cmake_build_no_openmp
+        if: ${{ matrix.sanitizer == 'THREAD' }}
         run: |
           cmake -B build \
-              -DLLAMA_NATIVE=OFF \
+              -DGGML_NATIVE=OFF \
+              -DLLAMA_BUILD_SERVER=ON \
+              -DLLAMA_CURL=ON \
+              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
+              -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON \
+              -DGGML_OPENMP=OFF ;
+          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
+
+      - name: Build
+        id: cmake_build
+        if: ${{ matrix.sanitizer != 'THREAD' }}
+        run: |
+          cmake -B build \
+              -DGGML_NATIVE=OFF \
               -DLLAMA_BUILD_SERVER=ON \
               -DLLAMA_CURL=ON \
               -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
               -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ;
-          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target server
+          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
 
       - name: Tests
         id: server_integration_tests
@@ -115,7 +127,7 @@ jobs:
 
 
   server-windows:
-    runs-on: windows-latest
+    runs-on: windows-2019
 
     steps:
       - name: Clone
@@ -138,7 +150,7 @@ jobs:
         id: cmake_build
         run: |
           cmake -B build -DLLAMA_CURL=ON -DCURL_LIBRARY="$env:RUNNER_TEMP/libcurl/lib/libcurl.dll.a" -DCURL_INCLUDE_DIR="$env:RUNNER_TEMP/libcurl/include"
-          cmake --build build --config Release -j ${env:NUMBER_OF_PROCESSORS} --target server
+          cmake --build build --config Release -j ${env:NUMBER_OF_PROCESSORS} --target llama-server
 
       - name: Python setup
         id: setup_python

.gitignore

@@ -1,126 +1,125 @@
-*.o
+# Extensions
+
 *.a
-*.so
+*.bat
+*.bin
+*.dll
+*.dot
+*.etag
+*.exe
+*.gcda
+*.gcno
+*.gcov
 *.gguf
 *.gguf.json
-*.bin
-*.exe
-*.dll
-*.log
-*.gcov
-*.gcno
-*.gcda
-*.dot
-*.bat
-*.tmp
-*.metallib
-*.etag
 *.lastModified
-.DS_Store
-.build/
+*.log
+*.metallib
+*.o
+*.so
+*.tmp
+
+# IDE / OS
+
 .cache/
 .ccls-cache/
 .direnv/
+.DS_Store
 .envrc
+.idea/
 .swiftpm
-.venv
-.clang-tidy
 .vs/
 .vscode/
-.idea/
+nppBackup
 
-ggml-metal-embed.metal
 
-lcov-report/
+# Coverage
+
 gcovr-report/
+lcov-report/
 
+# Build Artifacts
+
+tags
+.build/
 build*
+!build-info.cmake
+!build-info.cpp.in
+!build-info.sh
 !build.zig
+/libllama.so
+/llama-*
+android-ndk-*
+arm_neon.h
 cmake-build-*
+CMakeSettings.json
+compile_commands.json
+ggml-metal-embed.metal
+llama-batched-swift
+/rpc-server
 out/
 tmp/
 
+# CI
+
+!.github/workflows/*.yml
+
+# Models
+
 models/*
 models-mnt
+!models/.editorconfig
+!models/ggml-vocab-*.gguf*
 
-/Pipfile
-/baby-llama
-/beam-search
-/benchmark-matmult
-/convert-llama2c-to-ggml
-/embd-input-test
-/embedding
-/eval-callback
-/gguf
-/gguf-llama-simple
-/gguf-split
-/gritlm
-/imatrix
-/infill
-/libllama.so
-/llama-bench
-/llava-cli
-/lookahead
-/lookup
-/lookup-create
-/lookup-merge
-/lookup-stats
-/main
-/metal
-/passkey
-/perplexity
-/q8dot
-/quantize
-/quantize-stats
-/result
-/save-load-state
-/server
-/simple
-/batched
-/batched-bench
-/export-lora
-/finetune
-/retrieval
-/speculative
-/parallel
-/train-text-from-scratch
-/tokenize
-/vdot
-/common/build-info.cpp
-arm_neon.h
-compile_commands.json
-CMakeSettings.json
-
-__pycache__
-dist
+# Zig
 
 zig-out/
 zig-cache/
 
+# Logs
+
 ppl-*.txt
 qnt-*.txt
 perf-*.txt
 
+# Examples
+
 examples/jeopardy/results.txt
+examples/server/*.css.hpp
 examples/server/*.html.hpp
 examples/server/*.js.hpp
 examples/server/*.mjs.hpp
+!build_64.sh
+!examples/*.bat
+!examples/*/*.kts
+!examples/*/*/*.kts
+!examples/sycl/*.bat
+!examples/sycl/*.sh
 
-poetry.lock
+# Python
+
+/.venv
+__pycache__/
+*/poetry.lock
 poetry.toml
-nppBackup
+
+# Nix
+/result
 
 # Test binaries
-/tests/test-grammar-parser
-/tests/test-llama-grammar
+/tests/test-backend-ops
 /tests/test-double-float
 /tests/test-grad0
+/tests/test-grammar-parser
+/tests/test-llama-grammar
 /tests/test-opt
 /tests/test-quantize-fns
 /tests/test-quantize-perf
+/tests/test-rope
 /tests/test-sampling
 /tests/test-tokenizer-0
-/tests/test-tokenizer-1-spm
 /tests/test-tokenizer-1-bpe
-/tests/test-rope
-/tests/test-backend-ops
+/tests/test-tokenizer-1-spm
+
+# Scripts
+!/scripts/install-oneapi.bat

.gitmodules

@@ -1,3 +1,3 @@
 [submodule "kompute"]
-	path = kompute
+	path = ggml/src/kompute
 	url = https://github.com/nomic-ai/kompute.git

AUTHORS

@@ -1,8 +1,9 @@
-# date: Tue Apr  9 09:17:14 EEST 2024
+# date: Wed Jun 26 19:36:34 EEST 2024
 # this file is auto-generated by scripts/gen-authors.sh
 
 0cc4m <picard12@live.de>
 0xspringtime <110655352+0xspringtime@users.noreply.github.com>
+20kdc <asdd2808@gmail.com>
 2f38b454 <dxf@protonmail.com>
 3ooabkhxtn <31479382+3ooabkhxtn@users.noreply.github.com>
 44670 <44670@users.noreply.github.com>
@@ -11,14 +12,18 @@ AT <manyoso@users.noreply.github.com>
 Aarni Koskela <akx@iki.fi>
 Aaron Miller <apage43@ninjawhale.com>
 Aaryaman Vasishta <aaryaman.vasishta@amd.com>
+Abheek Gulati <abheekg@hotmail.com>
 Abhilash Majumder <30946547+abhilash1910@users.noreply.github.com>
 Abhishek Gopinath K <31348521+overtunned@users.noreply.github.com>
 Adithya Balaji <adithya.b94@gmail.com>
 AdithyanI <adithyan.i4internet@gmail.com>
 Adrian <smith.adriane@gmail.com>
 Adrian Hesketh <a-h@users.noreply.github.com>
+Ahmet Zeer <ahmed.zeer@std.yildiz.edu.tr>
 AidanBeltonS <87009434+AidanBeltonS@users.noreply.github.com>
 Aisuko <urakiny@gmail.com>
+Akarshan Biswas <akarshanbiswas@fedoraproject.org>
+Albert Jin <albert.jin@gmail.com>
 Alberto <57916483+albbus-stack@users.noreply.github.com>
 Alex <awhill19@icloud.com>
 Alex Azarov <alex@azarov.by>
@@ -35,19 +40,24 @@ Ali Nehzat <ali.nehzat@thanks.dev>
 Ali Tariq <ali.tariq@10xengineers.ai>
 Alon <alonfaraj@gmail.com>
 AlpinDale <52078762+AlpinDale@users.noreply.github.com>
+Amir <amir_zia@outlook.com>
 AmirAli Mirian <37371367+amiralimi@users.noreply.github.com>
 Ananta Bastola <anantarajbastola@gmail.com>
 Anas Ahouzi <112881240+aahouzi@users.noreply.github.com>
 András Salamon <ott2@users.noreply.github.com>
 Andrei <abetlen@gmail.com>
 Andrew Canis <andrew.canis@gmail.com>
+Andrew Downing <andrew2085@gmail.com>
 Andrew Duffy <a10y@users.noreply.github.com>
 Andrew Godfrey <AndrewGodfrey@users.noreply.github.com>
+Andy Tai <andy-tai@users.noreply.github.com>
 Arik Poznanski <arikpoz@users.noreply.github.com>
 Artem <guinmoon@gmail.com>
+Artem Zinnatullin <ceo@abstractny.gay>
 Artyom Lebedev <vagran.ast@gmail.com>
 Asbjørn Olling <asbjornolling@gmail.com>
 Ásgeir Bjarni Ingvarsson <asgeir@fundinn.org>
+Ashish <1856117+ashishdatta@users.noreply.github.com>
 Ashok Gelal <401055+ashokgelal@users.noreply.github.com>
 Ashraful Islam <ashraful.meche@gmail.com>
 Atsushi Tatsuma <yoshoku@outlook.com>
@@ -57,35 +67,46 @@ BADR <contact@pythops.com>
 Bach Le <bach@bullno1.com>
 Bailey Chittle <39804642+bachittle@users.noreply.github.com>
 BarfingLemurs <128182951+BarfingLemurs@users.noreply.github.com>
+Bartowski <ckealty1182@gmail.com>
 Behnam M <58621210+ibehnam@users.noreply.github.com>
+Ben Ashbaugh <ben.ashbaugh@intel.com>
 Ben Garney <bengarney@users.noreply.github.com>
 Ben Siraphob <bensiraphob@gmail.com>
 Ben Williams <ben@719ben.com>
+Benjamin Findley <39356821+Kartoffelsaft@users.noreply.github.com>
 Benjamin Lecaillon <84293038+blecaillon@users.noreply.github.com>
 Bernat Vadell <hounter.caza@gmail.com>
+Bingan <70050083+binganao@users.noreply.github.com>
 Bodo Graumann <mail@bodograumann.de>
 Bono Lv <lvscar@users.noreply.github.com>
 Borislav Stanimirov <b.stanimirov@abv.bg>
 Branden Butler <bwtbutler@hotmail.com>
 Brian <mofosyne@gmail.com>
 Bruce MacDonald <brucewmacdonald@gmail.com>
+Bryan Honof <bryanhonof@gmail.com>
 CJ Pais <cj@cjpais.com>
 CRD716 <crd716@gmail.com>
+Calvin Laurenson <calvin@laurenson.dev>
 Cameron <csteele@steelecameron.com>
 Cameron Kaiser <classilla@users.noreply.github.com>
+Carolinabanana <140120812+Carolinabanana@users.noreply.github.com>
 Casey Primozic <casey@cprimozic.net>
 Casey Primozic <me@ameo.link>
 CausalLM <148736309+CausalLM@users.noreply.github.com>
 Cebtenzzre <cebtenzzre@gmail.com>
 Chad Brewbaker <crb002@gmail.com>
+Chao Jiang <jc19chaoj@zoho.com>
 Cheng Shao <terrorjack@type.dance>
+Chris Elrod <elrodc@gmail.com>
 Chris Kuehl <ckuehl@ckuehl.me>
 Christian Demsar <christian@github.email.demsar.us>
 Christian Demsar <crasm@git.vczf.us>
 Christian Falch <875252+chrfalch@users.noreply.github.com>
 Christian Kögler <ck3d@gmx.de>
+Christian Zhou-Zheng <59622928+christianazinn@users.noreply.github.com>
 Clark Saben <76020733+csaben@users.noreply.github.com>
 Clint Herron <hanclinto@gmail.com>
+CrispStrobe <154636388+CrispStrobe@users.noreply.github.com>
 Cuong Trinh Manh <nguoithichkhampha@gmail.com>
 DAN™ <dranger003@gmail.com>
 Damian Stewart <d@damianstewart.com>
@@ -95,8 +116,12 @@ Daniel Bevenius <daniel.bevenius@gmail.com>
 Daniel Drake <drake@endlessos.org>
 Daniel Hiltgen <dhiltgen@users.noreply.github.com>
 Daniel Illescas Romero <illescas.daniel@protonmail.com>
+Daniele <57776841+daniandtheweb@users.noreply.github.com>
 DannyDaemonic <DannyDaemonic@gmail.com>
 Dat Quoc Nguyen <2412555+datquocnguyen@users.noreply.github.com>
+Dave <dave-fl@users.noreply.github.com>
+Dave Airlie <airlied@gmail.com>
+Dave Airlie <airlied@redhat.com>
 Dave Della Costa <ddellacosta+github@gmail.com>
 David Friehs <david@friehs.info>
 David Kennedy <dakennedyd@gmail.com>
@@ -104,10 +129,13 @@ David Pflug <david@pflug.email>
 David Renshaw <dwrenshaw@gmail.com>
 David Sommers <12738+databyte@users.noreply.github.com>
 David Yang <davidyang6us@gmail.com>
+Dawid Potocki <github@dawidpotocki.com>
 Dawid Wysocki <62249621+TortillaZHawaii@users.noreply.github.com>
 Dean <Dean.Sinaean@gmail.com>
 Deins <deinsegle@gmail.com>
+Deven Mistry <31466137+deven367@users.noreply.github.com>
 Didzis Gosko <didzis@users.noreply.github.com>
+Djip007 <djip.perois@free.fr>
 Don Mahurin <dmahurin@users.noreply.github.com>
 DooWoong Lee (David) <manics99@naver.com>
 Doomsdayrs <38189170+Doomsdayrs@users.noreply.github.com>
@@ -116,8 +144,11 @@ Dr. Tom Murphy VII Ph.D <499244+tom7@users.noreply.github.com>
 Ebey Abraham <ebey97@gmail.com>
 Ed Lee <edilee@mozilla.com>
 Ed Lepedus <ed.lepedus@googlemail.com>
+Eddie-Wang <wangjinheng1120@163.com>
 Edward Taylor <edeetee@gmail.com>
+Elaine <elaine.zosa@gmail.com>
 Elbios <141279586+Elbios@users.noreply.github.com>
+Elton Kola <eltonkola@gmail.com>
 Engininja2 <139037756+Engininja2@users.noreply.github.com>
 Equim <sayaka@ekyu.moe>
 Eric Sommerlade <es0m@users.noreply.github.com>
@@ -143,37 +174,47 @@ Firat <firatkiral@gmail.com>
 Folko-Ven <71110216+Folko-Ven@users.noreply.github.com>
 Foul-Tarnished <107711110+Foul-Tarnished@users.noreply.github.com>
 Francisco Melo <43780565+francis2tm@users.noreply.github.com>
+Frank Mai <thxcode0824@gmail.com>
 FrankHB <frankhb1989@gmail.com>
+Fred Douglas <43351173+fredlas@users.noreply.github.com>
 Frederik Vogel <Schaltfehler@users.noreply.github.com>
 Gabe Goodhart <gabe.l.hart@gmail.com>
 GainLee <perfecter.gen@gmail.com>
 Galunid <karolek1231456@gmail.com>
 Gary Linscott <glinscott@gmail.com>
 Gary Mulder <gjmulder@gmail.com>
+Gavin Zhao <gavinzhaojw@protonmail.com>
 Genkagaku.GPT <hlhr202@163.com>
 Georgi Gerganov <ggerganov@gmail.com>
 Gilad S <giladgd@users.noreply.github.com>
+Giuseppe Scrivano <giuseppe@scrivano.org>
 GiviMAD <GiviMAD@users.noreply.github.com>
 Govlzkoy <gotope@users.noreply.github.com>
 Guillaume "Vermeille" Sanchez <Guillaume.V.Sanchez@gmail.com>
 Guillaume Wenzek <gwenzek@users.noreply.github.com>
 Guoteng <32697156+SolenoidWGT@users.noreply.github.com>
 Gustavo Rocha Dias <91472747+gustrd@users.noreply.github.com>
+Haggai Nuchi <h.nuchi@gmail.com>
 Halalaluyafail3 <55773281+Halalaluyafail3@users.noreply.github.com>
+Hamdoud Hakem <90524568+hamdoudhakem@users.noreply.github.com>
+HanishKVC <hanishkvc@gmail.com>
 Haohui Mai <ricetons@gmail.com>
 Haoxiang Fei <tonyfettes@tonyfettes.com>
 Harald Fernengel <harald.fernengel@here.com>
 Hatsune Miku <129688334+at8u@users.noreply.github.com>
+HatsuneMikuUwU33 <173229399+HatsuneMikuUwU33@users.noreply.github.com>
 Henk Poley <HenkPoley@gmail.com>
 Henri Vasserman <henv@hot.ee>
 Henrik Forstén <henrik.forsten@gmail.com>
 Herman Semenov <GermanAizek@yandex.ru>
 Hesen Peng <hesen.peng@gmail.com>
 Hoang Nguyen <hugo53@users.noreply.github.com>
+Hong Bo PENG <penghb@cn.ibm.com>
 Hongyu Ouyang <96765450+casavaca@users.noreply.github.com>
 Howard Su <howard0su@gmail.com>
 Hua Jiang <allenhjiang@outlook.com>
 Huawei Lin <huaweilin.cs@gmail.com>
+Hugo Roussel <hugo.rous@gmail.com>
 Ian Bull <irbull@eclipsesource.com>
 Ian Bull <irbull@gmail.com>
 Ian Scrivener <github@zilogy.asia>
@@ -190,8 +231,10 @@ Ivan Stepanov <ivanstepanovftw@gmail.com>
 JH23X <165871467+JH23X@users.noreply.github.com>
 Jack Mousseau <jmousseau@users.noreply.github.com>
 JackJollimore <130917767+JackJollimore@users.noreply.github.com>
+Jaemin Son <woalsdnd@gmail.com>
 Jag Chadha <jagtesh@gmail.com>
 Jakub N <jakubniemczyk97@gmail.com>
+James A Capozzoli <157492257+jac-jim@users.noreply.github.com>
 James Reynolds <magnusviri@users.noreply.github.com>
 Jan Boon <jan.boon@kaetemi.be>
 Jan Boon <kaetemi@gmail.com>
@@ -205,12 +248,17 @@ Jean-Michaël Celerier <jeanmichael.celerier+github@gmail.com>
 Jed Fox <git@jedfox.com>
 Jeffrey Quesnelle <emozilla@nousresearch.com>
 Jesse Jojo Johnson <williamsaintgeorge@gmail.com>
+Jeximo <jeximo@gmail.com>
 Jhen-Jie Hong <iainst0409@gmail.com>
 Jiahao Li <liplus17@163.com>
 Jian Liao <jianliao@users.noreply.github.com>
 JidongZhang-THU <1119708529@qq.com>
 Jinwoo Jeong <33892306+williamjeong2@users.noreply.github.com>
 Jiří Podivín <66251151+jpodivin@users.noreply.github.com>
+Jiří Sejkora <Sejseloid@gmail.com>
+Joan Fontanals <jfontanalsmartinez@gmail.com>
+Joan Fontanals <joan.fontanals.martinez@jina.ai>
+Johan <JohanAR@users.noreply.github.com>
 Johannes Gäßler <johannesg@5d6.de>
 Johannes Rudolph <johannes.rudolph@gmail.com>
 John <78893154+cmp-nct@users.noreply.github.com>
@@ -221,15 +269,19 @@ Jonas Wunderlich <32615971+jonas-w@users.noreply.github.com>
 Jorge A <161275481+jorgealias@users.noreply.github.com>
 Jose Maldonado <63384398+yukiteruamano@users.noreply.github.com>
 Joseph Stahl <1269177+josephst@users.noreply.github.com>
+Josh Ramer <josh.ramer@icloud.com>
 Joyce <joycebrum@google.com>
 Juan Calderon-Perez <835733+gaby@users.noreply.github.com>
 Judd <foldl@users.noreply.github.com>
 Julius Arkenberg <arki05@users.noreply.github.com>
 Jun Jie <71215065+junnjiee16@users.noreply.github.com>
+Junyang Lin <justinlin930319@hotmail.com>
 Juraj Bednar <juraj@bednar.io>
 Justin Parker <jparkerweb@gmail.com>
 Justin Suess <justin.suess@westpoint.edu>
+Justina Cho <justcho5@gmail.com>
 Justine Tunney <jtunney@gmail.com>
+Justine Tunney <jtunney@mozilla.com>
 Juuso Alasuutari <juuso.alasuutari@gmail.com>
 KASR <karim.asrih@gmail.com>
 Kamil Tomšík <info@tomsik.cz>
@@ -242,6 +294,7 @@ Kawrakow <48489457+ikawrakow@users.noreply.github.com>
 Keiichi Tabata <keiichi.tabata@outlook.com>
 Kenvix ⭐ <kenvixzure@live.com>
 Kerfuffle <44031344+KerfuffleV2@users.noreply.github.com>
+Kevin Gibbons <bakkot@gmail.com>
 Kevin Ji <1146876+kevinji@users.noreply.github.com>
 Kevin Kwok <antimatter15@gmail.com>
 Kevin Lo <kevlo@kevlo.org>
@@ -257,6 +310,7 @@ Laura <Tijntje_7@msn.com>
 Lee <44310445+lx200916@users.noreply.github.com>
 Lee Drake <b.lee.drake@gmail.com>
 Leng Yue <lengyue@lengyue.me>
+Leon Knauer <git@leonknauer.com>
 LeonEricsson <70749762+LeonEricsson@users.noreply.github.com>
 Leonardo Neumann <leonardo@neumann.dev.br>
 Li Tan <tanliboy@gmail.com>
@@ -265,20 +319,26 @@ LoganDark <github@logandark.mozmail.com>
 LostRuins <39025047+LostRuins@users.noreply.github.com>
 Luciano <lucianostrika44@gmail.com>
 Luo Tian <lt@basecity.com>
+Lyle Dean <dean@lyle.dev>
 M. Yusuf Sarıgöz <yusufsarigoz@gmail.com>
 Maarten ter Huurne <maarten@treewalker.org>
 Mack Straight <eiz@users.noreply.github.com>
 Maël Kerbiriou <m431.kerbiriou@gmail.com>
 MaggotHATE <clay1326@gmail.com>
+Manuel <44313466+makuche@users.noreply.github.com>
 Marc Köhlbrugge <subscriptions@marckohlbrugge.com>
 Marco Matthies <71844+marcom@users.noreply.github.com>
 Marcus Dunn <51931484+MarcusDunn@users.noreply.github.com>
 Marian Cepok <marian.cepok@gmail.com>
 Mark Fairbairn <thebaron88@gmail.com>
 Marko Tasic <mtasic85@gmail.com>
+Markus Tavenrath <mtavenrath@users.noreply.github.com>
+Martin Delille <martin@delille.org>
 Martin Krasser <krasserm@googlemail.com>
 Martin Schwaighofer <mschwaig@users.noreply.github.com>
 Marvin Gießing <marvin.giessing@gmail.com>
+Masaya, Kato <62578291+msy-kato@users.noreply.github.com>
+MasterYi1024 <39848311+MasterYi1024@users.noreply.github.com>
 Mateusz Charytoniuk <mateusz.charytoniuk@protonmail.com>
 Matheus C. França <matheus-catarino@hotmail.com>
 Matheus Gabriel Alves Silva <matheusgasource@gmail.com>
@@ -287,8 +347,11 @@ Mathijs de Bruin <mathijs@mathijsfietst.nl>
 Matt Clayton <156335168+mattjcly@users.noreply.github.com>
 Matt Pulver <matt.pulver@heavy.ai>
 Matteo Boschini <12133566+mbosc@users.noreply.github.com>
+Mattheus Chediak <shammcity00@gmail.com>
 Matthew Tejo <matthew.tejo@gmail.com>
 Matvey Soloviev <blackhole89@gmail.com>
+Max Krasnyansky <max.krasnyansky@gmail.com>
+Max Krasnyansky <quic_maxk@quicinc.com>
 Maxime <672982+maximegmd@users.noreply.github.com>
 Maximilian Winter <maximilian.winter.91@gmail.com>
 Meng Zhang <meng@tabbyml.com>
@@ -300,32 +363,41 @@ Michael Kesper <mkesper@schokokeks.org>
 Michael Klimenko <mklimenko29@gmail.com>
 Michael Podvitskiy <podvitskiymichael@gmail.com>
 Michael Potter <NanoTekGuy@Gmail.com>
+Michael de Gans <michael.john.degans@gmail.com>
 Michaël de Vries <vriesdemichael@gmail.com>
 Mihai <mihai.chirculescu@yahoo.com>
 Mike <ytianhui2004@gmail.com>
+Mikko Juola <mikjuo@gmail.com>
 Minsoo Cheong <54794500+mscheong01@users.noreply.github.com>
 Mirko185 <mirkosig@gmail.com>
 Mirror Azure <54669636+MirrorAzure@users.noreply.github.com>
 Miwa / Ensan <63481257+ensan-hcl@users.noreply.github.com>
 Mohammadreza Hendiani <hendiani.mohammadreza@gmail.com>
+Mohammadreza Hendiani <mohammad.r.hendiani@gmail.com>
 Murilo Santana <mvrilo@gmail.com>
 Musab Gultekin <musabgultekin@users.noreply.github.com>
 Nam D. Tran <42194884+namtranase@users.noreply.github.com>
+Nathan Epstein <nate2@umbc.edu>
 NawafAlansari <72708095+NawafAlansari@users.noreply.github.com>
 Nebula <infinitewormhole@gmail.com>
+Neo Zhang <14088817+arthw@users.noreply.github.com>
+Neo Zhang <zhang.jianyu@outlook.com>
 Neo Zhang Jianyu <jianyu.zhang@intel.com>
 Neuman Vong <neuman.vong@gmail.com>
 Nexesenex <124105151+Nexesenex@users.noreply.github.com>
 Niall Coates <1349685+Niall-@users.noreply.github.com>
 Nicolai Weitkemper <kontakt@nicolaiweitkemper.de>
+Nicolás Pérez <nicolas_perez@brown.edu>
 Nigel Bosch <pnigelb@gmail.com>
 Niklas Korz <niklas@niklaskorz.de>
+Nikolas <127742645+nneubacher@users.noreply.github.com>
 Nindaleth <Nindaleth@users.noreply.github.com>
 Oleksandr Nikitin <oleksandr@tvori.info>
 Oleksii Maryshchenko <oleksii.maryshchenko@gmail.com>
 Olivier Chafik <ochafik@users.noreply.github.com>
 Ondřej Čertík <ondrej@certik.us>
 Ouadie EL FAROUKI <ouadie.elfarouki@codeplay.com>
+Patrice Ferlet <metal3d@gmail.com>
 Paul Tsochantaris <ptsochantaris@icloud.com>
 Pavol Rusnak <pavol@rusnak.io>
 Pedro Cuenca <pedro@huggingface.co>
@@ -343,9 +415,14 @@ RJ Adriaansen <adriaansen@eshcc.eur.nl>
 Radoslav Gerganov <rgerganov@gmail.com>
 Radosław Gryta <radek.gryta@gmail.com>
 Rahul Vivek Nair <68507071+RahulVivekNair@users.noreply.github.com>
+Raj Hammeer Singh Hada <hammeerraj@gmail.com>
+Ralph Soika <ralph.soika@imixs.com>
 Rand Xie <randxiexyy29@gmail.com>
 Randall Fitzgerald <randall@dasaku.net>
 Reinforce-II <fate@eastal.com>
+Ren Xuancheng <jklj077@users.noreply.github.com>
+Rene Leonhardt <65483435+reneleonhardt@users.noreply.github.com>
+RhinoDevel <RhinoDevel@users.noreply.github.com>
 Riceball LEE <snowyu.lee@gmail.com>
 Richard Kiss <him@richardkiss.com>
 Richard Roberson <richardr1126@gmail.com>
@@ -373,6 +450,7 @@ Rowan Hart <rowanbhart@gmail.com>
 Rune <43761327+Rune-AI@users.noreply.github.com>
 Ryan Landay <rlanday@gmail.com>
 Ryder Wishart <ryderwishart@gmail.com>
+Ryuei <louixs@users.noreply.github.com>
 Rőczey Barnabás <31726601+An0nie@users.noreply.github.com>
 SakuraUmi <yukinon244@gmail.com>
 Salvador E. Tropea <stropea@inti.gob.ar>
@@ -386,6 +464,7 @@ SebastianApel <13675545+SebastianApel@users.noreply.github.com>
 Senemu <10880819+Senemu@users.noreply.github.com>
 Sergey Alirzaev <zl29ah@gmail.com>
 Sergio López <slp@sinrega.org>
+Sertaç Özercan <852750+sozercan@users.noreply.github.com>
 SeungWon Jeong <65549245+redlion0929@users.noreply.github.com>
 ShadovvBeast <ShadovvBeast@gmail.com>
 Shakhar Dasgupta <shakhardasgupta@gmail.com>
@@ -394,6 +473,7 @@ Shijie <821898965@qq.com>
 Shintarou Okada <kokuzen@gmail.com>
 Shouzheng Liu <61452103+lshzh-ww@users.noreply.github.com>
 Shouzheng Liu <lshzh.hi@gmail.com>
+Shuichi Tsutsumi <shuichi0526@gmail.com>
 Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
 Simon Willison <swillison@gmail.com>
 Siwen Yu <yusiwen@gmail.com>
@@ -405,11 +485,14 @@ Someone <sergei.kozlukov@aalto.fi>
 Someone Serge <sergei.kozlukov@aalto.fi>
 Sourab Mangrulkar <13534540+pacman100@users.noreply.github.com>
 Spencer Sutton <spencersutton@users.noreply.github.com>
+Srihari-mcw <96763064+Srihari-mcw@users.noreply.github.com>
 Srinivas Billa <nivibilla@gmail.com>
 Stefan Sydow <stefan@sydow.email>
+Steffen Röcker <sroecker@gmail.com>
 Stephan Walter <stephan@walter.name>
 Stephen Nichols <snichols@users.noreply.github.com>
 Steve Grubb <ausearch.1@gmail.com>
+Steven Prichard <spprichard20@gmail.com>
 Steven Roussey <sroussey@gmail.com>
 Steward Garcia <57494570+FSSRepo@users.noreply.github.com>
 Suaj Carrot <72162667+SuajCarrot@users.noreply.github.com>
@@ -434,16 +517,19 @@ Tom C <tom.corelis@gmail.com>
 Tom Jobbins <784313+TheBloke@users.noreply.github.com>
 Tomas <tom.tomas.36478119@gmail.com>
 Tomáš Pazdiora <tomas.pazdiora@gmail.com>
+Tristan Druyen <tristan@vault81.mozmail.com>
 Tristan Ross <rosscomputerguy@protonmail.com>
 Tungsten842 <886724vf@anonaddy.me>
 Tungsten842 <quantmint@protonmail.com>
 Tushar <ditsuke@protonmail.com>
 UEXTM.com <84163508+uextm@users.noreply.github.com>
+Ulrich Drepper <drepper@gmail.com>
 Uzo Nweke <uzoechi@gmail.com>
 Vaibhav Srivastav <vaibhavs10@gmail.com>
 Val Kharitonov <mail@kharvd.com>
 Valentin Konovalov <valle.ketsujin@gmail.com>
 Valentyn Bezshapkin <61702053+valentynbez@users.noreply.github.com>
+Victor Nogueira <felladrin@gmail.com>
 Victor Z. Peng <ziliangdotme@gmail.com>
 Vlad <spitfireage@gmail.com>
 Vladimir <bogdad@gmail.com>
@@ -455,7 +541,9 @@ Weird Constructor <weirdconstructor@gmail.com>
 Welby Seely <welbyseely@gmail.com>
 Wentai Zhang <rchardx@gmail.com>
 WillCorticesAI <150854901+WillCorticesAI@users.noreply.github.com>
+William Tambellini <william.tambellini@gmail.com>
 Willy Tarreau <w@1wt.eu>
+Wouter <9594229+DifferentialityDevelopment@users.noreply.github.com>
 Wu Jian Ping <wujjpp@hotmail.com>
 Wu Jian Ping <wujp@greatld.com>
 Xiake Sun <xiake.sun@intel.com>
@@ -466,6 +554,8 @@ Xiaoyi Chen <cxychina@gmail.com>
 Xingchen Song(宋星辰) <xingchensong1996@163.com>
 Xuan Son Nguyen <thichthat@gmail.com>
 Yann Follet <131855179+YannFollet@users.noreply.github.com>
+Yaroslav <yaroslav.yashin@me.com>
+Yazan Agha-Schrader <mountaiin@icloud.com>
 Yiming Cui <conandiy@vip.qq.com>
 Yishuo Wang <MeouSker77@outlook.com>
 Yueh-Po Peng <94939112+y10ab1@users.noreply.github.com>
@@ -477,6 +567,7 @@ Zane Shannon <z@zcs.me>
 Zay <95888118+isaiahbjork@users.noreply.github.com>
 Zenix <zenixls2@gmail.com>
 Zhang Peiyuan <a1286225768@gmail.com>
+Zheng.Deng <32841220+dengzheng-cloud@users.noreply.github.com>
 ZhouYuChen <zhouyuchen@naver.com>
 Ziad Ben Hadj-Alouane <zied.benhadjalouane@gmail.com>
 Ziang Wu <97337387+ZiangWu-77@users.noreply.github.com>
@@ -484,14 +575,18 @@ Zsapi <martin1.zsapka@gmail.com>
 a-n-n-a-l-e-e <150648636+a-n-n-a-l-e-e@users.noreply.github.com>
 adel boussaken <netdur@gmail.com>
 afrideva <95653597+afrideva@users.noreply.github.com>
+agray3 <agray3@users.noreply.github.com>
 akawrykow <142945436+akawrykow@users.noreply.github.com>
 alexpinel <93524949+alexpinel@users.noreply.github.com>
 alonfaraj <alonfaraj@gmail.com>
+alwqx <kenan3015@gmail.com>
+amd-lalithnc <lalithnc@amd.com>
 andrijdavid <david@geek.mg>
 anon998 <131767832+anon998@users.noreply.github.com>
 anzz1 <anzz1@live.com>
 apaz <aarpazdera@gmail.com>
 apcameron <37645737+apcameron@users.noreply.github.com>
+arch-btw <57669023+arch-btw@users.noreply.github.com>
 arcrank <arcrank@gmail.com>
 arlo-phoenix <140345165+arlo-phoenix@users.noreply.github.com>
 at8u <129688334+at8u@users.noreply.github.com>
@@ -514,13 +609,17 @@ cocktailpeanut <121128867+cocktailpeanut@users.noreply.github.com>
 coezbek <c.oezbek@gmail.com>
 comex <comexk@gmail.com>
 compilade <113953597+compilade@users.noreply.github.com>
+compilade <git@compilade.net>
+cpumaxx <163466046+cpumaxx@users.noreply.github.com>
 crasm <crasm@git.vczf.net>
 crasm <crasm@git.vczf.us>
 daboe01 <daboe01@googlemail.com>
 david raistrick <keen99@users.noreply.github.com>
+ddh0 <dylanhalladay02@icloud.com>
 ddpasa <112642920+ddpasa@users.noreply.github.com>
 deepdiffuser <112834445+deepdiffuser@users.noreply.github.com>
 divinity76 <divinity76@gmail.com>
+dm4 <sunrisedm4@gmail.com>
 dotpy314 <33351922+dotpy314@users.noreply.github.com>
 drbh <david.richard.holtz@gmail.com>
 ds5t5 <145942675+ds5t5@users.noreply.github.com>
@@ -529,6 +628,7 @@ eastriver <lee@eastriver.dev>
 ebraminio <ebraminio@gmail.com>
 eiery <19350831+eiery@users.noreply.github.com>
 eric8607242 <e0928021388@gmail.com>
+fairydreaming <166155368+fairydreaming@users.noreply.github.com>
 fraxy-v <65565042+fraxy-v@users.noreply.github.com>
 github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 gliptic <gliptic@users.noreply.github.com>
@@ -539,6 +639,7 @@ h-h-h-h <13482553+h-h-h-h@users.noreply.github.com>
 hankcs <cnhankmc@gmail.com>
 hoangmit <hoangmit@users.noreply.github.com>
 hongbo.mo <352280764@qq.com>
+hopkins385 <98618192+hopkins385@users.noreply.github.com>
 howlger <eclipse@voormann.de>
 howlger <github@voormann.de>
 hutli <6594598+hutli@users.noreply.github.com>
@@ -549,14 +650,22 @@ hydai <z54981220@gmail.com>
 iSma <ismail.senhaji@gmail.com>
 iacore <74560659+iacore@users.noreply.github.com>
 igarnier <igarnier@protonmail.com>
+intelmatt <61025942+intelmatt@users.noreply.github.com>
 iohub <rickyang.pro@gmail.com>
 jacobi petrucciani <8117202+jpetrucciani@users.noreply.github.com>
+jaime-m-p <167997752+jaime-m-p@users.noreply.github.com>
 jameswu2014 <545426914@qq.com>
+jiez <373447296@qq.com>
 jneem <joeneeman@gmail.com>
+joecryptotoo <80373433+joecryptotoo@users.noreply.github.com>
 johnson442 <56517414+johnson442@users.noreply.github.com>
+jojorne <jojorne@users.noreply.github.com>
 jon-chuang <9093549+jon-chuang@users.noreply.github.com>
 jp-x-g <jpxg-dev@protonmail.com>
+jukofyork <69222624+jukofyork@users.noreply.github.com>
+junchao-loongson <68935141+junchao-loongson@users.noreply.github.com>
 jwj7140 <32943891+jwj7140@users.noreply.github.com>
+k.h.lai <adrian.k.h.lai@outlook.com>
 kaizau <kaizau@users.noreply.github.com>
 kalomaze <66376113+kalomaze@users.noreply.github.com>
 kang <tpdns9032100@gmail.com>
@@ -575,11 +684,15 @@ ldwang <ftgreat@163.com>
 le.chang <cljs118@126.com>
 leejet <leejet714@gmail.com>
 limitedAtonement <limitedAtonement@users.noreply.github.com>
+liuwei-git <14815172+liuwei-git@users.noreply.github.com>
 lon <114724657+longregen@users.noreply.github.com>
+loonerin <132926317+loonerin@users.noreply.github.com>
+luoyu-intel <yu.luo@intel.com>
 m3ndax <adrian.goessl@outlook.com>
 maddes8cht <55592906+maddes8cht@users.noreply.github.com>
 makomk <makosoft@googlemail.com>
 manikbhandari <mbbhandarimanik2@gmail.com>
+maor-ps <154728172+maor-ps@users.noreply.github.com>
 mdrokz <mohammadmunshi@gmail.com>
 mgroeber9110 <45620825+mgroeber9110@users.noreply.github.com>
 minarchist <minarchist@users.noreply.github.com>
@@ -593,15 +706,19 @@ ngc92 <7938269+ngc92@users.noreply.github.com>
 nhamanasu <45545786+nhamanasu@users.noreply.github.com>
 niansa/tuxifan <anton-sa@web.de>
 niansa/tuxifan <tuxifan@posteo.de>
+nickp27 <nb.porter@gmail.com>
 ningshanwutuobang <ningshanwutuobang@gmail.com>
 nold <Nold360@users.noreply.github.com>
 nopperl <54780682+nopperl@users.noreply.github.com>
 nusu-github <29514220+nusu-github@users.noreply.github.com>
 olexiyb <olexiyb@gmail.com>
+omahs <73983677+omahs@users.noreply.github.com>
 oobabooga <112222186+oobabooga@users.noreply.github.com>
 opparco <parco.opaai@gmail.com>
 ostix360 <55257054+ostix360@users.noreply.github.com>
+pengxin99 <pengxin.yuan@intel.com>
 perserk <perserk@gmail.com>
+pmysl <piotr.myslinski@outlook.com>
 postmasters <namnguyen@google.com>
 pudepiedj <pudepiedj@gmail.com>
 qingfengfenga <41416092+qingfengfenga@users.noreply.github.com>
@@ -614,16 +731,19 @@ rhuddleston <ryan.huddleston@percona.com>
 rimoliga <53384203+rimoliga@users.noreply.github.com>
 runfuture <runfuture@users.noreply.github.com>
 sandyiscool <sandyiscool@gmail.com>
+sasha0552 <admin@sasha0552.org>
 semidark <me@semidark.net>
 sharpHL <132747147+sharpHL@users.noreply.github.com>
 shibe2 <shibe@tuta.io>
 singularity <12184989+singularity-s0@users.noreply.github.com>
 sjinzh <sjinzh@gmail.com>
+sjxx <63994076+ylsdamxssjxxdd@users.noreply.github.com>
 slaren <2141330+slaren@users.noreply.github.com>
 slaren <slarengh@gmail.com>
 snadampal <87143774+snadampal@users.noreply.github.com>
 staviq <staviq@gmail.com>
 stduhpf <stephduh@live.fr>
+strawberrymelonpanda <152940198+strawberrymelonpanda@users.noreply.github.com>
 swittk <switt1995@gmail.com>
 takov751 <40316768+takov751@users.noreply.github.com>
 tarcey <cey.tarik@gmail.com>
@@ -636,12 +756,16 @@ uint256_t <konndennsa@gmail.com>
 uint256_t <maekawatoshiki1017@gmail.com>
 unbounded <haakon@likedan.net>
 valiray <133289098+valiray@users.noreply.github.com>
+vik <vikhyatk@gmail.com>
+viric <viric@viric.name>
 vodkaslime <646329483@qq.com>
 vvhg1 <94630311+vvhg1@users.noreply.github.com>
 vxiiduu <73044267+vxiiduu@users.noreply.github.com>
 wbpxre150 <100937007+wbpxre150@users.noreply.github.com>
 whoreson <139810751+whoreson@users.noreply.github.com>
+woachk <24752637+woachk@users.noreply.github.com>
 wonjun Jang <strutive07@gmail.com>
+woodx <124784234+woodx9@users.noreply.github.com>
 wzy <32936898+Freed-Wu@users.noreply.github.com>
 xaedes <xaedes@gmail.com>
 xaedes <xaedes@googlemail.com>
@@ -649,7 +773,10 @@ xloem <0xloem@gmail.com>
 yangli2 <yangli2@gmail.com>
 yuiseki <yuiseki@gmail.com>
 zakkor <edward.partenie@gmail.com>
+zhangkaihuo <zhangkaihuo@gmail.com>
 zhouwg <6889919+zhouwg@users.noreply.github.com>
+zhouwg <zhouwg2000@gmail.com>
 zrm <trustiosity.zrm@gmail.com>
+Ștefan-Gabriel Muscalu <legraphista@users.noreply.github.com>
 源文雨 <41315874+fumiama@users.noreply.github.com>
 Нияз Гарифзянов <112617865+garrnizon@users.noreply.github.com>

CMakePresets.json

@@ -1,4 +1,4 @@
-{
+{
   "version": 4,
   "configurePresets": [
     {
@@ -11,10 +11,23 @@
             "CMAKE_INSTALL_RPATH": "$ORIGIN;$ORIGIN/.."
         }
     },
-
+    {
+        "name": "sycl-base",
+        "hidden": true,
+        "generator": "Ninja",
+        "binaryDir": "${sourceDir}/build-${presetName}",
+        "cacheVariables": {
+            "CMAKE_EXPORT_COMPILE_COMMANDS": "ON",
+            "CMAKE_CXX_COMPILER": "icx",
+            "CMAKE_C_COMPILER": "cl",
+            "GGML_SYCL": "ON",
+            "CMAKE_INSTALL_RPATH": "$ORIGIN;$ORIGIN/.."
+        }
+    },
     { "name": "debug",   "hidden": true, "cacheVariables": { "CMAKE_BUILD_TYPE": "Debug" } },
-    { "name": "release", "hidden": true, "cacheVariables": { "CMAKE_BUILD_TYPE": "RelWithDebInfo" } },
-    { "name": "static",  "hidden": true, "cacheVariables": { "LLAMA_STATIC": "ON" } },
+    { "name": "release", "hidden": true, "cacheVariables": { "CMAKE_BUILD_TYPE": "Release" } },
+    { "name": "reldbg",  "hidden": true, "cacheVariables": { "CMAKE_BUILD_TYPE": "RelWithDebInfo" } },
+    { "name": "static",  "hidden": true, "cacheVariables": { "GGML_STATIC": "ON" } },
 
     {
         "name": "arm64-windows-msvc", "hidden": true,
@@ -35,11 +48,18 @@
     },
 
     { "name": "arm64-windows-llvm-debug"  , "inherits": [ "base", "arm64-windows-llvm",  "debug"   ] },
-    { "name": "arm64-windows-llvm-release", "inherits": [ "base", "arm64-windows-llvm",  "release" ] },
-    { "name": "arm64-windows-llvm+static-release", "inherits": [ "base", "arm64-windows-llvm",  "release", "static" ] },
+    { "name": "arm64-windows-llvm-release", "inherits": [ "base", "arm64-windows-llvm",  "reldbg" ] },
+    { "name": "arm64-windows-llvm+static-release", "inherits": [ "base", "arm64-windows-llvm",  "reldbg", "static" ] },
 
     { "name": "arm64-windows-msvc-debug"  , "inherits": [ "base", "arm64-windows-msvc",  "debug"   ] },
-    { "name": "arm64-windows-msvc-release", "inherits": [ "base", "arm64-windows-msvc",  "release" ] },
-    { "name": "arm64-windows-msvc+static-release", "inherits": [ "base", "arm64-windows-msvc",  "release", "static" ] }
+    { "name": "arm64-windows-msvc-release", "inherits": [ "base", "arm64-windows-msvc",  "reldbg" ] },
+    { "name": "arm64-windows-msvc+static-release", "inherits": [ "base", "arm64-windows-msvc",  "reldbg", "static" ] },
+
+    { "name": "x64-windows-msvc-debug"  , "inherits": [ "base", "debug"   ] },
+    { "name": "x64-windows-msvc-release", "inherits": [ "base", "reldbg" ] },
+    { "name": "x64-windows-msvc+static-release", "inherits": [ "base", "reldbg", "static" ] },
+
+    { "name": "x64-windows-sycl-debug"  , "inherits": [ "sycl-base", "debug"   ] },
+    { "name": "x64-windows-sycl-release", "inherits": [ "sycl-base", "release" ] }
   ]
 }

CONTRIBUTING.md

@@ -0,0 +1,14 @@
+# Contributing Guidelines
+
+## Checklist
+
+* Make sure your PR follows the [coding guidelines](https://github.com/ggerganov/llama.cpp/blob/master/README.md#coding-guidelines)
+* Test your changes using the commands in the [`tests`](tests) folder. For instance, running the `./tests/test-backend-ops` command tests different backend implementations of the GGML library
+* Execute [the full CI locally on your machine](ci/README.md) before publishing
+
+## PR formatting
+
+* Please rate the complexity of your PR (i.e. `Review Complexity : Low`, `Review Complexity : Medium`, `Review Complexity : High`). This makes it easier for maintainers to triage the PRs.
+    - The PR template has a series of review complexity checkboxes `[ ]` that you can mark as `[X]` for your conveience. Refer to [About task lists](https://docs.github.com/en/get-started/writing-on-github/working-with-advanced-formatting/about-task-lists) for more information.
+* If the pull request only contains documentation changes (e.g., updating READMEs, adding new wiki pages), please add `[no ci]` to the commit title. This will skip unnecessary CI checks and help reduce build times.
+* When squashing multiple commits on merge, use the following format for your commit title: `<module> : <commit title> (#<issue_number>)`. For example: `utils : Fix typo in utils.py (#1234)`

Package.swift

@@ -3,14 +3,13 @@
 import PackageDescription
 
 var sources = [
-    "ggml.c",
-    "sgemm.cpp",
-    "llama.cpp",
-    "unicode.cpp",
-    "unicode-data.cpp",
-    "ggml-alloc.c",
-    "ggml-backend.c",
-    "ggml-quants.c",
+    "src/llama.cpp",
+    "src/unicode.cpp",
+    "src/unicode-data.cpp",
+    "ggml/src/ggml.c",
+    "ggml/src/ggml-alloc.c",
+    "ggml/src/ggml-backend.c",
+    "ggml/src/ggml-quants.c",
 ]
 
 var resources: [Resource] = []
@@ -26,8 +25,8 @@ var cSettings: [CSetting] =  [
 ]
 
 #if canImport(Darwin)
-sources.append("ggml-metal.m")
-resources.append(.process("ggml-metal.metal"))
+sources.append("ggml/src/ggml-metal.m")
+resources.append(.process("ggml/src/ggml-metal.metal"))
 linkerSettings.append(.linkedFramework("Accelerate"))
 cSettings.append(
     contentsOf: [
@@ -63,8 +62,6 @@ let package = Package(
                "models",
                "tests",
                "CMakeLists.txt",
-               "ggml-cuda.cu",
-               "ggml-cuda.h",
                "Makefile"
             ],
             sources: sources,

README-sycl.md

@@ -1,6 +1,7 @@
 # llama.cpp for SYCL
 
 - [Background](#background)
+- [Recommended Release](#recommended-release)
 - [News](#news)
 - [OS](#os)
 - [Hardware](#hardware)
@@ -29,10 +30,25 @@ The llama.cpp SYCL backend is designed to support **Intel GPU** firstly. Based o
 
 When targeting **Intel CPU**, it is recommended to use llama.cpp for [Intel oneMKL](README.md#intel-onemkl) backend.
 
-It has the similar design of other llama.cpp BLAS-based paths such as *OpenBLAS, cuBLAS, CLBlast etc..*. In beginning work, the oneAPI's [SYCLomatic](https://github.com/oneapi-src/SYCLomatic) open-source migration tool (Commercial release [Intel® DPC++ Compatibility Tool](https://www.intel.com/content/www/us/en/developer/tools/oneapi/dpc-compatibility-tool.html)) was used for this purpose.
+It has the similar design of other llama.cpp BLAS-based paths such as *OpenBLAS, cuBLAS, etc..*. In beginning work, the oneAPI's [SYCLomatic](https://github.com/oneapi-src/SYCLomatic) open-source migration tool (Commercial release [Intel® DPC++ Compatibility Tool](https://www.intel.com/content/www/us/en/developer/tools/oneapi/dpc-compatibility-tool.html)) was used for this purpose.
+
+## Recommended Release
+
+The SYCL backend would be broken by some PRs due to no online CI.
+
+The following release is verified with good quality:
+
+|Commit ID|Tag|Release|Verified  Platform|
+|-|-|-|-|
+|fb76ec31a9914b7761c1727303ab30380fd4f05c|b3038 |[llama-b3038-bin-win-sycl-x64.zip](https://github.com/ggerganov/llama.cpp/releases/download/b3038/llama-b3038-bin-win-sycl-x64.zip) |Arc770/Linux/oneAPI 2024.1<br>MTL Arc GPU/Windows 11/oneAPI 2024.1|
+
 
 ## News
 
+- 2024.5
+  - Performance is increased: 34 -> 37 tokens/s of llama-2-7b.Q4_0 on Arc770.
+  - Arch Linux is verified successfully.
+
 - 2024.4
   - Support data types: GGML_TYPE_IQ4_NL, GGML_TYPE_IQ4_XS, GGML_TYPE_IQ3_XXS, GGML_TYPE_IQ3_S, GGML_TYPE_IQ2_XXS, GGML_TYPE_IQ2_XS, GGML_TYPE_IQ2_S, GGML_TYPE_IQ1_S, GGML_TYPE_IQ1_M.
 
@@ -54,10 +70,10 @@ It has the similar design of other llama.cpp BLAS-based paths such as *OpenBLAS,
 
 ## OS
 
-| OS      | Status  | Verified                           |
-|---------|---------|------------------------------------|
-| Linux   | Support | Ubuntu 22.04, Fedora Silverblue 39 |
-| Windows | Support | Windows 11                         |
+| OS      | Status  | Verified                                       |
+|---------|---------|------------------------------------------------|
+| Linux   | Support | Ubuntu 22.04, Fedora Silverblue 39, Arch Linux |
+| Windows | Support | Windows 11                                     |
 
 
 ## Hardware
@@ -70,14 +86,14 @@ It has the similar design of other llama.cpp BLAS-based paths such as *OpenBLAS,
 |-------------------------------|---------|---------------------------------------|
 | Intel Data Center Max Series  | Support | Max 1550, 1100                        |
 | Intel Data Center Flex Series | Support | Flex 170                              |
-| Intel Arc Series              | Support | Arc 770, 730M                         |
+| Intel Arc Series              | Support | Arc 770, 730M, Arc A750               |
 | Intel built-in Arc GPU        | Support | built-in Arc GPU in Meteor Lake       |
 | Intel iGPU                    | Support | iGPU in i5-1250P, i7-1260P, i7-1165G7 |
 
 *Notes:*
 
 - **Memory**
-  - The device memory is a limitation when running a large model. The loaded model size, *`llm_load_tensors: buffer_size`*, is displayed in the log when running `./bin/main`.
+  - The device memory is a limitation when running a large model. The loaded model size, *`llm_load_tensors: buffer_size`*, is displayed in the log when running `./bin/llama-cli`.
 
   - Please make sure the GPU shared memory from the host is large enough to account for the model's size. For e.g. the *llama-2-7b.Q4_0* requires at least 8.0GB for integrated GPU and 4.0GB for discrete GPU.
 
@@ -99,14 +115,14 @@ The docker build option is currently limited to *intel GPU* targets.
 ### Build image
 ```sh
 # Using FP16
-docker build -t llama-cpp-sycl --build-arg="LLAMA_SYCL_F16=ON" -f .devops/main-intel.Dockerfile .
+docker build -t llama-cpp-sycl --build-arg="GGML_SYCL_F16=ON" -f .devops/llama-cli-intel.Dockerfile .
 ```
 
 *Notes*:
 
-To build in default FP32 *(Slower than FP16 alternative)*, you can remove the `--build-arg="LLAMA_SYCL_F16=ON"` argument from the previous command.
+To build in default FP32 *(Slower than FP16 alternative)*, you can remove the `--build-arg="GGML_SYCL_F16=ON"` argument from the previous command.
 
-You can also use the `.devops/server-intel.Dockerfile`, which builds the *"server"* alternative.
+You can also use the `.devops/llama-server-intel.Dockerfile`, which builds the *"server"* alternative.
 
 ### Run container
 
@@ -228,10 +244,10 @@ source /opt/intel/oneapi/setvars.sh
 # Build LLAMA with MKL BLAS acceleration for intel GPU
 
 # Option 1: Use FP32 (recommended for better performance in most cases)
-cmake -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
+cmake -B build -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
 
 # Option 2: Use FP16
-cmake -B build -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
+cmake -B build -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON
 
 # build all binary
 cmake --build build --config Release -j -v
@@ -248,10 +264,10 @@ export CPLUS_INCLUDE_DIR=/path/to/oneMKL/include:$CPLUS_INCLUDE_DIR
 # Build LLAMA with Nvidia BLAS acceleration through SYCL
 
 # Option 1: Use FP32 (recommended for better performance in most cases)
-cmake -B build -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
+cmake -B build -DGGML_SYCL=ON -DGGML_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
 
 # Option 2: Use FP16
-cmake -B build -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
+cmake -B build -DGGML_SYCL=ON -DGGML_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON
 
 # build all binary
 cmake --build build --config Release -j -v
@@ -275,7 +291,7 @@ source /opt/intel/oneapi/setvars.sh
 Similar to the native `sycl-ls`, available SYCL devices can be queried as follow:
 
 ```sh
-./build/bin/ls-sycl-device
+./build/bin/llama-ls-sycl-device
 ```
 A example of such log in a system with 1 *intel CPU* and 1 *intel GPU* can look like the following:
 ```
@@ -313,7 +329,7 @@ Examples:
 - Use device 0:
 
 ```sh
-ZES_ENABLE_SYSMAN=1 ./build/bin/main -m models/llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 -sm none -mg 0
+ZES_ENABLE_SYSMAN=1 ./build/bin/llama-cli -m models/llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 -sm none -mg 0
 ```
 or run by script:
 
@@ -324,7 +340,7 @@ or run by script:
 - Use multiple devices:
 
 ```sh
-ZES_ENABLE_SYSMAN=1 ./build/bin/main -m models/llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 -sm layer
+ZES_ENABLE_SYSMAN=1 ./build/bin/llama-cli -m models/llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 -sm layer
 ```
 
 Otherwise, you can run the script:
@@ -394,15 +410,9 @@ Output (example):
 
 4. Install build tools
 
-a. Download & install cmake for Windows: https://cmake.org/download/
+a. Download & install cmake for Windows: https://cmake.org/download/ (CMake can also be installed from Visual Studio Installer)
+b. The new Visual Studio will install Ninja as default. (If not, please install it manually: https://ninja-build.org/)
 
-b. Download & install mingw-w64 make for Windows provided by w64devkit
-
-- Download the 1.19.0 version of [w64devkit](https://github.com/skeeto/w64devkit/releases/download/v1.19.0/w64devkit-1.19.0.zip).
-
-- Extract `w64devkit` on your pc.
-
-- Add the **bin** folder path in the Windows system PATH environment (for e.g. `C:\xxx\w64devkit\bin\`).
 
 ### II. Build llama.cpp
 
@@ -412,10 +422,10 @@ On the oneAPI command line window, step into the llama.cpp main directory and ru
 @call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat" intel64 --force
 
 # Option 1: Use FP32 (recommended for better performance in most cases)
-cmake -B build -G "MinGW Makefiles" -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icx  -DCMAKE_BUILD_TYPE=Release
+cmake -B build -G "Ninja" -DGGML_SYCL=ON -DCMAKE_C_COMPILER=cl -DCMAKE_CXX_COMPILER=icx  -DCMAKE_BUILD_TYPE=Release
 
 # Option 2: Or FP16
-cmake -B build -G "MinGW Makefiles" -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icx  -DCMAKE_BUILD_TYPE=Release -DLLAMA_SYCL_F16=ON
+cmake -B build -G "Ninja" -DGGML_SYCL=ON -DCMAKE_C_COMPILER=cl -DCMAKE_CXX_COMPILER=icx  -DCMAKE_BUILD_TYPE=Release -DGGML_SYCL_F16=ON
 
 cmake --build build --config Release -j
 ```
@@ -425,9 +435,23 @@ Otherwise, run the `win-build-sycl.bat` wrapper which encapsulates the former in
 .\examples\sycl\win-build-sycl.bat
 ```
 
+Or, use CMake presets to build:
+```sh
+cmake --preset x64-windows-sycl-release
+cmake --build build-x64-windows-sycl-release -j --target llama-cli
+
+cmake -DGGML_SYCL_F16=ON --preset x64-windows-sycl-release
+cmake --build build-x64-windows-sycl-release -j --target llama-cli
+
+cmake --preset x64-windows-sycl-debug
+cmake --build build-x64-windows-sycl-debug -j --target llama-cli
+```
+
+Or, you can use Visual Studio to open llama.cpp folder as a CMake project. Choose the sycl CMake presets (`x64-windows-sycl-release` or `x64-windows-sycl-debug`) before you compile the project.
+
 *Notes:*
 
-- By default, calling `make` will build all target binary files. In case of a minimal experimental setup, the user can build the inference executable only through `make main`.
+- In case of a minimal experimental setup, the user can build the inference executable only through `cmake --build build --config Release -j --target llama-cli`.
 
 ### III. Run the inference
 
@@ -488,13 +512,13 @@ Examples:
 - Use device 0:
 
 ```
-build\bin\main.exe -m models\llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e -ngl 33 -s 0 -sm none -mg 0
+build\bin\llama-cli.exe -m models\llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e -ngl 33 -s 0 -sm none -mg 0
 ```
 
 - Use multiple devices:
 
 ```
-build\bin\main.exe -m models\llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e -ngl 33 -s 0 -sm layer
+build\bin\llama-cli.exe -m models\llama-2-7b.Q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e -ngl 33 -s 0 -sm layer
 ```
 Otherwise, run the following wrapper script:
 
@@ -520,9 +544,9 @@ use 1 SYCL GPUs: [0] with Max compute units:512
 
 | Name               | Value                             | Function                                    |
 |--------------------|-----------------------------------|---------------------------------------------|
-| LLAMA_SYCL         | ON (mandatory)                    | Enable build with SYCL code path.           |
-| LLAMA_SYCL_TARGET  | INTEL *(default)* \| NVIDIA       | Set the SYCL target device type.            |
-| LLAMA_SYCL_F16     | OFF *(default)* \|ON *(optional)* | Enable FP16 build with SYCL code path.      |
+| GGML_SYCL          | ON (mandatory)                    | Enable build with SYCL code path.           |
+| GGML_SYCL_TARGET   | INTEL *(default)* \| NVIDIA       | Set the SYCL target device type.            |
+| GGML_SYCL_F16      | OFF *(default)* \|ON *(optional)* | Enable FP16 build with SYCL code path.      |
 | CMAKE_C_COMPILER   | icx                               | Set *icx* compiler for SYCL code path.      |
 | CMAKE_CXX_COMPILER | icpx *(Linux)*, icx *(Windows)*   | Set `icpx/icx` compiler for SYCL code path. |
 

README.md

@@ -2,14 +2,20 @@
 
 ![llama](https://user-images.githubusercontent.com/1991296/230134379-7181e485-c521-4d23-a0d6-f7b3b61ba524.png)
 
-[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT) [![Server](https://github.com/ggerganov/llama.cpp/actions/workflows/server.yml/badge.svg?branch=master&event=schedule)](https://github.com/ggerganov/llama.cpp/actions/workflows/server.yml)
+[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
+[![Server](https://github.com/ggerganov/llama.cpp/actions/workflows/server.yml/badge.svg?branch=master&event=schedule)](https://github.com/ggerganov/llama.cpp/actions/workflows/server.yml)
+[![Conan Center](https://shields.io/conan/v/llama-cpp)](https://conan.io/center/llama-cpp)
 
 [Roadmap](https://github.com/users/ggerganov/projects/7) / [Project status](https://github.com/ggerganov/llama.cpp/discussions/3471) / [Manifesto](https://github.com/ggerganov/llama.cpp/discussions/205) / [ggml](https://github.com/ggerganov/ggml)
 
 Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others) in pure C/C++
 
+> [!IMPORTANT]
+[2024 Jun 12] Binaries have been renamed w/ a `llama-` prefix. `main` is now `llama-cli`, `server` is `llama-server`, etc (https://github.com/ggerganov/llama.cpp/pull/7809)
+
 ### Recent API changes
 
+- [2024 Jun 26] The source code and CMake build scripts have been restructured https://github.com/ggerganov/llama.cpp/pull/8006
 - [2024 Apr 21] `llama_token_to_piece` can now optionally render special tokens https://github.com/ggerganov/llama.cpp/pull/6807
 - [2024 Apr 4] State and session file functions reorganized under `llama_state_*` https://github.com/ggerganov/llama.cpp/pull/6341
 - [2024 Mar 26] Logits and embeddings API updated for compactness https://github.com/ggerganov/llama.cpp/pull/6122
@@ -20,7 +26,8 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others)
 
 ### Hot topics
 
-- **Initial Flash-Attention support: https://github.com/ggerganov/llama.cpp/pull/5021**
+- **`convert.py` has been deprecated and moved to `examples/convert-legacy-llama.py`, please use `convert-hf-to-gguf.py`** https://github.com/ggerganov/llama.cpp/pull/7430
+- Initial Flash-Attention support: https://github.com/ggerganov/llama.cpp/pull/5021
 - BPE pre-tokenization support has been added: https://github.com/ggerganov/llama.cpp/pull/6920
 - MoE memory layout has been updated - reconvert models for `mmap` support and regenerate `imatrix` https://github.com/ggerganov/llama.cpp/pull/6387
 - Model sharding instructions using `gguf-split` https://github.com/ggerganov/llama.cpp/discussions/6404
@@ -50,7 +57,6 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others)
         <li><a href="#quantization">Quantization</a></li>
         <li><a href="#interactive-mode">Interactive mode</a></li>
         <li><a href="#constrained-output-with-grammars">Constrained output with grammars</a></li>
-        <li><a href="#instruct-mode">Instruct mode</a></li>
         <li><a href="#obtaining-and-using-the-facebook-llama-2-model">Obtaining and using the Facebook LLaMA 2 model</a></li>
         <li><a href="#seminal-papers-and-background-on-the-models">Seminal papers and background on the models</a></li>
         <li><a href="#perplexity-measuring-model-quality">Perplexity (measuring model quality)</a></li>
@@ -74,7 +80,7 @@ variety of hardware - locally and in the cloud.
 - AVX, AVX2 and AVX512 support for x86 architectures
 - 1.5-bit, 2-bit, 3-bit, 4-bit, 5-bit, 6-bit, and 8-bit integer quantization for faster inference and reduced memory use
 - Custom CUDA kernels for running LLMs on NVIDIA GPUs (support for AMD GPUs via HIP)
-- Vulkan, SYCL, and (partial) OpenCL backend support
+- Vulkan and SYCL backend support
 - CPU+GPU hybrid inference to partially accelerate models larger than the total VRAM capacity
 
 Since its [inception](https://github.com/ggerganov/llama.cpp/issues/33#issuecomment-1465108022), the project has
@@ -102,6 +108,7 @@ Typically finetunes of the base models below are supported as well.
 - [X] [Falcon](https://huggingface.co/models?search=tiiuae/falcon)
 - [X] [Chinese LLaMA / Alpaca](https://github.com/ymcui/Chinese-LLaMA-Alpaca) and [Chinese LLaMA-2 / Alpaca-2](https://github.com/ymcui/Chinese-LLaMA-Alpaca-2)
 - [X] [Vigogne (French)](https://github.com/bofenghuang/vigogne)
+- [X] [BERT](https://github.com/ggerganov/llama.cpp/pull/5423)
 - [X] [Koala](https://bair.berkeley.edu/blog/2023/04/03/koala/)
 - [X] [Baichuan 1 & 2](https://huggingface.co/models?search=baichuan-inc/Baichuan) + [derivations](https://huggingface.co/hiyouga/baichuan-7b-sft)
 - [X] [Aquila 1 & 2](https://huggingface.co/models?search=BAAI/Aquila)
@@ -147,6 +154,8 @@ Typically finetunes of the base models below are supported as well.
 
 [llama.cpp web server](./examples/server) is a lightweight [OpenAI API](https://github.com/openai/openai-openapi) compatible HTTP server that can be used to serve local models and easily connect them to existing clients.
 
+[simplechat](./examples/server/public_simplechat) is a simple chat client, which can be used to chat with the model exposed using above web server (use --path to point to simplechat), from a local web browser.
+
 **Bindings:**
 
 - Python: [abetlen/llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
@@ -188,6 +197,7 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 - [cztomsik/ava](https://github.com/cztomsik/ava) (MIT)
 - [ptsochantaris/emeltal](https://github.com/ptsochantaris/emeltal)
 - [pythops/tenere](https://github.com/pythops/tenere) (AGPL)
+- [RAGNA Desktop](https://ragna.app/) (proprietary)
 - [RecurseChat](https://recurse.chat/) (proprietary)
 - [semperai/amica](https://github.com/semperai/amica)
 - [withcatai/catai](https://github.com/withcatai/catai)
@@ -200,19 +210,26 @@ Unless otherwise noted these projects are open-source with permissive licensing:
 - [KodiBot](https://github.com/firatkiral/kodibot) (GPL)
 - [eva](https://github.com/ylsdamxssjxxdd/eva) (MIT)
 - [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)
 
 *(to have a project listed here, it should clearly state that it depends on `llama.cpp`)*
 
 **Tools:**
 
 - [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
+
+**Infrastructure:**
+
+- [Paddler](https://github.com/distantmagic/paddler) - Stateful load balancer custom-tailored for llama.cpp
 
 ---
 
 Here is a typical run using LLaMA v2 13B on M2 Ultra:
 
 ```
-$ make -j && ./main -m models/llama-13b-v2/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e
+$ make -j && ./llama-cli -m models/llama-13b-v2/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e
 I llama.cpp build info:
 I UNAME_S:  Darwin
 I UNAME_P:  arm
@@ -315,8 +332,6 @@ In order to build llama.cpp you have four different options.
       make
       ```
 
-      **Note**: for `Debug` builds, run `make LLAMA_DEBUG=1`
-
   - On Windows:
 
     1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases).
@@ -328,40 +343,38 @@ In order to build llama.cpp you have four different options.
         make
         ```
 
+  - Notes:
+    - For faster compilation, add the `-j` argument to run multiple jobs in parallel. For example, `make -j 8` will run 8 jobs in parallel.
+    - For faster repeated compilation, install [ccache](https://ccache.dev/).
+    - For debug builds, run `make LLAMA_DEBUG=1`
+
 - Using `CMake`:
 
-    ```bash
-    cmake -B build
-    cmake --build build --config Release
-    ```
+  ```bash
+  cmake -B build
+  cmake --build build --config Release
+  ```
 
-    **Note**: for `Debug` builds, there are two cases:
+  **Notes**:
 
-    - Single-config generators (e.g. default = `Unix Makefiles`; note that they just ignore the `--config` flag):
+    - For faster compilation, add the `-j` argument to run multiple jobs in parallel. For example, `cmake --build build --config Release -j 8` will run 8 jobs in parallel.
+    - For faster repeated compilation, install [ccache](https://ccache.dev/).
+    - For debug builds, there are two cases:
+
+      1. Single-config generators (e.g. default = `Unix Makefiles`; note that they just ignore the `--config` flag):
 
       ```bash
       cmake -B build -DCMAKE_BUILD_TYPE=Debug
       cmake --build build
       ```
 
-    - Multi-config generators (`-G` param set to Visual Studio, XCode...):
+      2. Multi-config generators (`-G` param set to Visual Studio, XCode...):
 
       ```bash
       cmake -B build -G "Xcode"
       cmake --build build --config Debug
       ```
 
-- Using `Zig` (version 0.11 or later):
-
-    Building for optimization levels and CPU features can be accomplished using standard build arguments, for example AVX2, FMA, F16C,
-    it's also possible to cross compile for other operating systems and architectures:
-
-    ```bash
-    zig build -Doptimize=ReleaseFast -Dtarget=x86_64-windows-gnu -Dcpu=x86_64+avx2+fma+f16c
-    ```
-
-    The `zig targets` command will give you valid options to use.
-
 -   Using `gmake` (FreeBSD):
 
     1. Install and activate [DRM in FreeBSD](https://wiki.freebsd.org/Graphics)
@@ -369,27 +382,54 @@ In order to build llama.cpp you have four different options.
     3. Install compilation dependencies.
 
         ```bash
-        sudo pkg install gmake automake autoconf pkgconf llvm15 clinfo clover \
-            opencl clblast openblas
+        sudo pkg install gmake automake autoconf pkgconf llvm15 openblas
 
         gmake CC=/usr/local/bin/clang15 CXX=/usr/local/bin/clang++15 -j4
         ```
 
-    **Notes:** With this packages you can build llama.cpp with OPENBLAS and
-    CLBLAST support for use OpenCL GPU acceleration in FreeBSD. Please read
-    the instructions for use and activate this options in this document below.
+### Homebrew
+
+On Mac and Linux, the homebrew package manager can be used via
+```
+brew install llama.cpp
+```
+The formula is automatically updated with new `llama.cpp` releases. More info: https://github.com/ggerganov/llama.cpp/discussions/7668
+
+### Nix
+
+On Mac and Linux, the Nix package manager can be used via
+```
+nix profile install nixpkgs#llama-cpp
+```
+For flake enabled installs.
+
+Or
+```
+nix-env --file '<nixpkgs>' --install --attr llama-cpp
+```
+For non-flake enabled installs.
+
+This expression is automatically updated within the [nixpkgs repo](https://github.com/NixOS/nixpkgs/blob/nixos-24.05/pkgs/by-name/ll/llama-cpp/package.nix#L164).
+
+#### Flox
+
+On Mac and Linux, Flox can be used to install llama.cpp within a Flox environment via
+```
+flox install llama-cpp
+```
+Flox follows the nixpkgs build of llama.cpp.
 
 ### Metal Build
 
 On MacOS, Metal is enabled by default. Using Metal makes the computation run on the GPU.
-To disable the Metal build at compile time use the `LLAMA_NO_METAL=1` flag or the `LLAMA_METAL=OFF` cmake option.
+To disable the Metal build at compile time use the `GGML_NO_METAL=1` flag or the `GGML_METAL=OFF` cmake option.
 
 When built with Metal support, you can explicitly disable GPU inference with the `--n-gpu-layers|-ngl 0` command-line
 argument.
 
 ### BLAS Build
 
-Building the program with BLAS support may lead to some performance improvements in prompt processing using batch sizes higher than 32 (the default is 512). Support with CPU-only BLAS implementations doesn't affect the normal generation performance. We may see generation performance improvements with GPU-involved BLAS implementations, e.g. cuBLAS, hipBLAS and CLBlast. There are currently several different BLAS implementations available for build and use:
+Building the program with BLAS support may lead to some performance improvements in prompt processing using batch sizes higher than 32 (the default is 512). Support with CPU-only BLAS implementations doesn't affect the normal generation performance. We may see generation performance improvements with GPU-involved BLAS implementations, e.g. cuBLAS, hipBLAS. There are currently several different BLAS implementations available for build and use:
 
 - #### Accelerate Framework:
 
@@ -402,7 +442,7 @@ Building the program with BLAS support may lead to some performance improvements
   - Using `make`:
     - On Linux:
       ```bash
-      make LLAMA_OPENBLAS=1
+      make GGML_OPENBLAS=1
       ```
 
     - On Windows:
@@ -417,13 +457,13 @@ Building the program with BLAS support may lead to some performance improvements
       8. From here you can run:
 
           ```bash
-          make LLAMA_OPENBLAS=1
+          make GGML_OPENBLAS=1
           ```
 
   - Using `CMake` on Linux:
 
       ```bash
-      cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
+      cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS
       cmake --build build --config Release
       ```
 
@@ -442,10 +482,10 @@ Building the program with BLAS support may lead to some performance improvements
   Building through oneAPI compilers will make avx_vnni instruction set available for intel processors that do not support avx512 and avx512_vnni. Please note that this build config **does not support Intel GPU**. For Intel GPU support, please refer to [llama.cpp for SYCL](./README-sycl.md).
 
   - Using manual oneAPI installation:
-    By default, `LLAMA_BLAS_VENDOR` is set to `Generic`, so if you already sourced intel environment script and assign `-DLLAMA_BLAS=ON` in cmake, the mkl version of Blas will automatically been selected. Otherwise please install oneAPI and follow the below steps:
+    By default, `GGML_BLAS_VENDOR` is set to `Generic`, so if you already sourced intel environment script and assign `-DGGML_BLAS=ON` in cmake, the mkl version of Blas will automatically been selected. Otherwise please install oneAPI and follow the below steps:
       ```bash
       source /opt/intel/oneapi/setvars.sh # You can skip this step if  in oneapi-basekit docker image, only required for manual installation
-      cmake -B build -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_NATIVE=ON
+      cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_NATIVE=ON
       cmake --build build --config Release
       ```
 
@@ -462,25 +502,28 @@ Building the program with BLAS support may lead to some performance improvements
 
   - Using `make`:
     ```bash
-    make LLAMA_CUDA=1
+    make GGML_CUDA=1
     ```
   - Using `CMake`:
 
     ```bash
-    cmake -B build -DLLAMA_CUDA=ON
+    cmake -B build -DGGML_CUDA=ON
     cmake --build build --config Release
     ```
 
   The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. The following compilation options are also available to tweak performance:
 
-  | Option                         | Legal values           | Default | Description                                                                                                                                                                                                                                                                             |
-  |--------------------------------|------------------------|---------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
-  | LLAMA_CUDA_FORCE_DMMV          | Boolean                | false   | Force the use of dequantization + matrix vector multiplication kernels instead of using kernels that do matrix vector multiplication on quantized data. By default the decision is made based on compute capability (MMVQ for 6.1/Pascal/GTX 1000 or higher). Does not affect k-quants. |
-  | LLAMA_CUDA_DMMV_X              | Positive integer >= 32 | 32      | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants.                                         |
-  | LLAMA_CUDA_MMV_Y               | Positive integer       | 1       | Block size in y direction for the CUDA mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended.                                                                                                                                         |
-  | LLAMA_CUDA_F16                 | Boolean                | false   | If enabled, use half-precision floating point arithmetic for the CUDA dequantization + mul mat vec kernels and for the q4_1 and q5_1 matrix matrix multiplication kernels. Can improve performance on relatively recent GPUs.                                                           |
-  | LLAMA_CUDA_KQUANTS_ITER        | 1 or 2                 | 2       | Number of values processed per iteration and per CUDA thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs.                                                                                                                     |
-  | LLAMA_CUDA_PEER_MAX_BATCH_SIZE | Positive integer       | 128     | Maximum batch size for which to enable peer access between multiple GPUs. Peer access requires either Linux or NVLink. When using NVLink enabling peer access for larger batch sizes is potentially beneficial.                                                                         |
+  | Option                        | Legal values           | Default | Description                                                                                                                                                                                                                                                                             |
+  |-------------------------------|------------------------|---------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+  | GGML_CUDA_FORCE_DMMV          | Boolean                | false   | Force the use of dequantization + matrix vector multiplication kernels instead of using kernels that do matrix vector multiplication on quantized data. By default the decision is made based on compute capability (MMVQ for 6.1/Pascal/GTX 1000 or higher). Does not affect k-quants. |
+  | GGML_CUDA_DMMV_X              | Positive integer >= 32 | 32      | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants.                                         |
+  | GGML_CUDA_MMV_Y               | Positive integer       | 1       | Block size in y direction for the CUDA mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended.                                                                                                                                         |
+  | GGML_CUDA_FORCE_MMQ           | Boolean                | false   | Force the use of custom matrix multiplication kernels for quantized models instead of FP16 cuBLAS even if there is no int8 tensor core implementation available (affects V100, RDNA3). MMQ kernels are enabled by default on GPUs with int8 tensor core support. With MMQ force enabled, speed for large batch sizes will be worse but VRAM consumption will be lower.                       |
+  | GGML_CUDA_FORCE_CUBLAS        | Boolean                | false   | Force the use of FP16 cuBLAS instead of custom matrix multiplication kernels for quantized models                                                                                                                                                                                       |
+  | GGML_CUDA_F16                 | Boolean                | false   | If enabled, use half-precision floating point arithmetic for the CUDA dequantization + mul mat vec kernels and for the q4_1 and q5_1 matrix matrix multiplication kernels. Can improve performance on relatively recent GPUs.                                                           |
+  | GGML_CUDA_KQUANTS_ITER        | 1 or 2                 | 2       | Number of values processed per iteration and per CUDA thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs.                                                                                                                     |
+  | GGML_CUDA_PEER_MAX_BATCH_SIZE | Positive integer       | 128     | Maximum batch size for which to enable peer access between multiple GPUs. Peer access requires either Linux or NVLink. When using NVLink enabling peer access for larger batch sizes is potentially beneficial.                                                                         |
+  | GGML_CUDA_FA_ALL_QUANTS       | Boolean                | false   | Compile support for all KV cache quantization type (combinations) for the FlashAttention CUDA kernels. More fine-grained control over KV cache size but compilation takes much longer.                                                                                                  |
 
 - #### hipBLAS
 
@@ -490,15 +533,15 @@ Building the program with BLAS support may lead to some performance improvements
 
   - Using `make`:
     ```bash
-    make LLAMA_HIPBLAS=1
+    make GGML_HIPBLAS=1
     ```
   - Using `CMake` for Linux (assuming a gfx1030-compatible AMD GPU):
     ```bash
     HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \
-        cmake -S . -B build -DLLAMA_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \
+        cmake -S . -B build -DGGML_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \
         && cmake --build build --config Release -- -j 16
     ```
-    On Linux it is also possible to use unified memory architecture (UMA) to share main memory between the CPU and integrated GPU by setting `-DLLAMA_HIP_UMA=ON`.
+    On Linux it is also possible to use unified memory architecture (UMA) to share main memory between the CPU and integrated GPU by setting `-DGGML_HIP_UMA=ON`.
     However, this hurts performance for non-integrated GPUs (but enables working with integrated GPUs).
 
     Note that if you get the following error:
@@ -512,19 +555,19 @@ Building the program with BLAS support may lead to some performance improvements
     ```bash
     HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -p)" \
     HIP_DEVICE_LIB_PATH=<directory-you-just-found> \
-        cmake -S . -B build -DLLAMA_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \
+        cmake -S . -B build -DGGML_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \
         && cmake --build build -- -j 16
     ```
 
   - Using `make` (example for target gfx1030, build with 16 CPU threads):
     ```bash
-    make -j16 LLAMA_HIPBLAS=1 LLAMA_HIP_UMA=1 AMDGPU_TARGETS=gfx1030
+    make -j16 GGML_HIPBLAS=1 GGML_HIP_UMA=1 AMDGPU_TARGETS=gfx1030
     ```
 
   - Using `CMake` for Windows (using x64 Native Tools Command Prompt for VS, and assuming a gfx1100-compatible AMD GPU):
     ```bash
     set PATH=%HIP_PATH%\bin;%PATH%
-    cmake -S . -B build -G Ninja -DAMDGPU_TARGETS=gfx1100 -DLLAMA_HIPBLAS=ON -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release
+    cmake -S . -B build -G Ninja -DAMDGPU_TARGETS=gfx1100 -DGGML_HIPBLAS=ON -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release
     cmake --build build
     ```
     Make sure that `AMDGPU_TARGETS` is set to the GPU arch you want to compile for. The above example uses `gfx1100` that corresponds to Radeon RX 7900XTX/XT/GRE. You can find a list of targets [here](https://llvm.org/docs/AMDGPUUsage.html#processors)
@@ -535,116 +578,11 @@ Building the program with BLAS support may lead to some performance improvements
   If your GPU is not officially supported you can use the environment variable [`HSA_OVERRIDE_GFX_VERSION`] set to a similar GPU, for example 10.3.0 on RDNA2 (e.g. gfx1030, gfx1031, or gfx1035) or 11.0.0 on RDNA3.
   The following compilation options are also available to tweak performance (yes, they refer to CUDA, not HIP, because it uses the same code as the cuBLAS version above):
 
-  | Option                  | Legal values           | Default | Description                                                                                                                                                                                                                                    |
-  |-------------------------|------------------------|---------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
-  | LLAMA_CUDA_DMMV_X       | Positive integer >= 32 | 32      | Number of values in x direction processed by the HIP dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. |
-  | LLAMA_CUDA_MMV_Y        | Positive integer       | 1       | Block size in y direction for the HIP mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants.                                                                       |
-  | LLAMA_CUDA_KQUANTS_ITER | 1 or 2                 | 2       | Number of values processed per iteration and per HIP thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs.                                                                             |
-
-- #### CLBlast
-
-  OpenCL acceleration is provided by the matrix multiplication kernels from the [CLBlast](https://github.com/CNugteren/CLBlast) project and custom kernels for ggml that can generate tokens on the GPU.
-
-  You will need the [OpenCL SDK](https://github.com/KhronosGroup/OpenCL-SDK).
-    - For Ubuntu, Debian, and Fedora the packages `opencl-headers`, `ocl-icd` may be needed.
-
-    - For Windows, a pre-built SDK is available on the [OpenCL Releases](https://github.com/KhronosGroup/OpenCL-SDK/releases) page.
-
-    - <details>
-        <summary>Installing the OpenCL SDK from source</summary>
-
-        ```sh
-        git clone --recurse-submodules https://github.com/KhronosGroup/OpenCL-SDK.git
-        cd OpenCL-SDK
-        cmake -B build -DBUILD_DOCS=OFF \
-          -DBUILD_EXAMPLES=OFF \
-          -DBUILD_TESTING=OFF \
-          -DOPENCL_SDK_BUILD_SAMPLES=OFF \
-          -DOPENCL_SDK_TEST_SAMPLES=OFF
-        cmake --build build
-        cmake --install build --prefix /some/path
-        ```
-      </details>
-
-  ##### Installing CLBlast
-
-  Pre-built CLBlast binaries may be found on the [CLBlast Releases](https://github.com/CNugteren/CLBlast/releases) page. For Unix variants, it may also be found in your operating system's packages.
-
-  Linux packaging:
-  Fedora Linux:
-  ```bash
-  sudo dnf install clblast
-  ```
-
-  Alternatively, they may be built from source.
-
-  - <details>
-    <summary>Windows:</summary>
-
-      ```cmd
-      set OPENCL_SDK_ROOT="C:/OpenCL-SDK-v2023.04.17-Win-x64"
-      git clone https://github.com/CNugteren/CLBlast.git
-      cd CLBlast
-      cmake -B build -DBUILD_SHARED_LIBS=OFF -DOVERRIDE_MSVC_FLAGS_TO_MT=OFF -DTUNERS=OFF -DOPENCL_ROOT=%OPENCL_SDK_ROOT% -G "Visual Studio 17 2022" -A x64
-      cmake --build build --config Release
-      cmake --install build --prefix C:/CLBlast
-      ```
-
-      (note: `--config Release` at build time is the default and only relevant for Visual Studio builds - or multi-config Ninja builds)
-
-  - <details>
-    <summary>Unix:</summary>
-
-      ```sh
-      git clone https://github.com/CNugteren/CLBlast.git
-      cd CLBlast
-      cmake -B build -DBUILD_SHARED_LIBS=OFF -DTUNERS=OFF
-      cmake --build build --config Release
-      cmake --install build --prefix /some/path
-      ```
-
-      Where `/some/path` is where the built library will be installed (default is `/usr/local`).
-    </details>
-
-  ##### Building Llama with CLBlast
-
-  - Build with make:
-    ```sh
-    make LLAMA_CLBLAST=1
-    ```
-  - CMake (Unix):
-    ```sh
-    cmake -B build -DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path
-    cmake --build build --config Release
-    ```
-  - CMake (Windows):
-    ```cmd
-    set CL_BLAST_CMAKE_PKG="C:/CLBlast/lib/cmake/CLBlast"
-    git clone https://github.com/ggerganov/llama.cpp
-    cd llama.cpp
-    cmake -B build -DBUILD_SHARED_LIBS=OFF -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH=%CL_BLAST_CMAKE_PKG% -G "Visual Studio 17 2022" -A x64
-    cmake --build build --config Release
-    cmake --install build --prefix C:/LlamaCPP
-    ```
-
-  ##### Running Llama with CLBlast
-
-  The CLBlast build supports `--gpu-layers|-ngl` like the CUDA version does.
-
-  To select the correct platform (driver) and device (GPU), you can use the environment variables `GGML_OPENCL_PLATFORM` and `GGML_OPENCL_DEVICE`.
-  The selection can be a number (starting from 0) or a text string to search:
-
-  ```sh
-  GGML_OPENCL_PLATFORM=1 ./main ...
-  GGML_OPENCL_DEVICE=2 ./main ...
-  GGML_OPENCL_PLATFORM=Intel ./main ...
-  GGML_OPENCL_PLATFORM=AMD GGML_OPENCL_DEVICE=1 ./main ...
-  ```
-
-  The default behavior is to find the first GPU device, but when it is an integrated GPU on a laptop, for instance, the selectors are useful.
-  Using the variables it is possible to select a CPU-based driver as well, if so desired.
-
-  You can get a list of platforms and devices from the `clinfo -l` command, etc.
+  | Option                 | Legal values           | Default | Description                                                                                                                                                                                                                                    |
+  |------------------------|------------------------|---------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+  | GGML_CUDA_DMMV_X       | Positive integer >= 32 | 32      | Number of values in x direction processed by the HIP dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. |
+  | GGML_CUDA_MMV_Y        | Positive integer       | 1       | Block size in y direction for the HIP mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants.                                                                       |
+  | GGML_CUDA_KQUANTS_ITER | 1 or 2                 | 2       | Number of values processed per iteration and per HIP thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs.                                                                             |
 
 - #### Vulkan
 
@@ -654,7 +592,7 @@ Building the program with BLAS support may lead to some performance improvements
 
   ```sh
   # Build the image
-  docker build -t llama-cpp-vulkan -f .devops/main-vulkan.Dockerfile .
+  docker build -t llama-cpp-vulkan -f .devops/llama-cli-vulkan.Dockerfile .
 
   # Then, use it:
   docker run -it --rm -v "$(pwd):/app:Z" --device /dev/dri/renderD128:/dev/dri/renderD128 --device /dev/dri/card1:/dev/dri/card1 llama-cpp-vulkan -m "/app/models/YOUR_MODEL_FILE" -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33
@@ -675,15 +613,17 @@ Building the program with BLAS support may lead to some performance improvements
   vulkaninfo
   ```
 
-  Alternatively your package manager might be able to provide the appropiate libraries. For example for Ubuntu 22.04 you can install `libvulkan-dev` instead.
+  Alternatively your package manager might be able to provide the appropriate libraries.
+  For example for Ubuntu 22.04 you can install `libvulkan-dev` instead.
+  For Fedora 40, you can install `vulkan-devel`, `glslc` and `glslang` packages.
 
   Then, build llama.cpp using the cmake command below:
 
   ```bash
-  cmake -B build -DLLAMA_VULKAN=1
+  cmake -B build -DGGML_VULKAN=1
   cmake --build build --config Release
   # Test the output binary (with "-ngl 33" to offload all layers to GPU)
-  ./bin/main -m "PATH_TO_MODEL" -p "Hi you how are you" -n 50 -e -ngl 33 -t 4
+  ./bin/llama-cli -m "PATH_TO_MODEL" -p "Hi you how are you" -n 50 -e -ngl 33 -t 4
 
   # You should see in the output, ggml_vulkan detected your GPU. For example:
   # ggml_vulkan: Using Intel(R) Graphics (ADL GT2) | uma: 1 | fp16: 1 | warp size: 32
@@ -696,7 +636,8 @@ Building the program with BLAS support may lead to some performance improvements
 
 To obtain the official LLaMA 2 weights please see the <a href="#obtaining-and-using-the-facebook-llama-2-model">Obtaining and using the Facebook LLaMA 2 model</a> section. There is also a large selection of pre-quantized `gguf` models available on Hugging Face.
 
-Note: `convert.py` does not support LLaMA 3, you can use `convert-hf-to-gguf.py` with LLaMA 3 downloaded from Hugging Face.
+Note: `convert.py` has been moved to `examples/convert-legacy-llama.py` and shouldn't be used for anything other than `Llama/Llama2/Mistral` models and their derivatives.
+It does not support LLaMA 3, you can use `convert-hf-to-gguf.py` with LLaMA 3 downloaded from Hugging Face.
 
 ```bash
 # obtain the official LLaMA model weights and place them in ./models
@@ -713,23 +654,20 @@ ls ./models
 python3 -m pip install -r requirements.txt
 
 # convert the model to ggml FP16 format
-python3 convert.py models/mymodel/
-
-# [Optional] for models using BPE tokenizers
-python convert.py models/mymodel/ --vocab-type bpe
+python3 convert-hf-to-gguf.py models/mymodel/
 
 # quantize the model to 4-bits (using Q4_K_M method)
-./quantize ./models/mymodel/ggml-model-f16.gguf ./models/mymodel/ggml-model-Q4_K_M.gguf Q4_K_M
+./llama-quantize ./models/mymodel/ggml-model-f16.gguf ./models/mymodel/ggml-model-Q4_K_M.gguf Q4_K_M
 
 # update the gguf filetype to current version if older version is now unsupported
-./quantize ./models/mymodel/ggml-model-Q4_K_M.gguf ./models/mymodel/ggml-model-Q4_K_M-v2.gguf COPY
+./llama-quantize ./models/mymodel/ggml-model-Q4_K_M.gguf ./models/mymodel/ggml-model-Q4_K_M-v2.gguf COPY
 ```
 
 ### Run the quantized model
 
 ```bash
 # start inference on a gguf model
-./main -m ./models/mymodel/ggml-model-Q4_K_M.gguf -n 128
+./llama-cli -m ./models/mymodel/ggml-model-Q4_K_M.gguf -n 128
 ```
 
 When running the larger models, make sure you have enough disk space to store all the intermediate files.
@@ -804,7 +742,7 @@ The time per token is measured on a MacBook M1 Pro 32GB RAM using 4 and 8 thread
 #### How to run
 
 1. Download/extract: https://huggingface.co/datasets/ggml-org/ci/resolve/main/wikitext-2-raw-v1.zip
-2. Run `./perplexity -m models/7B/ggml-model-q4_0.gguf -f wiki.test.raw`
+2. Run `./llama-perplexity -m models/7B/ggml-model-q4_0.gguf -f wiki.test.raw`
 3. Output:
 ```
 perplexity : calculating perplexity over 655 chunks
@@ -828,16 +766,16 @@ Here is an example of a few-shot interaction, invoked with the command
 ./examples/chat-13B.sh
 
 # custom arguments using a 13B model
-./main -m ./models/13B/ggml-model-q4_0.gguf -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt
+./llama-cli -m ./models/13B/ggml-model-q4_0.gguf -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt
 ```
 
-Note the use of `--color` to distinguish between user input and generated text. Other parameters are explained in more detail in the [README](examples/main/README.md) for the `main` example program.
+Note the use of `--color` to distinguish between user input and generated text. Other parameters are explained in more detail in the [README](examples/main/README.md) for the `llama-cli` example program.
 
 ![image](https://user-images.githubusercontent.com/1991296/224575029-2af3c7dc-5a65-4f64-a6bb-517a532aea38.png)
 
 ### Persistent Interaction
 
-The prompt, user inputs, and model generations can be saved and resumed across calls to `./main` by leveraging `--prompt-cache` and `--prompt-cache-all`. The `./examples/chat-persistent.sh` script demonstrates this with support for long-running, resumable chat sessions. To use this example, you must provide a file to cache the initial chat prompt and a directory to save the chat session, and may optionally provide the same variables as `chat-13B.sh`. The same prompt cache can be reused for new chat sessions. Note that both prompt cache and chat directory are tied to the initial prompt (`PROMPT_TEMPLATE`) and the model file.
+The prompt, user inputs, and model generations can be saved and resumed across calls to `./llama-cli` by leveraging `--prompt-cache` and `--prompt-cache-all`. The `./examples/chat-persistent.sh` script demonstrates this with support for long-running, resumable chat sessions. To use this example, you must provide a file to cache the initial chat prompt and a directory to save the chat session, and may optionally provide the same variables as `chat-13B.sh`. The same prompt cache can be reused for new chat sessions. Note that both prompt cache and chat directory are tied to the initial prompt (`PROMPT_TEMPLATE`) and the model file.
 
 ```bash
 # Start a new chat
@@ -859,41 +797,13 @@ PROMPT_TEMPLATE=./prompts/chat-with-bob.txt PROMPT_CACHE_FILE=bob.prompt.bin \
 `llama.cpp` supports grammars to constrain model output. For example, you can force the model to output JSON only:
 
 ```bash
-./main -m ./models/13B/ggml-model-q4_0.gguf -n 256 --grammar-file grammars/json.gbnf -p 'Request: schedule a call at 8pm; Command:'
+./llama-cli -m ./models/13B/ggml-model-q4_0.gguf -n 256 --grammar-file grammars/json.gbnf -p 'Request: schedule a call at 8pm; Command:'
 ```
 
 The `grammars/` folder contains a handful of sample grammars. To write your own, check out the [GBNF Guide](./grammars/README.md).
 
 For authoring more complex JSON grammars, you can also check out https://grammar.intrinsiclabs.ai/, a browser app that lets you write TypeScript interfaces which it compiles to GBNF grammars that you can save for local use. Note that the app is built and maintained by members of the community, please file any issues or FRs on [its repo](http://github.com/intrinsiclabsai/gbnfgen) and not this one.
 
-### Instruct mode
-
-1. First, download and place the `ggml` model into the `./models` folder
-2. Run the `main` tool like this:
-
-```
-./examples/alpaca.sh
-```
-
-Sample run:
-
-```
-== Running in interactive mode. ==
- - Press Ctrl+C to interject at any time.
- - Press Return to return control to LLaMA.
- - If you want to submit another line, end your input in '\'.
-
- Below is an instruction that describes a task. Write a response that appropriately completes the request.
-
-> How many letters are there in the English alphabet?
-There 26 letters in the English Alphabet
-> What is the most common way of transportation in Amsterdam?
-The majority (54%) are using public transit. This includes buses, trams and metros with over 100 lines throughout the city which make it very accessible for tourists to navigate around town as well as locals who commute by tram or metro on a daily basis
-> List 5 words that start with "ca".
-cadaver, cauliflower, cabbage (vegetable), catalpa (tree) and Cailleach.
->
-```
-
 ### Obtaining and using the Facebook LLaMA 2 model
 
 - Refer to [Facebook's LLaMA download page](https://ai.meta.com/resources/models-and-libraries/llama-downloads/) if you want to access the model data.
@@ -966,7 +876,7 @@ $mv /sdcard/llama.cpp/llama-2-7b-chat.Q4_K_M.gguf /data/data/com.termux/files/ho
 Now, you can start chatting:
 ```
 $cd /data/data/com.termux/files/home/bin
-$./main -m ../model/llama-2-7b-chat.Q4_K_M.gguf -n 128 -cml
+$./llama-cli -m ../model/llama-2-7b-chat.Q4_K_M.gguf -n 128 -cml
 ```
 
 Here's a demo of an interactive session running on Pixel 5 phone:
@@ -1033,8 +943,8 @@ Assuming one has the [nvidia-container-toolkit](https://github.com/NVIDIA/nvidia
 
 ```bash
 docker build -t local/llama.cpp:full-cuda -f .devops/full-cuda.Dockerfile .
-docker build -t local/llama.cpp:light-cuda -f .devops/main-cuda.Dockerfile .
-docker build -t local/llama.cpp:server-cuda -f .devops/server-cuda.Dockerfile .
+docker build -t local/llama.cpp:light-cuda -f .devops/llama-cli-cuda.Dockerfile .
+docker build -t local/llama.cpp:server-cuda -f .devops/llama-server-cuda.Dockerfile .
 ```
 
 You may want to pass in some different `ARGS`, depending on the CUDA environment supported by your container host, as well as the GPU architecture.
@@ -1084,7 +994,7 @@ docker run --gpus all -v /path/to/models:/models local/llama.cpp:server-cuda -m
 
 ### Docs
 
-- [main](./examples/main/README.md)
+- [main (cli)](./examples/main/README.md)
 - [server](./examples/server/README.md)
 - [jeopardy](./examples/jeopardy/README.md)
 - [BLIS](./docs/BLIS.md)

ci/run.sh

@@ -36,11 +36,11 @@ SRC=`pwd`
 CMAKE_EXTRA="-DLLAMA_FATAL_WARNINGS=ON"
 
 if [ ! -z ${GG_BUILD_METAL} ]; then
-    CMAKE_EXTRA="${CMAKE_EXTRA} -DLLAMA_METAL_SHADER_DEBUG=ON"
+    CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_METAL=ON"
 fi
 
 if [ ! -z ${GG_BUILD_CUDA} ]; then
-    CMAKE_EXTRA="${CMAKE_EXTRA} -DLLAMA_CUDA=1"
+    CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_CUDA=1"
 fi
 
 if [ ! -z ${GG_BUILD_SYCL} ]; then
@@ -50,7 +50,7 @@ if [ ! -z ${GG_BUILD_SYCL} ]; then
         exit 1
     fi
 
-    CMAKE_EXTRA="${CMAKE_EXTRA} -DLLAMA_SYCL=1 DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON"
+    CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_SYCL=1 DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON"
 fi
 ## helpers
 
@@ -284,10 +284,10 @@ function gg_run_open_llama_7b_v2 {
 
     set -e
 
-    (time cmake -DCMAKE_BUILD_TYPE=Release ${CMAKE_EXTRA} -DLLAMA_CUDA=1 .. ) 2>&1 | tee -a $OUT/${ci}-cmake.log
+    (time cmake -DCMAKE_BUILD_TYPE=Release ${CMAKE_EXTRA} -DGGML_CUDA=1 .. ) 2>&1 | tee -a $OUT/${ci}-cmake.log
     (time make -j                                                           ) 2>&1 | tee -a $OUT/${ci}-make.log
 
-    python3 ../convert.py ${path_models} --outfile ${path_models}/ggml-model-f16.gguf
+    python3 ../examples/convert-legacy-llama.py ${path_models} --outfile ${path_models}/ggml-model-f16.gguf
 
     model_f16="${path_models}/ggml-model-f16.gguf"
     model_q8_0="${path_models}/ggml-model-q8_0.gguf"
@@ -303,47 +303,47 @@ function gg_run_open_llama_7b_v2 {
 
     wiki_test="${path_wiki}/wiki.test.raw"
 
-    ./bin/quantize ${model_f16} ${model_q8_0} q8_0
-    ./bin/quantize ${model_f16} ${model_q4_0} q4_0
-    ./bin/quantize ${model_f16} ${model_q4_1} q4_1
-    ./bin/quantize ${model_f16} ${model_q5_0} q5_0
-    ./bin/quantize ${model_f16} ${model_q5_1} q5_1
-    ./bin/quantize ${model_f16} ${model_q2_k} q2_k
-    ./bin/quantize ${model_f16} ${model_q3_k} q3_k
-    ./bin/quantize ${model_f16} ${model_q4_k} q4_k
-    ./bin/quantize ${model_f16} ${model_q5_k} q5_k
-    ./bin/quantize ${model_f16} ${model_q6_k} q6_k
+    ./bin/llama-quantize ${model_f16} ${model_q8_0} q8_0
+    ./bin/llama-quantize ${model_f16} ${model_q4_0} q4_0
+    ./bin/llama-quantize ${model_f16} ${model_q4_1} q4_1
+    ./bin/llama-quantize ${model_f16} ${model_q5_0} q5_0
+    ./bin/llama-quantize ${model_f16} ${model_q5_1} q5_1
+    ./bin/llama-quantize ${model_f16} ${model_q2_k} q2_k
+    ./bin/llama-quantize ${model_f16} ${model_q3_k} q3_k
+    ./bin/llama-quantize ${model_f16} ${model_q4_k} q4_k
+    ./bin/llama-quantize ${model_f16} ${model_q5_k} q5_k
+    ./bin/llama-quantize ${model_f16} ${model_q6_k} q6_k
 
-    (time ./bin/main --model ${model_f16}  -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/main --model ${model_q8_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/main --model ${model_q4_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/main --model ${model_q4_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/main --model ${model_q5_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/main --model ${model_q5_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/main --model ${model_q2_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
-    (time ./bin/main --model ${model_q3_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/main --model ${model_q4_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/main --model ${model_q5_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/main --model ${model_q6_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/llama-cli --model ${model_f16}  -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-cli --model ${model_q8_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-cli --model ${model_q4_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/llama-cli --model ${model_q4_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/llama-cli --model ${model_q5_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/llama-cli --model ${model_q5_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/llama-cli --model ${model_q2_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
+    (time ./bin/llama-cli --model ${model_q3_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/llama-cli --model ${model_q4_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/llama-cli --model ${model_q5_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/llama-cli --model ${model_q6_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
-    (time ./bin/perplexity --model ${model_f16}  -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/perplexity --model ${model_q8_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/perplexity --model ${model_q4_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/perplexity --model ${model_q4_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/perplexity --model ${model_q5_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/perplexity --model ${model_q5_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/perplexity --model ${model_q2_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
-    (time ./bin/perplexity --model ${model_q3_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/perplexity --model ${model_q4_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/perplexity --model ${model_q5_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/perplexity --model ${model_q6_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/llama-perplexity --model ${model_f16}  -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-perplexity --model ${model_q8_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-perplexity --model ${model_q4_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/llama-perplexity --model ${model_q4_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/llama-perplexity --model ${model_q5_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/llama-perplexity --model ${model_q5_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/llama-perplexity --model ${model_q2_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
+    (time ./bin/llama-perplexity --model ${model_q3_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/llama-perplexity --model ${model_q4_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/llama-perplexity --model ${model_q5_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/llama-perplexity --model ${model_q6_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
-    (time ./bin/imatrix --model ${model_f16} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log
+    (time ./bin/llama-imatrix --model ${model_f16} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log
 
-    (time ./bin/save-load-state     -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state -fa -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state     -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state -fa -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state     -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state -fa -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state     -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state -fa -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
 
     function check_ppl {
         qnt="$1"
@@ -437,45 +437,45 @@ function gg_run_pythia_1_4b {
 
     wiki_test_60="${path_wiki}/wiki.test-60.raw"
 
-    ./bin/quantize ${model_f16} ${model_q8_0} q8_0
-    ./bin/quantize ${model_f16} ${model_q4_0} q4_0
-    ./bin/quantize ${model_f16} ${model_q4_1} q4_1
-    ./bin/quantize ${model_f16} ${model_q5_0} q5_0
-    ./bin/quantize ${model_f16} ${model_q5_1} q5_1
-    ./bin/quantize ${model_f16} ${model_q2_k} q2_k
-    ./bin/quantize ${model_f16} ${model_q3_k} q3_k
-    ./bin/quantize ${model_f16} ${model_q4_k} q4_k
-    ./bin/quantize ${model_f16} ${model_q5_k} q5_k
-    ./bin/quantize ${model_f16} ${model_q6_k} q6_k
+    ./bin/llama-quantize ${model_f16} ${model_q8_0} q8_0
+    ./bin/llama-quantize ${model_f16} ${model_q4_0} q4_0
+    ./bin/llama-quantize ${model_f16} ${model_q4_1} q4_1
+    ./bin/llama-quantize ${model_f16} ${model_q5_0} q5_0
+    ./bin/llama-quantize ${model_f16} ${model_q5_1} q5_1
+    ./bin/llama-quantize ${model_f16} ${model_q2_k} q2_k
+    ./bin/llama-quantize ${model_f16} ${model_q3_k} q3_k
+    ./bin/llama-quantize ${model_f16} ${model_q4_k} q4_k
+    ./bin/llama-quantize ${model_f16} ${model_q5_k} q5_k
+    ./bin/llama-quantize ${model_f16} ${model_q6_k} q6_k
 
-    (time ./bin/main --model ${model_f16}  -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/main --model ${model_q8_0} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/main --model ${model_q4_0} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/main --model ${model_q4_1} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/main --model ${model_q5_0} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/main --model ${model_q5_1} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/main --model ${model_q2_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
-    (time ./bin/main --model ${model_q3_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/main --model ${model_q4_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/main --model ${model_q5_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/main --model ${model_q6_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/llama-cli --model ${model_f16}  -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-cli --model ${model_q8_0} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-cli --model ${model_q4_0} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/llama-cli --model ${model_q4_1} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/llama-cli --model ${model_q5_0} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/llama-cli --model ${model_q5_1} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/llama-cli --model ${model_q2_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
+    (time ./bin/llama-cli --model ${model_q3_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/llama-cli --model ${model_q4_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/llama-cli --model ${model_q5_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/llama-cli --model ${model_q6_k} -s 1234 -n 64 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
-    (time ./bin/perplexity --model ${model_f16}  -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/perplexity --model ${model_q8_0} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/perplexity --model ${model_q4_0} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/perplexity --model ${model_q4_1} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/perplexity --model ${model_q5_0} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/perplexity --model ${model_q5_1} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/perplexity --model ${model_q2_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
-    (time ./bin/perplexity --model ${model_q3_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/perplexity --model ${model_q4_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/perplexity --model ${model_q5_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/perplexity --model ${model_q6_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/llama-perplexity --model ${model_f16}  -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-perplexity --model ${model_q8_0} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-perplexity --model ${model_q4_0} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/llama-perplexity --model ${model_q4_1} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/llama-perplexity --model ${model_q5_0} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/llama-perplexity --model ${model_q5_1} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/llama-perplexity --model ${model_q2_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
+    (time ./bin/llama-perplexity --model ${model_q3_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/llama-perplexity --model ${model_q4_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/llama-perplexity --model ${model_q5_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/llama-perplexity --model ${model_q6_k} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
-    (time ./bin/imatrix --model ${model_f16} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log
+    (time ./bin/llama-imatrix --model ${model_f16} -f ${wiki_test_60} -c 128 -b 128 --chunks 1 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log
 
-    (time ./bin/save-load-state     --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state -fa --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state     --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state -fa --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
 
     function check_ppl {
         qnt="$1"
@@ -550,7 +550,7 @@ function gg_run_pythia_2_8b {
 
     set -e
 
-    (time cmake -DCMAKE_BUILD_TYPE=Release ${CMAKE_EXTRA} -DLLAMA_CUDA=1 .. ) 2>&1 | tee -a $OUT/${ci}-cmake.log
+    (time cmake -DCMAKE_BUILD_TYPE=Release ${CMAKE_EXTRA} -DGGML_CUDA=1 .. ) 2>&1 | tee -a $OUT/${ci}-cmake.log
     (time make -j                                                           ) 2>&1 | tee -a $OUT/${ci}-make.log
 
     python3 ../convert-hf-to-gguf.py ${path_models} --outfile ${path_models}/ggml-model-f16.gguf
@@ -569,47 +569,47 @@ function gg_run_pythia_2_8b {
 
     wiki_test="${path_wiki}/wiki.test.raw"
 
-    ./bin/quantize ${model_f16} ${model_q8_0} q8_0
-    ./bin/quantize ${model_f16} ${model_q4_0} q4_0
-    ./bin/quantize ${model_f16} ${model_q4_1} q4_1
-    ./bin/quantize ${model_f16} ${model_q5_0} q5_0
-    ./bin/quantize ${model_f16} ${model_q5_1} q5_1
-    ./bin/quantize ${model_f16} ${model_q2_k} q2_k
-    ./bin/quantize ${model_f16} ${model_q3_k} q3_k
-    ./bin/quantize ${model_f16} ${model_q4_k} q4_k
-    ./bin/quantize ${model_f16} ${model_q5_k} q5_k
-    ./bin/quantize ${model_f16} ${model_q6_k} q6_k
+    ./bin/llama-quantize ${model_f16} ${model_q8_0} q8_0
+    ./bin/llama-quantize ${model_f16} ${model_q4_0} q4_0
+    ./bin/llama-quantize ${model_f16} ${model_q4_1} q4_1
+    ./bin/llama-quantize ${model_f16} ${model_q5_0} q5_0
+    ./bin/llama-quantize ${model_f16} ${model_q5_1} q5_1
+    ./bin/llama-quantize ${model_f16} ${model_q2_k} q2_k
+    ./bin/llama-quantize ${model_f16} ${model_q3_k} q3_k
+    ./bin/llama-quantize ${model_f16} ${model_q4_k} q4_k
+    ./bin/llama-quantize ${model_f16} ${model_q5_k} q5_k
+    ./bin/llama-quantize ${model_f16} ${model_q6_k} q6_k
 
-    (time ./bin/main --model ${model_f16}  -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/main --model ${model_q8_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/main --model ${model_q4_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/main --model ${model_q4_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/main --model ${model_q5_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/main --model ${model_q5_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/main --model ${model_q2_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
-    (time ./bin/main --model ${model_q3_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/main --model ${model_q4_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/main --model ${model_q5_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/main --model ${model_q6_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/llama-cli --model ${model_f16}  -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-cli --model ${model_q8_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-cli --model ${model_q4_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/llama-cli --model ${model_q4_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/llama-cli --model ${model_q5_0} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/llama-cli --model ${model_q5_1} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/llama-cli --model ${model_q2_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
+    (time ./bin/llama-cli --model ${model_q3_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/llama-cli --model ${model_q4_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/llama-cli --model ${model_q5_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/llama-cli --model ${model_q6_k} -t 1 -ngl 999 -s 1234 -n 256 --ignore-eos -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
-    (time ./bin/perplexity --model ${model_f16}  -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/perplexity --model ${model_q8_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/perplexity --model ${model_q4_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/perplexity --model ${model_q4_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/perplexity --model ${model_q5_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/perplexity --model ${model_q5_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/perplexity --model ${model_q2_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
-    (time ./bin/perplexity --model ${model_q3_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/perplexity --model ${model_q4_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/perplexity --model ${model_q5_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/perplexity --model ${model_q6_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/llama-perplexity --model ${model_f16}  -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-perplexity --model ${model_q8_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-perplexity --model ${model_q4_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/llama-perplexity --model ${model_q4_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/llama-perplexity --model ${model_q5_0} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/llama-perplexity --model ${model_q5_1} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/llama-perplexity --model ${model_q2_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q2_k.log
+    (time ./bin/llama-perplexity --model ${model_q3_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/llama-perplexity --model ${model_q4_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/llama-perplexity --model ${model_q5_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/llama-perplexity --model ${model_q6_k} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
-    (time ./bin/imatrix --model ${model_f16} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log
+    (time ./bin/llama-imatrix --model ${model_f16} -f ${wiki_test} -t 1 -ngl 999 -c 2048 -b 512 --chunks 4 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log
 
-    (time ./bin/save-load-state     -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state -fa -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state     -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
-    (time ./bin/save-load-state -fa -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state     -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state -fa -ngl 10 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state     -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
+    (time ./bin/llama-save-load-state -fa -ngl 99 --model ${model_q4_0} ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
 
     function check_ppl {
         qnt="$1"
@@ -688,15 +688,15 @@ function gg_run_embd_bge_small {
     (time cmake -DCMAKE_BUILD_TYPE=Release ${CMAKE_EXTRA} .. ) 2>&1 | tee -a $OUT/${ci}-cmake.log
     (time make -j                                            ) 2>&1 | tee -a $OUT/${ci}-make.log
 
-    python3 ../convert-hf-to-gguf.py ${path_models} --outfile ${path_models}/ggml-model-f16.gguf
+    python3 ../convert_hf_to_gguf.py ${path_models} --outfile ${path_models}/ggml-model-f16.gguf
 
     model_f16="${path_models}/ggml-model-f16.gguf"
     model_q8_0="${path_models}/ggml-model-q8_0.gguf"
 
-    ./bin/quantize ${model_f16} ${model_q8_0} q8_0
+    ./bin/llama-quantize ${model_f16} ${model_q8_0} q8_0
 
-    (time ./bin/embedding --model ${model_f16}  -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/embedding --model ${model_q8_0} -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/llama-embedding --model ${model_f16}  -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/llama-embedding --model ${model_q8_0} -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
 
     set +e
 }

cmake/arm64-windows-llvm.cmake

@@ -9,7 +9,7 @@ set( CMAKE_CXX_COMPILER  clang++ )
 set( CMAKE_C_COMPILER_TARGET   ${target} )
 set( CMAKE_CXX_COMPILER_TARGET ${target} )
 
-set( arch_c_flags "-march=armv8.7-a -fvectorize -ffp-model=fast" )
+set( arch_c_flags "-march=armv8.7-a -fvectorize -ffp-model=fast -fno-finite-math-only" )
 set( warn_c_flags "-Wno-format -Wno-unused-variable -Wno-unused-function -Wno-gnu-zero-variadic-macro-arguments" )
 
 set( CMAKE_C_FLAGS_INIT   "${arch_c_flags} ${warn_c_flags}" )

cmake/git-vars.cmake

@@ -0,0 +1,22 @@
+find_package(Git)
+
+# the commit's SHA1
+execute_process(COMMAND
+    "${GIT_EXECUTABLE}" describe --match=NeVeRmAtCh --always --abbrev=8
+    WORKING_DIRECTORY "${CMAKE_SOURCE_DIR}"
+    OUTPUT_VARIABLE GIT_SHA1
+    ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
+
+# the date of the commit
+execute_process(COMMAND
+    "${GIT_EXECUTABLE}" log -1 --format=%ad --date=local
+    WORKING_DIRECTORY "${CMAKE_SOURCE_DIR}"
+    OUTPUT_VARIABLE GIT_DATE
+    ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
+
+# the subject of the commit
+execute_process(COMMAND
+    "${GIT_EXECUTABLE}" log -1 --format=%s
+    WORKING_DIRECTORY "${CMAKE_SOURCE_DIR}"
+    OUTPUT_VARIABLE GIT_COMMIT_SUBJECT
+    ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)

cmake/llama-config.cmake.in

@@ -1,46 +1,43 @@
-set(LLAMA_VERSION @LLAMA_INSTALL_VERSION@)
+set(LLAMA_VERSION      @LLAMA_INSTALL_VERSION@)
 set(LLAMA_BUILD_COMMIT @LLAMA_BUILD_COMMIT@)
 set(LLAMA_BUILD_NUMBER @LLAMA_BUILD_NUMBER@)
-set(LLAMA_SHARED_LIB @BUILD_SHARED_LIBS@)
-set(LLAMA_BLAS @LLAMA_BLAS@)
-set(LLAMA_CUDA @LLAMA_CUDA@)
-set(LLAMA_METAL @LLAMA_METAL@)
-set(LLAMA_CLBLAST @LLAMA_CLBLAST@)
-set(LLAMA_HIPBLAS @LLAMA_HIPBLAS@)
-set(LLAMA_ACCELERATE @LLAMA_ACCELERATE@)
+set(LLAMA_SHARED_LIB   @BUILD_SHARED_LIBS@)
+
+set(GGML_BLAS       @GGML_BLAS@)
+set(GGML_CUDA       @GGML_CUDA@)
+set(GGML_METAL      @GGML_METAL@)
+set(GGML_HIPBLAS    @GGML_HIPBLAS@)
+set(GGML_ACCELERATE @GGML_ACCELERATE@)
 
 @PACKAGE_INIT@
 
 set_and_check(LLAMA_INCLUDE_DIR "@PACKAGE_LLAMA_INCLUDE_INSTALL_DIR@")
-set_and_check(LLAMA_LIB_DIR "@PACKAGE_LLAMA_LIB_INSTALL_DIR@")
-set_and_check(LLAMA_BIN_DIR "@PACKAGE_LLAMA_BIN_INSTALL_DIR@")
+set_and_check(LLAMA_LIB_DIR     "@PACKAGE_LLAMA_LIB_INSTALL_DIR@")
+set_and_check(LLAMA_BIN_DIR     "@PACKAGE_LLAMA_BIN_INSTALL_DIR@")
 
 # Ensure transient dependencies satisfied
 
 find_package(Threads REQUIRED)
-if (APPLE AND LLAMA_ACCELERATE)
+
+if (APPLE AND GGML_ACCELERATE)
     find_library(ACCELERATE_FRAMEWORK Accelerate REQUIRED)
 endif()
 
-if (LLAMA_BLAS)
+if (GGML_BLAS)
     find_package(BLAS REQUIRED)
 endif()
 
-if (LLAMA_CUDA)
+if (GGML_CUDA)
     find_package(CUDAToolkit REQUIRED)
 endif()
 
-if (LLAMA_METAL)
+if (GGML_METAL)
     find_library(FOUNDATION_LIBRARY Foundation REQUIRED)
     find_library(METAL_FRAMEWORK Metal REQUIRED)
     find_library(METALKIT_FRAMEWORK MetalKit REQUIRED)
 endif()
 
-if (LLAMA_CLBLAST)
-    find_package(CLBlast REQUIRED)
-endif()
-
-if (LLAMA_HIPBLAS)
+if (GGML_HIPBLAS)
     find_package(hip REQUIRED)
     find_package(hipblas REQUIRED)
     find_package(rocblas REQUIRED)
@@ -52,7 +49,9 @@ find_library(llama_LIBRARY llama
 
 set(_llama_link_deps "Threads::Threads" "@LLAMA_EXTRA_LIBS@")
 set(_llama_transient_defines "@LLAMA_TRANSIENT_DEFINES@")
+
 add_library(llama UNKNOWN IMPORTED)
+
 set_target_properties(llama
     PROPERTIES
         INTERFACE_INCLUDE_DIRECTORIES "${LLAMA_INCLUDE_DIR}"

cmake/llama.pc.in

@@ -0,0 +1,10 @@
+prefix=@CMAKE_INSTALL_PREFIX@
+exec_prefix=${prefix}
+libdir=${exec_prefix}/lib
+includedir=${prefix}/include
+
+Name: llama
+Description: Port of Facebook's LLaMA model in C/C++
+Version: @PROJECT_VERSION@
+Libs: -L${libdir} -lllama
+Cflags: -I${includedir}

codecov.yml

@@ -1,14 +0,0 @@
-comment: off
-
-coverage:
-  status:
-    project:
-      default:
-        target: auto
-        threshold: 0
-        base: auto
-    patch:
-      default:
-        target: auto
-        threshold: 0
-        base: auto

common/CMakeLists.txt

@@ -1,5 +1,6 @@
 # common
 
+find_package(Threads REQUIRED)
 
 # Build info header
 #
@@ -36,7 +37,7 @@ add_custom_command(
     COMMENT "Generating build details from Git"
     COMMAND ${CMAKE_COMMAND} -DMSVC=${MSVC} -DCMAKE_C_COMPILER_VERSION=${CMAKE_C_COMPILER_VERSION}
             -DCMAKE_C_COMPILER_ID=${CMAKE_C_COMPILER_ID} -DCMAKE_VS_PLATFORM_NAME=${CMAKE_VS_PLATFORM_NAME}
-            -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -P "${CMAKE_CURRENT_SOURCE_DIR}/../scripts/gen-build-info-cpp.cmake"
+            -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -P "${CMAKE_CURRENT_SOURCE_DIR}/cmake/build-info-gen-cpp.cmake"
     WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}/.."
     DEPENDS "${CMAKE_CURRENT_SOURCE_DIR}/build-info.cpp.in" ${GIT_INDEX}
     VERBATIM
@@ -83,5 +84,5 @@ if (LLAMA_CURL)
 endif ()
 
 target_include_directories(${TARGET} PUBLIC .)
-target_compile_features(${TARGET} PUBLIC cxx_std_11)
-target_link_libraries(${TARGET} PRIVATE ${LLAMA_COMMON_EXTRA_LIBS} PUBLIC llama)
+target_compile_features   (${TARGET} PUBLIC cxx_std_11)
+target_link_libraries     (${TARGET} PRIVATE ${LLAMA_COMMON_EXTRA_LIBS} PUBLIC llama Threads::Threads)

common/cmake/build-info-gen-cpp.cmake

@@ -1,7 +1,7 @@
-include(${CMAKE_CURRENT_SOURCE_DIR}/scripts/build-info.cmake)
+include(${CMAKE_CURRENT_SOURCE_DIR}/cmake/build-info.cmake)
 
 set(TEMPLATE_FILE "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp.in")
-set(OUTPUT_FILE "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp")
+set(OUTPUT_FILE   "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp")
 
 # Only write the build info if it changed
 if(EXISTS ${OUTPUT_FILE})

common/common.h

@@ -52,70 +52,76 @@ int32_t cpu_get_num_math();
 // CLI argument parsing
 //
 
+// dimensionality reduction methods, used by cvector-generator
+enum dimre_method {
+    DIMRE_METHOD_PCA,
+    DIMRE_METHOD_MEAN,
+};
+
 struct gpt_params {
     uint32_t seed                 = LLAMA_DEFAULT_SEED; // RNG seed
 
     int32_t n_threads             = cpu_get_num_math();
-    int32_t n_threads_draft       = -1;
-    int32_t n_threads_batch       = -1;    // number of threads to use for batch processing (-1 = use n_threads)
-    int32_t n_threads_batch_draft = -1;
-    int32_t n_predict             = -1;    // new tokens to predict
-    int32_t n_ctx                 = 512;   // context size
-    int32_t n_batch               = 2048;  // logical batch size for prompt processing (must be >=32 to use BLAS)
-    int32_t n_ubatch              = 512;   // physical batch size for prompt processing (must be >=32 to use BLAS)
-    int32_t n_keep                = 0;     // number of tokens to keep from initial prompt
-    int32_t n_draft               = 5;     // number of tokens to draft during speculative decoding
-    int32_t n_chunks              = -1;    // max number of chunks to process (-1 = unlimited)
-    int32_t n_parallel            = 1;     // number of parallel sequences to decode
-    int32_t n_sequences           = 1;     // number of sequences to decode
-    float   p_split               = 0.1f;  // speculative decoding split probability
-    int32_t n_gpu_layers          = -1;    // number of layers to store in VRAM (-1 - use default)
-    int32_t n_gpu_layers_draft    = -1;    // number of layers to store in VRAM for the draft model (-1 - use default)
-    llama_split_mode split_mode   = LLAMA_SPLIT_MODE_LAYER; // how to split the model across GPUs
-    int32_t main_gpu              = 0;     // the GPU that is used for scratch and small tensors
-    float   tensor_split[128]     = {0};   // how split tensors should be distributed across GPUs
-    int32_t n_beams               = 0;     // if non-zero then use beam search of given width.
-    int32_t grp_attn_n            = 1;     // group-attention factor
-    int32_t grp_attn_w            = 512;   // group-attention width
-    int32_t n_print               = -1;    // print token count every n tokens (-1 = disabled)
-    float   rope_freq_base        = 0.0f;  // RoPE base frequency
-    float   rope_freq_scale       = 0.0f;  // RoPE frequency scaling factor
+    int32_t n_threads_draft       =    -1;
+    int32_t n_threads_batch       =    -1; // number of threads to use for batch processing (-1 = use n_threads)
+    int32_t n_threads_batch_draft =    -1;
+    int32_t n_predict             =    -1; // new tokens to predict
+    int32_t n_ctx                 =     0; // context size
+    int32_t n_batch               =  2048; // logical batch size for prompt processing (must be >=32 to use BLAS)
+    int32_t n_ubatch              =   512; // physical batch size for prompt processing (must be >=32 to use BLAS)
+    int32_t n_keep                =     0; // number of tokens to keep from initial prompt
+    int32_t n_draft               =     5; // number of tokens to draft during speculative decoding
+    int32_t n_chunks              =    -1; // max number of chunks to process (-1 = unlimited)
+    int32_t n_parallel            =     1; // number of parallel sequences to decode
+    int32_t n_sequences           =     1; // number of sequences to decode
+    float   p_split               =  0.1f; // speculative decoding split probability
+    int32_t n_gpu_layers          =    -1; // number of layers to store in VRAM (-1 - use default)
+    int32_t n_gpu_layers_draft    =    -1; // number of layers to store in VRAM for the draft model (-1 - use default)
+    int32_t main_gpu              =     0; // the GPU that is used for scratch and small tensors
+    float   tensor_split[128]     =   {0}; // how split tensors should be distributed across GPUs
+    int32_t grp_attn_n            =     1; // group-attention factor
+    int32_t grp_attn_w            =   512; // group-attention width
+    int32_t n_print               =    -1; // print token count every n tokens (-1 = disabled)
+    float   rope_freq_base        =  0.0f; // RoPE base frequency
+    float   rope_freq_scale       =  0.0f; // RoPE frequency scaling factor
     float   yarn_ext_factor       = -1.0f; // YaRN extrapolation mix factor
-    float   yarn_attn_factor      = 1.0f;  // YaRN magnitude scaling factor
+    float   yarn_attn_factor      =  1.0f; // YaRN magnitude scaling factor
     float   yarn_beta_fast        = 32.0f; // YaRN low correction dim
-    float   yarn_beta_slow        = 1.0f;  // YaRN high correction dim
-    int32_t yarn_orig_ctx         = 0;     // YaRN original context length
+    float   yarn_beta_slow        =  1.0f; // YaRN high correction dim
+    int32_t yarn_orig_ctx         =     0; // YaRN original context length
     float   defrag_thold          = -1.0f; // KV cache defragmentation threshold
-    std::string rpc_servers       = "";    // comma separated list of RPC servers
 
     ggml_backend_sched_eval_callback cb_eval = nullptr;
     void * cb_eval_user_data                 = nullptr;
 
     ggml_numa_strategy numa = GGML_NUMA_STRATEGY_DISABLED;
 
+    enum llama_split_mode        split_mode        = LLAMA_SPLIT_MODE_LAYER; // how to split the model across GPUs
     enum llama_rope_scaling_type rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED;
     enum llama_pooling_type      pooling_type      = LLAMA_POOLING_TYPE_UNSPECIFIED; // pooling type for embeddings
 
     // // sampling parameters
     struct llama_sampling_params sparams;
 
-    std::string model                = "";  // model path
-    std::string model_draft          = "";  // draft model for speculative decoding
+    std::string model                = ""; // model path
+    std::string model_draft          = ""; // draft model for speculative decoding
     std::string model_alias          = "unknown"; // model alias
-    std::string model_url            = "";  // model url to download
-    std::string hf_repo              = "";  // HF repo
-    std::string hf_file              = "";  // HF file
+    std::string model_url            = ""; // model url to download
+    std::string hf_repo              = ""; // HF repo
+    std::string hf_file              = ""; // HF file
     std::string prompt               = "";
-    std::string prompt_file          = "";  // store the external prompt file name
-    std::string path_prompt_cache    = "";  // path to file for saving/loading prompt eval state
-    std::string input_prefix         = "";  // string to prefix user inputs with
-    std::string input_suffix         = "";  // string to suffix user inputs with
-    std::vector<std::string> antiprompt; // string upon seeing which more user input is prompted
-    std::string logdir               = "";  // directory in which to save YAML log files
+    std::string prompt_file          = ""; // store the external prompt file name
+    std::string path_prompt_cache    = ""; // path to file for saving/loading prompt eval state
+    std::string input_prefix         = ""; // string to prefix user inputs with
+    std::string input_suffix         = ""; // string to suffix user inputs with
+    std::string logdir               = ""; // directory in which to save YAML log files
     std::string lookup_cache_static  = ""; // path of static ngram cache file for lookup decoding
     std::string lookup_cache_dynamic = ""; // path of dynamic ngram cache file for lookup decoding
-    std::string logits_file          = "";  // file for saving *all* logits
+    std::string logits_file          = ""; // file for saving *all* logits
+    std::string rpc_servers          = ""; // comma separated list of RPC servers
 
+    std::vector<std::string> in_files;   // all input files
+    std::vector<std::string> antiprompt; // strings upon which more user input is prompted (a.k.a. reverse prompts)
     std::vector<llama_model_kv_override> kv_overrides;
 
     // TODO: avoid tuple, use struct
@@ -124,37 +130,35 @@ struct gpt_params {
 
     std::vector<llama_control_vector_load_info> control_vectors; // control vector with user defined scale
 
+    int32_t verbosity                  = 0;
     int32_t control_vector_layer_start = -1; // layer range for control vector
     int32_t control_vector_layer_end   = -1; // layer range for control vector
 
-    int  ppl_stride        = 0;     // stride for perplexity calculations. If left at 0, the pre-existing approach will be used.
-    int  ppl_output_type   = 0;     // = 0 -> ppl output is as usual, = 1 -> ppl output is num_tokens, ppl, one per line
-                                    //                                       (which is more convenient to use for plotting)
-                                    //
-    bool   hellaswag       = false; // compute HellaSwag score over random tasks from datafile supplied in prompt
-    size_t hellaswag_tasks = 400;   // number of tasks to use when computing the HellaSwag score
+    int32_t ppl_stride      = 0;     // stride for perplexity calculations. If left at 0, the pre-existing approach will be used.
+    int32_t ppl_output_type = 0;     // = 0 -> ppl output is as usual, = 1 -> ppl output is num_tokens, ppl, one per line
+                                     //                                       (which is more convenient to use for plotting)
+                                     //
+    bool   hellaswag        = false; // compute HellaSwag score over random tasks from datafile supplied in prompt
+    size_t hellaswag_tasks  = 400;   // number of tasks to use when computing the HellaSwag score
 
-    bool   winogrande      = false; // compute Winogrande score over random tasks from datafile supplied in prompt
-    size_t winogrande_tasks= 0;     // number of tasks to use when computing the Winogrande score. If 0, all tasks will be computed
+    bool   winogrande       = false; // compute Winogrande score over random tasks from datafile supplied in prompt
+    size_t winogrande_tasks = 0;     // number of tasks to use when computing the Winogrande score. If 0, all tasks will be computed
 
-    bool   multiple_choice = false; // compute TruthfulQA score over random tasks from datafile supplied in prompt
-    size_t multiple_choice_tasks = 0;     // number of tasks to use when computing the TruthfulQA score. If 0, all tasks will be computed
+    bool   multiple_choice  = false;  // compute TruthfulQA score over random tasks from datafile supplied in prompt
+    size_t multiple_choice_tasks = 0; // number of tasks to use when computing the TruthfulQA score. If 0, all tasks will be computed
 
-    bool   kl_divergence   = false; // compute KL divergence
+    bool   kl_divergence    = false; // compute KL divergence
 
-    bool random_prompt     = false; // do not randomize prompt if none provided
+    bool usage             = false; // print usage
     bool use_color         = false; // use color to distinguish generations and inputs
-    bool interactive       = false; // interactive mode
-    bool interactive_specials = false; // whether to allow special tokens from user, during interactive mode
     bool special           = false; // enable special token output
+    bool interactive       = false; // interactive mode
+    bool interactive_first = false; // wait for user input immediately
     bool conversation      = false; // conversation mode (does not print special tokens and suffix/prefix)
-    bool chatml            = false; // chatml mode (used for models trained on chatml syntax)
     bool prompt_cache_all  = false; // save user input and generations to prompt cache
     bool prompt_cache_ro   = false; // open the prompt cache read-only and do not update it
 
-    bool embedding         = false; // get only sentence embedding
-    bool escape            = false; // escape "\n", "\r", "\t", "\'", "\"", and "\\"
-    bool interactive_first = false; // wait for user input immediately
+    bool escape            = true;  // escape "\n", "\r", "\t", "\'", "\"", and "\\"
     bool multiline_input   = false; // reverse the usage of `\`
     bool simple_io         = false; // improves compatibility with subprocesses and limited consoles
     bool cont_batching     = true;  // insert new sequences for decoding on-the-fly
@@ -162,7 +166,6 @@ struct gpt_params {
 
     bool input_prefix_bos  = false; // prefix BOS to user inputs, preceding input_prefix
     bool ignore_eos        = false; // ignore generated EOS tokens
-    bool instruct          = false; // instruction mode (used for Alpaca models)
     bool logits_all        = false; // return logits for all tokens in the batch
     bool use_mmap          = true;  // use mmap for faster loads
     bool use_mlock         = false; // use mlock to keep model in memory
@@ -180,6 +183,76 @@ struct gpt_params {
     // multimodal models (see examples/llava)
     std::string mmproj = "";        // path to multimodal projector
     std::vector<std::string> image; // path to image file(s)
+
+    // embedding
+    bool embedding         = false; // get only sentence embedding
+    int32_t embd_normalize = 2;     // normalisation for embendings (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)
+    std::string embd_out   = "";    // empty = default, "array" = [[],[]...], "json" = openai style, "json+" = same "json" + cosine similarity matrix
+    std::string embd_sep   = "\n";  // separator of embendings
+
+    // server params
+    int32_t port           = 8080;         // server listens on this network port
+    int32_t timeout_read   = 600;          // http read timeout in seconds
+    int32_t timeout_write  = timeout_read; // http write timeout in seconds
+    int32_t n_threads_http = -1;           // number of threads to process HTTP requests
+
+    std::string hostname      = "127.0.0.1";
+    std::string public_path   = "";
+    std::string chat_template = "";
+    std::string system_prompt = "";
+    bool enable_chat_template = true;
+
+    std::vector<std::string> api_keys;
+
+    std::string ssl_file_key  = "";
+    std::string ssl_file_cert = "";
+
+    bool endpoint_slots   = true;
+    bool endpoint_metrics = false;
+
+    bool log_json = false;
+
+    std::string slot_save_path;
+
+    float slot_prompt_similarity = 0.5f;
+
+    // batched-bench params
+    bool is_pp_shared = false;
+
+    std::vector<int32_t> n_pp;
+    std::vector<int32_t> n_tg;
+    std::vector<int32_t> n_pl;
+
+    // retrieval params
+    std::vector<std::string> context_files; // context files to embed
+
+    int32_t chunk_size = 64; // chunk size for context embedding
+
+    std::string chunk_separator = "\n"; // chunk separator for context embedding
+
+    // passkey params
+    int32_t n_junk = 250; // number of times to repeat the junk text
+    int32_t i_pos  = -1;  // position of the passkey in the junk text
+
+    // imatrix params
+    std::string out_file = "imatrix.dat"; // save the resulting imatrix to this file
+
+    int32_t n_out_freq  = 10; // output the imatrix every n_out_freq iterations
+    int32_t n_save_freq =  0; // save the imatrix every n_save_freq iterations
+    int32_t i_chunk     =  0; // start processing from this chunk
+
+    bool process_output = false; // collect data for the output tensor
+    bool compute_ppl    = true;  // whether to compute perplexity
+
+    // cvector-generator params
+    int n_pca_batch = 100;
+    int n_pca_iterations = 1000;
+    dimre_method cvector_dimre_method = DIMRE_METHOD_PCA;
+    std::string cvector_outfile       = "control_vector.gguf";
+    std::string cvector_positive_file = "examples/cvector-generator/positive.txt";
+    std::string cvector_negative_file = "examples/cvector-generator/negative.txt";
+
+    bool spm_infill = false; // suffix/prefix/middle pattern for infill
 };
 
 void gpt_params_handle_model_default(gpt_params & params);
@@ -199,7 +272,20 @@ std::vector<std::string> string_split(std::string input, char separator);
 
 std::string string_strip(const std::string & str);
 std::string string_get_sortable_timestamp();
-std::string string_random_prompt(std::mt19937 & rng);
+
+template<class T>
+static std::vector<T> string_split(const std::string & str, char delim) {
+    std::vector<T> values;
+    std::istringstream str_stream(str);
+    std::string token;
+    while (std::getline(str_stream, token, delim)) {
+        T value;
+        std::istringstream token_stream(token);
+        token_stream >> value;
+        values.push_back(value);
+    }
+    return values;
+}
 
 bool string_parse_kv_override(const char * data, std::vector<llama_model_kv_override> & overrides);
 void string_process_escapes(std::string & input);
@@ -212,6 +298,7 @@ bool fs_validate_filename(const std::string & filename);
 bool fs_create_directory_with_parents(const std::string & path);
 
 std::string fs_get_cache_directory();
+std::string fs_get_cache_file(const std::string & filename);
 
 //
 // Model utils
@@ -282,6 +369,38 @@ std::string llama_detokenize_bpe(
 // defaults to true when model type is SPM, otherwise false.
 bool llama_should_add_bos_token(const llama_model * model);
 
+//
+// Chat template utils
+//
+
+// same with llama_chat_message, but uses std::string
+struct llama_chat_msg {
+    std::string role;
+    std::string content;
+};
+
+// Check if the template supplied via "--chat-template" is supported or not. Returns true if it's valid
+bool llama_chat_verify_template(const std::string & tmpl);
+
+// CPP wrapper for llama_chat_apply_template
+// If the built-in template is not supported, we default to chatml
+// If the custom "tmpl" is not supported, we throw an error
+std::string llama_chat_apply_template(const struct llama_model * model,
+        const std::string & tmpl,
+        const std::vector<llama_chat_msg> & chat,
+        bool add_ass);
+
+// Format single message, while taking into account the position of that message in chat history
+std::string llama_chat_format_single(const struct llama_model * model,
+        const std::string & tmpl,
+        const std::vector<llama_chat_msg> & past_msg,
+        const llama_chat_msg & new_msg,
+        bool add_ass);
+
+// Returns an example of formatted chat
+std::string llama_chat_format_example(const struct llama_model * model,
+        const std::string & tmpl);
+
 //
 // KV cache utils
 //
@@ -296,7 +415,7 @@ void llama_kv_cache_dump_view_seqs(const llama_kv_cache_view & view, int row_siz
 // Embedding utils
 //
 
-void llama_embd_normalize(const float * inp, float * out, int n);
+void llama_embd_normalize(const float * inp, float * out, int n, int embd_norm = 2);
 
 float llama_embd_similarity_cos(const float * embd1, const float * embd2, int n);
 
@@ -340,4 +459,3 @@ void yaml_dump_string_multiline(FILE * stream, const char * prop_name, const cha
 void yaml_dump_non_result_info(
     FILE * stream, const gpt_params & params, const llama_context * lctx,
     const std::string & timestamp, const std::vector<int> & prompt_tokens, const char * model_desc);
-

common/grammar-parser.cpp

@@ -46,8 +46,12 @@ namespace grammar_parser {
         state.rules[rule_id] = rule;
     }
 
+    static bool is_digit_char(char c) {
+        return '0' <= c && c <= '9';
+    }
+
     static bool is_word_char(char c) {
-        return ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') || c == '-' || ('0' <= c && c <= '9');
+        return ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') || c == '-' || is_digit_char(c);
     }
 
     static std::pair<uint32_t, const char *> parse_hex(const char * src, int size) {
@@ -99,6 +103,17 @@ namespace grammar_parser {
         return pos;
     }
 
+    static const char * parse_int(const char * src) {
+        const char * pos = src;
+        while (is_digit_char(*pos)) {
+            pos++;
+        }
+        if (pos == src) {
+            throw std::runtime_error(std::string("expecting integer at ") + src);
+        }
+        return pos;
+    }
+
     static std::pair<uint32_t, const char *> parse_char(const char * src) {
         if (*src == '\\') {
             switch (src[1]) {
@@ -137,6 +152,60 @@ namespace grammar_parser {
             bool                                 is_nested) {
         size_t last_sym_start = out_elements.size();
         const char * pos = src;
+
+        auto handle_repetitions = [&](int min_times, int max_times) {
+
+            if (last_sym_start == out_elements.size()) {
+                throw std::runtime_error(std::string("expecting preceding item to */+/?/{ at ") + pos);
+            }
+
+            // apply transformation to previous symbol (last_sym_start to end) according to
+            // the following rewrite rules:
+            // S{m,n} --> S S S (m times) S'(n-m)
+            //            S'(x)   ::= S S'(x-1) |
+            //            (... n-m definitions of these S' rules ...)
+            //            S'(1)   ::= S |
+            // S{m,} -->  S S S (m times) S'
+            //            S'     ::= S S' |
+            // S*     --> S{0,}
+            //        --> S'     ::= S S' |
+            // S+     --> S{1,}
+            //        --> S S'
+            //            S'     ::= S S' |
+            // S?     --> S{0,1}
+            //        --> S'
+            //            S'     ::= S |
+
+            std::vector<llama_grammar_element> previous_elements(out_elements.begin() + last_sym_start, out_elements.end());
+            if (min_times == 0) {
+                out_elements.resize(last_sym_start);
+            } else {
+                // Repeat the previous elements (min_times - 1) times
+                for (int i = 1; i < min_times; i++) {
+                    out_elements.insert(out_elements.end(), previous_elements.begin(), previous_elements.end());
+                }
+            }
+
+            uint32_t last_rec_rule_id = 0;
+            auto n_opt = max_times < 0 ? 1 : max_times - min_times;
+
+            std::vector<llama_grammar_element> rec_rule(previous_elements);
+            for (int i = 0; i < n_opt; i++) {
+                rec_rule.resize(previous_elements.size());
+                uint32_t rec_rule_id = generate_symbol_id(state, rule_name);
+                if (i > 0 || max_times < 0) {
+                    rec_rule.push_back({LLAMA_GRETYPE_RULE_REF, max_times < 0 ? rec_rule_id : last_rec_rule_id});
+                }
+                rec_rule.push_back({LLAMA_GRETYPE_ALT, 0});
+                rec_rule.push_back({LLAMA_GRETYPE_END, 0});
+                add_rule(state, rec_rule_id, rec_rule);
+                last_rec_rule_id = rec_rule_id;
+            }
+            if (n_opt > 0) {
+                out_elements.push_back({LLAMA_GRETYPE_RULE_REF, last_rec_rule_id});
+            }
+        };
+
         while (*pos) {
             if (*pos == '"') { // literal string
                 pos++;
@@ -197,40 +266,51 @@ namespace grammar_parser {
                     throw std::runtime_error(std::string("expecting ')' at ") + pos);
                 }
                 pos = parse_space(pos + 1, is_nested);
-            } else if (*pos == '*' || *pos == '+' || *pos == '?') { // repetition operator
-                if (last_sym_start == out_elements.size()) {
-                    throw std::runtime_error(std::string("expecting preceding item to */+/? at ") + pos);
-                }
-
-                // apply transformation to previous symbol (last_sym_start to end) according to
-                // rewrite rules:
-                // S* --> S' ::= S S' |
-                // S+ --> S' ::= S S' | S
-                // S? --> S' ::= S |
-                uint32_t sub_rule_id = generate_symbol_id(state, rule_name);
-                std::vector<llama_grammar_element> sub_rule;
-                // add preceding symbol to generated rule
-                sub_rule.insert(
-                    sub_rule.end(), out_elements.begin() + last_sym_start, out_elements.end());
-                if (*pos == '*' || *pos == '+') {
-                    // cause generated rule to recurse
-                    sub_rule.push_back({LLAMA_GRETYPE_RULE_REF, sub_rule_id});
-                }
-                // mark start of alternate def
-                sub_rule.push_back({LLAMA_GRETYPE_ALT, 0});
-                if (*pos == '+') {
-                    // add preceding symbol as alternate only for '+' (otherwise empty)
-                    sub_rule.insert(
-                        sub_rule.end(), out_elements.begin() + last_sym_start, out_elements.end());
-                }
-                sub_rule.push_back({LLAMA_GRETYPE_END, 0});
-                add_rule(state, sub_rule_id, sub_rule);
-
-                // in original rule, replace previous symbol with reference to generated rule
-                out_elements.resize(last_sym_start);
-                out_elements.push_back({LLAMA_GRETYPE_RULE_REF, sub_rule_id});
-
+            } else if (*pos == '.') { // any char
+                last_sym_start = out_elements.size();
+                out_elements.push_back({LLAMA_GRETYPE_CHAR_ANY, 0});
                 pos = parse_space(pos + 1, is_nested);
+            } else if (*pos == '*') {
+                pos = parse_space(pos + 1, is_nested);
+                handle_repetitions(0, -1);
+            } else if (*pos == '+') {
+                pos = parse_space(pos + 1, is_nested);
+                handle_repetitions(1, -1);
+            } else if (*pos == '?') {
+                pos = parse_space(pos + 1, is_nested);
+                handle_repetitions(0, 1);
+            } else if (*pos == '{') {
+                pos = parse_space(pos + 1, is_nested);
+
+                if (!is_digit_char(*pos)) {
+                    throw std::runtime_error(std::string("expecting an int at ") + pos);
+                }
+                const char * int_end = parse_int(pos);
+                int min_times = std::stoul(std::string(pos, int_end - pos));
+                pos = parse_space(int_end, is_nested);
+
+                int max_times = -1;
+
+                if (*pos == '}') {
+                    max_times = min_times;
+                    pos = parse_space(pos + 1, is_nested);
+                } else if (*pos == ',') {
+                    pos = parse_space(pos + 1, is_nested);
+
+                    if (is_digit_char(*pos)) {
+                        const char * int_end = parse_int(pos);
+                        max_times = std::stoul(std::string(pos, int_end - pos));
+                        pos = parse_space(int_end, is_nested);
+                    }
+
+                    if (*pos != '}') {
+                        throw std::runtime_error(std::string("expecting '}' at ") + pos);
+                    }
+                    pos = parse_space(pos + 1, is_nested);
+                } else {
+                    throw std::runtime_error(std::string("expecting ',' at ") + pos);
+                }
+                handle_repetitions(min_times, max_times);
             } else {
                 break;
             }
@@ -325,6 +405,7 @@ namespace grammar_parser {
             case LLAMA_GRETYPE_CHAR_NOT:       return true;
             case LLAMA_GRETYPE_CHAR_ALT:       return true;
             case LLAMA_GRETYPE_CHAR_RNG_UPPER: return true;
+            case LLAMA_GRETYPE_CHAR_ANY:       return true;
             default:                           return false;
         }
     }
@@ -339,6 +420,7 @@ namespace grammar_parser {
                 case LLAMA_GRETYPE_CHAR_NOT:       fprintf(file, "CHAR_NOT");       break;
                 case LLAMA_GRETYPE_CHAR_RNG_UPPER: fprintf(file, "CHAR_RNG_UPPER"); break;
                 case LLAMA_GRETYPE_CHAR_ALT:       fprintf(file, "CHAR_ALT");       break;
+                case LLAMA_GRETYPE_CHAR_ANY:       fprintf(file, "CHAR_ANY");       break;
             }
             switch (elem.type) {
                 case LLAMA_GRETYPE_END:
@@ -350,6 +432,7 @@ namespace grammar_parser {
                 case LLAMA_GRETYPE_CHAR_NOT:
                 case LLAMA_GRETYPE_CHAR_RNG_UPPER:
                 case LLAMA_GRETYPE_CHAR_ALT:
+                case LLAMA_GRETYPE_CHAR_ANY:
                     fprintf(file, "(\"");
                     print_grammar_char(file, elem.value);
                     fprintf(file, "\") ");
@@ -407,11 +490,15 @@ namespace grammar_parser {
                     }
                     print_grammar_char(file, elem.value);
                     break;
+                case LLAMA_GRETYPE_CHAR_ANY:
+                    fprintf(file, ".");
+                    break;
             }
             if (is_char_element(elem)) {
                 switch (rule[i + 1].type) {
                     case LLAMA_GRETYPE_CHAR_ALT:
                     case LLAMA_GRETYPE_CHAR_RNG_UPPER:
+                    case LLAMA_GRETYPE_CHAR_ANY:
                         break;
                     default:
                         fprintf(file, "] ");

common/json-schema-to-grammar.cpp

@@ -16,92 +16,282 @@ static std::string join(Iterator begin, Iterator end, const std::string & separa
 
 static std::string repeat(const std::string & str, size_t n);
 
-static std::string build_repetition(const std::string & item_rule, int min_items, int max_items, const std::string & separator_rule = "", bool item_rule_is_literal = false) {
+static std::string build_repetition(const std::string & item_rule, int min_items, int max_items, const std::string & separator_rule = "") {
+    auto has_max = max_items != std::numeric_limits<int>::max();
+
+    if (min_items == 0 && max_items == 1) {
+        return item_rule + "?";
+    }
+
     if (separator_rule.empty()) {
-        if (min_items == 0 && max_items == 1) {
-            return item_rule + "?";
-        } else if (min_items == 1 && max_items == std::numeric_limits<int>::max()) {
+        if (min_items == 1 && !has_max) {
             return item_rule + "+";
-        }
-    }
-
-    std::string result;
-    if (min_items > 0) {
-        if (item_rule_is_literal && separator_rule.empty()) {
-            result = "\"" + repeat(std::string(item_rule.begin() + 1, item_rule.end() - 1), min_items) + "\"";
+        } else if (min_items == 0 && !has_max) {
+            return item_rule + "*";
         } else {
-            std::vector<std::string> items(min_items, item_rule);
-            result = join(items.begin(), items.end(), separator_rule.empty() ? " " : " " + separator_rule + " ");
+            return item_rule + "{" + std::to_string(min_items) + "," + (has_max ? std::to_string(max_items) : "") + "}";
         }
     }
 
-    std::function<std::string(int, bool)> opt_repetitions = [&](int up_to_n, bool prefix_with_sep) -> std::string {
-        auto content = prefix_with_sep && !separator_rule.empty() ? separator_rule + " " + item_rule : item_rule;
-
-        if (up_to_n == 0) {
-            return "";
-        } else if (up_to_n == 1) {
-            return "(" + content + ")?";
-        } else if (!separator_rule.empty() && !prefix_with_sep) {
-            return "(" + content + " " + opt_repetitions(up_to_n - 1, true) + ")?";
-        } else {
-            std::string res = repeat("(" + content + " ", up_to_n);
-            // strip trailing space
-            res = res.substr(0, res.length() - 1);
-            res += repeat(")?", up_to_n);
-            return res;
-        }
-    };
-
-    if (min_items > 0 && max_items != min_items) {
-        result += " ";
+    auto result = item_rule + " " + build_repetition("(" + separator_rule + " " + item_rule + ")", min_items == 0 ? 0 : min_items - 1, has_max ? max_items - 1 : max_items);
+    if (min_items == 0) {
+        result = "(" + result + ")?";
     }
-
-    if (max_items != std::numeric_limits<int>::max()) {
-        result += opt_repetitions(max_items - min_items, min_items > 0);
-    } else {
-        std::string item_operator = "(" + (separator_rule.empty() ? "" : separator_rule + " ") + item_rule + ")";
-        if (min_items == 0 && !separator_rule.empty()) {
-            result = "(" + item_rule + " " + item_operator + "*)?";
-        } else {
-            result += item_operator + "*";
-        }
-    }
-
     return result;
 }
 
-const std::string SPACE_RULE = "\" \"?";
+/* Minimalistic replacement for std::string_view, which is only available from C++17 onwards */
+class string_view {
+    const std::string & _str;
+    const size_t _start;
+    const size_t _end;
+public:
+    string_view(const std::string & str, size_t start = 0, size_t end  = std::string::npos) : _str(str), _start(start), _end(end == std::string::npos ? str.length() : end) {}
+
+    size_t size() const {
+        return _end - _start;
+    }
+
+    size_t length() const {
+        return size();
+    }
+
+    operator std::string() const {
+        return str();
+    }
+
+    std::string str() const {
+        return _str.substr(_start, _end - _start);
+    }
+
+    string_view substr(size_t pos, size_t len = std::string::npos) const {
+        return string_view(_str, _start + pos, len == std::string::npos ? _end : _start + pos + len);
+    }
+
+    char operator[](size_t pos) const {
+        auto index = _start + pos;
+        if (index >= _end) {
+            throw std::out_of_range("string_view index out of range");
+        }
+        return _str[_start + pos];
+    }
+
+    bool operator==(const string_view & other) const {
+        std::string this_str = *this;
+        std::string other_str = other;
+        return this_str == other_str;
+    }
+};
+
+static void _build_min_max_int(int min_value, int max_value, std::stringstream & out, int decimals_left = 16, bool top_level = true) {
+    auto has_min = min_value != std::numeric_limits<int>::min();
+    auto has_max = max_value != std::numeric_limits<int>::max();
+
+    auto digit_range = [&](char from, char to) {
+        out << "[";
+        if (from == to) {
+            out << from;
+        } else {
+            out << from << "-" << to;
+        }
+        out << "]";
+    };
+    auto more_digits = [&](int min_digits, int max_digits) {
+        out << "[0-9]";
+        if (min_digits == max_digits && min_digits == 1) {
+            return;
+        }
+        out << "{";
+        out << min_digits;
+        if (max_digits != min_digits) {
+            out << ",";
+            if (max_digits != std::numeric_limits<int>::max()) {
+                out << max_digits;
+            }
+        }
+        out << "}";
+    };
+    std::function<void(const string_view &, const string_view &)> uniform_range =
+        [&](const string_view & from, const string_view & to) {
+            size_t i = 0;
+            while (i < from.length() && i < to.length() && from[i] == to[i]) {
+                i++;
+            }
+            if (i > 0) {
+                out << "\"" << from.substr(0, i).str() << "\"";
+            }
+            if (i < from.length() && i < to.length()) {
+                if (i > 0) {
+                    out << " ";
+                }
+                auto sub_len = from.length() - i - 1;
+                if (sub_len > 0) {
+                    auto from_sub = from.substr(i + 1);
+                    auto to_sub = to.substr(i + 1);
+                    auto sub_zeros = repeat("0", sub_len);
+                    auto sub_nines = repeat("9", sub_len);
+
+                    auto to_reached = false;
+                    out << "(";
+                    if (from_sub == sub_zeros) {
+                        digit_range(from[i], to[i] - 1);
+                        out << " ";
+                        more_digits(sub_len, sub_len);
+                    } else {
+                        out << "[" << from[i] << "] ";
+                        out << "(";
+                        uniform_range(from_sub, sub_nines);
+                        out << ")";
+                        if (from[i] < to[i] - 1) {
+                            out << " | ";
+                            if (to_sub == sub_nines) {
+                                digit_range(from[i] + 1, to[i]);
+                                to_reached = true;
+                            } else {
+                                digit_range(from[i] + 1, to[i] - 1);
+                            }
+                            out << " ";
+                            more_digits(sub_len, sub_len);
+                        }
+                    }
+                    if (!to_reached) {
+                        out << " | ";
+                        digit_range(to[i], to[i]);
+                        out << " ";
+                        uniform_range(sub_zeros, to_sub);
+                    }
+                    out << ")";
+                } else {
+                    out << "[" << from[i] << "-" << to[i] << "]";
+                }
+            }
+        };
+
+    if (has_min && has_max) {
+        if (min_value < 0 && max_value < 0) {
+            out << "\"-\" (";
+            _build_min_max_int(-max_value, -min_value, out, decimals_left, /* top_level= */ true);
+            out << ")";
+            return;
+        }
+
+        if (min_value < 0) {
+            out << "\"-\" (";
+            _build_min_max_int(0, -min_value, out, decimals_left, /* top_level= */ true);
+            out << ") | ";
+            min_value = 0;
+        }
+
+        auto min_s = std::to_string(min_value);
+        auto max_s = std::to_string(max_value);
+        auto min_digits = min_s.length();
+        auto max_digits = max_s.length();
+
+        for (auto digits = min_digits; digits < max_digits; digits++) {
+            uniform_range(min_s, repeat("9", digits));
+            min_s = "1" + repeat("0", digits);
+            out << " | ";
+        }
+        uniform_range(min_s, max_s);
+        return;
+    }
+
+    auto less_decimals = std::max(decimals_left - 1, 1);
+
+    if (has_min) {
+        if (min_value < 0) {
+            out << "\"-\" (";
+            _build_min_max_int(std::numeric_limits<int>::min(), -min_value, out, decimals_left, /* top_level= */ false);
+            out << ") | [0] | [1-9] ";
+            more_digits(0, decimals_left - 1);
+        } else if (min_value == 0) {
+            if (top_level) {
+                out << "[0] | [1-9] ";
+                more_digits(0, less_decimals);
+            } else {
+                more_digits(1, decimals_left);
+            }
+        } else if (min_value <= 9) {
+            char c = '0' + min_value;
+            auto range_start = top_level ? '1' : '0';
+            if (c > range_start) {
+                digit_range(range_start, c - 1);
+                out << " ";
+                more_digits(1, less_decimals);
+                out << " | ";
+            }
+            digit_range(c, '9');
+            out << " ";
+            more_digits(0, less_decimals);
+        } else {
+            auto min_s = std::to_string(min_value);
+            auto len = min_s.length();
+            auto c = min_s[0];
+
+            if (c > '1') {
+                digit_range(top_level ? '1' : '0', c - 1);
+                out << " ";
+                more_digits(len, less_decimals);
+                out << " | ";
+            }
+            digit_range(c, c);
+            out << " (";
+            _build_min_max_int(std::stoi(min_s.substr(1)), std::numeric_limits<int>::max(), out, less_decimals, /* top_level= */ false);
+            out << ")";
+            if (c < '9') {
+                out << " | ";
+                digit_range(c + 1, '9');
+                out << " ";
+                more_digits(len - 1, less_decimals);
+            }
+        }
+        return;
+    }
+
+    if (has_max) {
+        if (max_value >= 0) {
+            if (top_level) {
+                out << "\"-\" [1-9] ";
+                more_digits(0, less_decimals);
+                out << " | ";
+            }
+            _build_min_max_int(0, max_value, out, decimals_left, /* top_level= */ true);
+        } else {
+            out << "\"-\" (";
+            _build_min_max_int(-max_value, std::numeric_limits<int>::max(), out, decimals_left, /* top_level= */ false);
+            out << ")";
+        }
+        return;
+    }
+
+    throw std::runtime_error("At least one of min_value or max_value must be set");
+}
+
+const std::string SPACE_RULE = "| \" \" | \"\\n\" [ \\t]{0,20}";
 
 struct BuiltinRule {
     std::string content;
     std::vector<std::string> deps;
 };
 
-const std::string _up_to_15_digits = build_repetition("[0-9]", 0, 15);
-
 std::unordered_map<std::string, BuiltinRule> PRIMITIVE_RULES = {
     {"boolean", {"(\"true\" | \"false\") space", {}}},
-    {"decimal-part", {"[0-9] " + _up_to_15_digits, {}}},
-    {"integral-part", {"[0-9] | [1-9] " + _up_to_15_digits, {}}},
+    {"decimal-part", {"[0-9]{1,16}", {}}},
+    {"integral-part", {"[0] | [1-9] [0-9]{0,15}", {}}},
     {"number", {"(\"-\"? integral-part) (\".\" decimal-part)? ([eE] [-+]? integral-part)? space", {"integral-part", "decimal-part"}}},
     {"integer", {"(\"-\"? integral-part) space", {"integral-part"}}},
     {"value", {"object | array | string | number | boolean | null", {"object", "array", "string", "number", "boolean", "null"}}},
     {"object", {"\"{\" space ( string \":\" space value (\",\" space string \":\" space value)* )? \"}\" space", {"string", "value"}}},
     {"array", {"\"[\" space ( value (\",\" space value)* )? \"]\" space", {"value"}}},
-    {"uuid", {"\"\\\"\" [0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F] "
-                "\"-\" [0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F] "
-                "\"-\" [0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F] "
-                "\"-\" [0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F] "
-                "\"-\" [0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F] \"\\\"\" space", {}}},
-    {"char",   {"[^\"\\\\] | \"\\\\\" ([\"\\\\/bfnrt] | \"u\" [0-9a-fA-F] [0-9a-fA-F] [0-9a-fA-F] [0-9a-fA-F])", {}}},
+    {"uuid", {"\"\\\"\" [0-9a-fA-F]{8} \"-\" [0-9a-fA-F]{4} \"-\" [0-9a-fA-F]{4} \"-\" [0-9a-fA-F]{4} \"-\" [0-9a-fA-F]{12} \"\\\"\" space", {}}},
+    {"char",   {"[^\"\\\\\\x7F\\x00-\\x1F] | [\\\\] ([\"\\\\bfnrt] | \"u\" [0-9a-fA-F]{4})", {}}},
     {"string", {"\"\\\"\" char* \"\\\"\" space", {"char"}}},
     {"null", {"\"null\" space", {}}},
 };
 
 std::unordered_map<std::string, BuiltinRule> STRING_FORMAT_RULES = {
-    {"date", {"[0-9] [0-9] [0-9] [0-9] \"-\" ( \"0\" [1-9] | \"1\" [0-2] ) \"-\" ( \"0\" [1-9] | [1-2] [0-9] | \"3\" [0-1] )", {}}},
-    {"time", {"([01] [0-9] | \"2\" [0-3]) \":\" [0-5] [0-9] \":\" [0-5] [0-9] ( \".\" [0-9] [0-9] [0-9] )? ( \"Z\" | ( \"+\" | \"-\" ) ( [01] [0-9] | \"2\" [0-3] ) \":\" [0-5] [0-9] )", {}}},
+    {"date", {"[0-9]{4} \"-\" ( \"0\" [1-9] | \"1\" [0-2] ) \"-\" ( \"0\" [1-9] | [1-2] [0-9] | \"3\" [0-1] )", {}}},
+    {"time", {"([01] [0-9] | \"2\" [0-3]) \":\" [0-5] [0-9] \":\" [0-5] [0-9] ( \".\" [0-9]{3} )? ( \"Z\" | ( \"+\" | \"-\" ) ( [01] [0-9] | \"2\" [0-3] ) \":\" [0-5] [0-9] )", {}}},
     {"date-time", {"date \"T\" time", {"date", "time"}}},
     {"date-string", {"\"\\\"\" date \"\\\"\" space", {"date"}}},
     {"time-string", {"\"\\\"\" time \"\\\"\" space", {"time"}}},
@@ -126,7 +316,7 @@ std::unordered_map<char, std::string> GRAMMAR_LITERAL_ESCAPES = {
 };
 
 std::unordered_set<char> NON_LITERAL_SET = {'|', '.', '(', ')', '[', ']', '{', '}', '*', '+', '?'};
-std::unordered_set<char> ESCAPED_IN_REGEXPS_BUT_NOT_IN_LITERALS = {'[', ']', '(', ')', '|', '{', '}', '*', '+', '?'};
+std::unordered_set<char> ESCAPED_IN_REGEXPS_BUT_NOT_IN_LITERALS = {'^', '$', '.', '[', ']', '(', ')', '|', '{', '}', '*', '+', '?'};
 
 template <typename Iterator>
 std::string join(Iterator begin, Iterator end, const std::string & separator) {
@@ -197,7 +387,6 @@ static std::string format_literal(const std::string & literal) {
     return "\"" + escaped + "\"";
 }
 
-
 class SchemaConverter {
 private:
     std::function<json(const std::string &)> _fetch_json;
@@ -385,8 +574,7 @@ private:
                         sub_is_literal ? "\"" + sub + "\"" : sub,
                         min_times,
                         max_times,
-                        "",
-                        sub_is_literal
+                        ""
                     );
                     seq.back().second = false;
                 } else {
@@ -426,6 +614,75 @@ private:
         return _add_rule(name, "\"\\\"\" " + to_rule(transform()) + " \"\\\"\" space");
     }
 
+    /*
+        Returns a rule that matches a JSON string that is none of the provided strings
+
+        not_strings({"a"})
+            -> ["] ( [a] char+ | [^"a] char* )? ["] space
+        not_strings({"and", "also"})
+            -> ["] ( [a] ([l] ([s] ([o] char+ | [^"o] char*) | [^"s] char*) | [n] ([d] char+ | [^"d] char*) | [^"ln] char*) | [^"a] char* )? ["] space
+    */
+    std::string _not_strings(const std::vector<std::string> & strings) {
+
+        struct TrieNode {
+            std::map<char, TrieNode> children;
+            bool is_end_of_string;
+
+            TrieNode() : is_end_of_string(false) {}
+
+            void insert(const std::string & string) {
+                auto node = this;
+                for (char c : string) {
+                    node = &node->children[c];
+                }
+                node->is_end_of_string = true;
+            }
+        };
+
+        TrieNode trie;
+        for (const auto & s : strings) {
+            trie.insert(s);
+        }
+
+        std::string char_rule = _add_primitive("char", PRIMITIVE_RULES.at("char"));
+        std::ostringstream out;
+        out << "[\"] ( ";
+        std::function<void(const TrieNode &)> visit = [&](const TrieNode & node) {
+            std::ostringstream rejects;
+            auto first = true;
+            for (const auto & kv : node.children) {
+                rejects << kv.first;
+                if (first) {
+                    first = false;
+                } else {
+                    out << " | ";
+                }
+                out << "[" << kv.first << "]";
+                if (!kv.second.children.empty()) {
+                    out << " (";
+                    visit(kv.second);
+                    out << ")";
+                } else if (kv.second.is_end_of_string) {
+                    out << " " << char_rule << "+";
+                }
+            }
+            if (!node.children.empty()) {
+                if (!first) {
+                    out << " | ";
+                }
+                out << "[^\"" << rejects.str() << "] " << char_rule << "*";
+            }
+        };
+        visit(trie);
+
+        out << " )";
+        if (!trie.is_end_of_string) {
+            out << "?";
+        }
+        out << " [\"] space";
+        return out.str();
+    }
+
     std::string _resolve_ref(const std::string & ref) {
         std::string ref_name = ref.substr(ref.find_last_of('/') + 1);
         if (_rules.find(ref_name) == _rules.end() && _refs_being_resolved.find(ref) == _refs_being_resolved.end()) {
@@ -446,6 +703,7 @@ private:
         std::vector<std::string> required_props;
         std::vector<std::string> optional_props;
         std::unordered_map<std::string, std::string> prop_kv_rule_names;
+        std::vector<std::string> prop_names;
         for (const auto & kv : properties) {
             const auto &prop_name = kv.first;
             const auto &prop_schema = kv.second;
@@ -460,11 +718,18 @@ private:
             } else {
                 optional_props.push_back(prop_name);
             }
+            prop_names.push_back(prop_name);
         }
-        if (additional_properties.is_object() || (additional_properties.is_boolean() && additional_properties.get<bool>())) {
+        if ((additional_properties.is_boolean() && additional_properties.get<bool>()) || additional_properties.is_object()) {
             std::string sub_name = name + (name.empty() ? "" : "-") + "additional";
-            std::string value_rule = visit(additional_properties.is_object() ? additional_properties : json::object(), sub_name + "-value");
-            std::string kv_rule = _add_rule(sub_name + "-kv", _add_primitive("string", PRIMITIVE_RULES.at("string")) + " \":\" space " + value_rule);
+            std::string value_rule =
+                additional_properties.is_object() ? visit(additional_properties, sub_name + "-value")
+                : _add_primitive("value", PRIMITIVE_RULES.at("value"));
+
+            auto key_rule =
+                prop_names.empty() ? _add_primitive("string", PRIMITIVE_RULES.at("string"))
+                : _add_rule(sub_name + "-k", _not_strings(prop_names));
+            std::string kv_rule = _add_rule(sub_name + "-kv", key_rule + " \":\" space " + value_rule);
             prop_kv_rule_names["*"] = kv_rule;
             optional_props.push_back("*");
         }
@@ -490,15 +755,11 @@ private:
                 }
                 std::string k = ks[0];
                 std::string kv_rule_name = prop_kv_rule_names[k];
-                if (k == "*") {
-                    res = _add_rule(
-                        name + (name.empty() ? "" : "-") + "additional-kvs",
-                        kv_rule_name + " ( \",\" space " + kv_rule_name + " )*"
-                    );
-                } else if (first_is_optional) {
-                    res = "( \",\" space " + kv_rule_name + " )?";
+                std::string comma_ref = "( \",\" space " + kv_rule_name + " )";
+                if (first_is_optional) {
+                    res = comma_ref + (k == "*" ? "*" : "?");
                 } else {
-                    res = kv_rule_name;
+                    res = kv_rule_name + (k == "*" ? " " + comma_ref + "*" : "");
                 }
                 if (ks.size() > 1) {
                     res += " " + _add_rule(
@@ -632,17 +893,19 @@ public:
         } else if (schema_type.is_array()) {
             std::vector<json> schema_types;
             for (const auto & t : schema_type) {
-                schema_types.push_back({{"type", t}});
+                json schema_copy(schema);
+                schema_copy["type"] = t;
+                schema_types.push_back(schema_copy);
             }
             return _add_rule(rule_name, _generate_union_rule(name, schema_types));
         } else if (schema.contains("const")) {
-            return _add_rule(rule_name, _generate_constant_rule(schema["const"]));
+            return _add_rule(rule_name, _generate_constant_rule(schema["const"]) + " space");
         } else if (schema.contains("enum")) {
             std::vector<std::string> enum_values;
             for (const auto & v : schema["enum"]) {
                 enum_values.push_back(_generate_constant_rule(v));
             }
-            return _add_rule(rule_name, join(enum_values.begin(), enum_values.end(), " | "));
+            return _add_rule(rule_name, "(" + join(enum_values.begin(), enum_values.end(), " | ") + ") space");
         } else if ((schema_type.is_null() || schema_type == "object")
                 && (schema.contains("properties") ||
                     (schema.contains("additionalProperties") && schema["additionalProperties"] != true))) {
@@ -724,6 +987,24 @@ public:
             int min_len = schema.contains("minLength") ? schema["minLength"].get<int>() : 0;
             int max_len = schema.contains("maxLength") ? schema["maxLength"].get<int>() : std::numeric_limits<int>::max();
             return _add_rule(rule_name, "\"\\\"\" " + build_repetition(char_rule, min_len, max_len) + " \"\\\"\" space");
+        } else if (schema_type == "integer" && (schema.contains("minimum") || schema.contains("exclusiveMinimum") || schema.contains("maximum") || schema.contains("exclusiveMaximum"))) {
+            int min_value = std::numeric_limits<int>::min();
+            int max_value = std::numeric_limits<int>::max();
+            if (schema.contains("minimum")) {
+                min_value = schema["minimum"].get<int>();
+            } else if (schema.contains("exclusiveMinimum")) {
+                min_value = schema["exclusiveMinimum"].get<int>() + 1;
+            }
+            if (schema.contains("maximum")) {
+                max_value = schema["maximum"].get<int>();
+            } else if (schema.contains("exclusiveMaximum")) {
+                max_value = schema["exclusiveMaximum"].get<int>() - 1;
+            }
+            std::stringstream out;
+            out << "(";
+            _build_min_max_int(min_value, max_value, out);
+            out << ") space";
+            return _add_rule(rule_name, out.str());
         } else if (schema.empty() || schema_type == "object") {
             return _add_rule(rule_name, _add_primitive("object", PRIMITIVE_RULES.at("object")));
         } else {

common/sampling.cpp

@@ -28,9 +28,13 @@ struct llama_sampling_context * llama_sampling_init(const struct llama_sampling_
 
         std::vector<const llama_grammar_element *> grammar_rules(result->parsed_grammar.c_rules());
 
-        result->grammar = llama_grammar_init(
+        struct llama_grammar * grammar = llama_grammar_init(
                 grammar_rules.data(),
                 grammar_rules.size(), result->parsed_grammar.symbol_ids.at("root"));
+        if (grammar == nullptr) {
+            throw std::runtime_error("Failed to initialize llama_grammar");
+        }
+        result->grammar = grammar;
     }
 
     result->prev.resize(params.n_prev);
@@ -59,9 +63,13 @@ void llama_sampling_reset(llama_sampling_context * ctx) {
     if (!ctx->parsed_grammar.rules.empty()) {
         std::vector<const llama_grammar_element *> grammar_rules(ctx->parsed_grammar.c_rules());
 
-        ctx->grammar = llama_grammar_init(
+        struct llama_grammar * grammar = llama_grammar_init(
                 grammar_rules.data(),
                 grammar_rules.size(), ctx->parsed_grammar.symbol_ids.at("root"));
+        if (grammar == nullptr) {
+            throw std::runtime_error("Failed to initialize llama_grammar");
+        }
+        ctx->grammar = grammar;
     }
 
     std::fill(ctx->prev.begin(), ctx->prev.end(), 0);

convert_hf_to_gguf_update.py

@@ -1,4 +1,5 @@
 #!/usr/bin/env python3
+# -*- coding: utf-8 -*-
 
 # This script downloads the tokenizer models of the specified models from Huggingface and
 # generates the get_vocab_base_pre() function for convert-hf-to-gguf.py
@@ -44,6 +45,7 @@ class TOKENIZER_TYPE(IntEnum):
     SPM = auto()
     BPE = auto()
     WPM = auto()
+    UGM = auto()
 
 
 # TODO: this string has to exercise as much pre-tokenizer functionality as possible
@@ -82,6 +84,13 @@ models = [
     {"name": "jina-v2-es",     "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jinaai/jina-embeddings-v2-base-es", },
     {"name": "jina-v2-de",     "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jinaai/jina-embeddings-v2-base-de", },
     {"name": "smaug-bpe",      "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/abacusai/Smaug-Llama-3-70B-Instruct", },
+    {"name": "poro-chat",      "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LumiOpen/Poro-34B-chat", },
+    {"name": "jina-v2-code",   "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jinaai/jina-embeddings-v2-base-code", },
+    {"name": "viking",         "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LumiOpen/Viking-7B", }, # Also used for Viking 13B and 33B
+    {"name": "gemma",          "tokt": TOKENIZER_TYPE.SPM, "repo": "https://huggingface.co/google/gemma-2b", },
+    {"name": "gemma-2",        "tokt": TOKENIZER_TYPE.SPM, "repo": "https://huggingface.co/google/gemma-2-9b", },
+    {"name": "jais",           "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/core42/jais-13b", },
+    {"name": "t5",             "tokt": TOKENIZER_TYPE.UGM, "repo": "https://huggingface.co/google-t5/t5-small", },
 ]
 
 
@@ -103,9 +112,13 @@ def download_model(model):
     os.makedirs(f"models/tokenizers/{name}", exist_ok=True)
 
     files = ["config.json", "tokenizer.json", "tokenizer_config.json"]
+
     if tokt == TOKENIZER_TYPE.SPM:
         files.append("tokenizer.model")
 
+    if tokt == TOKENIZER_TYPE.UGM:
+        files.append("spiece.model")
+
     for file in files:
         save_path = f"models/tokenizers/{name}/{file}"
         if os.path.isfile(save_path):
@@ -128,7 +141,7 @@ for model in models:
     name = model["name"]
     tokt = model["tokt"]
 
-    if tokt == TOKENIZER_TYPE.SPM:
+    if tokt == TOKENIZER_TYPE.SPM or tokt == TOKENIZER_TYPE.UGM:
         continue
 
     # Skip if the tokenizer folder does not exist or there are other download issues previously
@@ -138,7 +151,10 @@ for model in models:
 
     # create the tokenizer
     try:
-        tokenizer = AutoTokenizer.from_pretrained(f"models/tokenizers/{name}")
+        if name == "t5":
+            tokenizer = AutoTokenizer.from_pretrained(f"models/tokenizers/{name}", use_fast=False)
+        else:
+            tokenizer = AutoTokenizer.from_pretrained(f"models/tokenizers/{name}")
     except OSError as e:
         logger.error(f"Error loading tokenizer for model {name}. The model may not exist or is not accessible with the provided token. Error: {e}")
         continue  # Skip to the next model if the tokenizer can't be loaded
@@ -211,7 +227,7 @@ src_func = f"""
 """
 
 convert_py_pth = pathlib.Path("convert-hf-to-gguf.py")
-convert_py = convert_py_pth.read_text()
+convert_py = convert_py_pth.read_text(encoding="utf-8")
 convert_py = re.sub(
     r"(# Marker: Start get_vocab_base_pre)(.+?)( +# Marker: End get_vocab_base_pre)",
     lambda m: m.group(1) + src_func + m.group(3),
@@ -219,7 +235,7 @@ convert_py = re.sub(
     flags=re.DOTALL | re.MULTILINE,
 )
 
-convert_py_pth.write_text(convert_py)
+convert_py_pth.write_text(convert_py, encoding="utf-8")
 
 logger.info("+++ convert-hf-to-gguf.py was updated")
 
@@ -259,6 +275,7 @@ tests = [
     "\n =",
     "' era",
     "Hello, y'all! How are you 😁 ?我想在apple工作1314151天～",
+    "!!!!!!",
     "3",
     "33",
     "333",
@@ -268,7 +285,8 @@ tests = [
     "3333333",
     "33333333",
     "333333333",
-    # "Cửa Việt", # llama-bpe fails on this
+    "Cửa Việt", # llama-bpe fails on this
+    " discards",
     chktxt,
 ]
 
@@ -296,7 +314,10 @@ for model in models:
 
     # create the tokenizer
     try:
-        tokenizer = AutoTokenizer.from_pretrained(f"models/tokenizers/{name}")
+        if name == "t5":
+            tokenizer = AutoTokenizer.from_pretrained(f"models/tokenizers/{name}", use_fast=False)
+        else:
+            tokenizer = AutoTokenizer.from_pretrained(f"models/tokenizers/{name}")
     except OSError as e:
         logger.error(f"Failed to load tokenizer for model {name}. Error: {e}")
         continue  # Skip this model and continue with the next one in the loop

docs/BLIS.md

@@ -30,8 +30,8 @@ We recommend using openmp since it's easier to modify the cores being used.
 Makefile:
 
 ```bash
-make LLAMA_BLIS=1 -j
-# make LLAMA_BLIS=1 benchmark-matmult
+make GGML_BLIS=1 -j
+# make GGML_BLIS=1 llama-benchmark-matmult
 ```
 
 CMake:
@@ -39,7 +39,7 @@ CMake:
 ```bash
 mkdir build
 cd build
-cmake -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=FLAME ..
+cmake -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=FLAME ..
 make -j
 ```
 

docs/HOWTO-add-model.md

@@ -17,7 +17,7 @@ Also, it is important to check that the examples and main ggml backends (CUDA, M
 ### 1. Convert the model to GGUF
 
 This step is done in python with a `convert` script using the [gguf](https://pypi.org/project/gguf/) library.
-Depending on the model architecture, you can use either [convert.py](../convert.py) or [convert-hf-to-gguf.py](../convert-hf-to-gguf.py).
+Depending on the model architecture, you can use either [convert-hf-to-gguf.py](../convert-hf-to-gguf.py) or [examples/convert-legacy-llama.py](../examples/convert-legacy-llama.py) (for `llama/llama2` models in `.pth` format).
 
 The convert script reads the model configuration, tokenizer, tensor names+data and converts them to GGUF metadata and tensors.
 
@@ -100,7 +100,7 @@ Have a look at existing implementation like `build_llama`, `build_dbrx` or `buil
 
 When implementing a new graph, please note that the underlying `ggml` backends might not support them all, support for missing backend operations can be added in another PR.
 
-Note: to debug the inference graph: you can use [eval-callback](../examples/eval-callback).
+Note: to debug the inference graph: you can use [llama-eval-callback](../examples/eval-callback).
 
 ## GGUF specification
 

docs/token_generation_performance_tips.md

@@ -3,7 +3,7 @@
 ## Verifying that the model is running on the GPU with CUDA
 Make sure you compiled llama with the correct env variables according to [this guide](../README.md#CUDA), so that llama accepts the `-ngl N` (or `--n-gpu-layers N`) flag. When running llama, you may configure `N` to be very large, and llama will offload the maximum possible number of layers to the GPU, even if it's less than the number you configured. For example:
 ```shell
-./main -m "path/to/model.gguf" -ngl 200000 -p "Please sir, may I have some "
+./llama-cli -m "path/to/model.gguf" -ngl 200000 -p "Please sir, may I have some "
 ```
 
 When running llama, before it starts the inference work, it will output diagnostic information that shows whether cuBLAS is offloading work to the GPU. Look for these lines:
@@ -27,7 +27,7 @@ RAM: 32GB
 
 Model: `TheBloke_Wizard-Vicuna-30B-Uncensored-GGML/Wizard-Vicuna-30B-Uncensored.q4_0.gguf` (30B parameters, 4bit quantization, GGML)
 
-Run command: `./main -m "path/to/model.gguf" -p "An extremely detailed description of the 10 best ethnic dishes will follow, with recipes: " -n 1000 [additional benchmark flags]`
+Run command: `./llama-cli -m "path/to/model.gguf" -p "An extremely detailed description of the 10 best ethnic dishes will follow, with recipes: " -n 1000 [additional benchmark flags]`
 
 Result:
 

examples/CMakeLists.txt

@@ -12,44 +12,45 @@ include_directories(${CMAKE_CURRENT_SOURCE_DIR})
 
 if (EMSCRIPTEN)
 else()
+    add_subdirectory(cvector-generator)
     add_subdirectory(baby-llama)
-    add_subdirectory(batched)
     add_subdirectory(batched-bench)
-    add_subdirectory(beam-search)
+    add_subdirectory(batched)
     add_subdirectory(benchmark)
     add_subdirectory(convert-llama2c-to-ggml)
     add_subdirectory(embedding)
     add_subdirectory(eval-callback)
+    add_subdirectory(export-lora)
     add_subdirectory(finetune)
-    add_subdirectory(gritlm)
+    add_subdirectory(gbnf-validator)
     add_subdirectory(gguf-split)
+    add_subdirectory(gguf)
+    add_subdirectory(gritlm)
+    add_subdirectory(imatrix)
     add_subdirectory(infill)
     add_subdirectory(llama-bench)
     add_subdirectory(llava)
-    if (LLAMA_SYCL)
-        add_subdirectory(sycl)
-    endif()
-    add_subdirectory(main)
-    add_subdirectory(tokenize)
-    add_subdirectory(parallel)
-    add_subdirectory(perplexity)
-    add_subdirectory(quantize)
-    add_subdirectory(quantize-stats)
-    add_subdirectory(retrieval)
-    add_subdirectory(save-load-state)
-    add_subdirectory(simple)
-    add_subdirectory(passkey)
-    add_subdirectory(speculative)
     add_subdirectory(lookahead)
     add_subdirectory(lookup)
-    add_subdirectory(gguf)
-    add_subdirectory(train-text-from-scratch)
-    add_subdirectory(imatrix)
-    if (LLAMA_BUILD_SERVER)
-        add_subdirectory(server)
-    endif()
-    add_subdirectory(export-lora)
-    if (LLAMA_RPC)
+    add_subdirectory(main)
+    add_subdirectory(parallel)
+    add_subdirectory(passkey)
+    add_subdirectory(perplexity)
+    add_subdirectory(quantize-stats)
+    add_subdirectory(quantize)
+    add_subdirectory(retrieval)
+    if (GGML_RPC)
         add_subdirectory(rpc)
     endif()
+    if (LLAMA_BUILD_SERVER)
+    add_subdirectory(server)
+    endif()
+    if (GGML_SYCL)
+        add_subdirectory(sycl)
+    endif()
+    add_subdirectory(save-load-state)
+    add_subdirectory(simple)
+    add_subdirectory(speculative)
+    add_subdirectory(tokenize)
+    add_subdirectory(train-text-from-scratch)
 endif()

examples/Miku.sh

@@ -22,7 +22,7 @@ if [ -n "$N_THREAD" ]; then
     GEN_OPTIONS+=(--threads "$N_THREAD")
 fi
 
-./main "${GEN_OPTIONS[@]}" \
+./llama-cli "${GEN_OPTIONS[@]}" \
     --model "$MODEL" \
     --in-prefix " " \
     --in-suffix "${AI_NAME}:" \

examples/alpaca.sh

@@ -1,19 +0,0 @@
-#!/bin/bash
-
-#
-# Temporary script - will be removed in the future
-#
-
-cd `dirname $0`
-cd ..
-
-./main -m ./models/alpaca.13b.ggmlv3.q8_0.bin \
-       --color \
-       -f ./prompts/alpaca.txt \
-       --ctx_size 2048 \
-       -n -1 \
-       -ins -b 256 \
-       --top_k 10000 \
-       --temp 0.2 \
-       --repeat_penalty 1.1 \
-       -t 7

examples/baby-llama/CMakeLists.txt

@@ -1,4 +1,4 @@
-set(TARGET baby-llama)
+set(TARGET llama-baby-llama)
 add_executable(${TARGET} baby-llama.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})

examples/baby-llama/baby-llama.cpp

@@ -522,8 +522,8 @@ static struct ggml_tensor * forward(
             // wk   shape [n_embd, n_embd, 1, 1]
             // Qcur shape [n_embd/n_head, n_head, N, 1]
             // Kcur shape [n_embd/n_head, n_head, N, 1]
-            struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0, 0);
-            struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0, 0);
+            struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0);
+            struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0);
 
             // store key and value to memory
             {
@@ -759,8 +759,8 @@ static struct ggml_tensor * forward_batch(
             // wk   shape [n_embd, n_embd, 1, 1]
             // Qcur shape [n_embd/n_head, n_head, N, n_batch]
             // Kcur shape [n_embd/n_head, n_head, N, n_batch]
-            struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0, 0);
-            struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0, 0);
+            struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0);
+            struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0);
             assert_shape_4d(Qcur, n_embd/n_head, n_head, N, n_batch);
             assert_shape_4d(Kcur, n_embd/n_head, n_head, N, n_batch);
 
@@ -1056,7 +1056,7 @@ static struct ggml_tensor * forward_lora(
                                                         model->layers[il].wqb,
                                                         cur)),
                                                 n_embd/n_head, n_head, N),
-                                            KQ_pos, n_rot, 0, 0);
+                                            KQ_pos, n_rot, 0);
             struct ggml_tensor * Kcur = ggml_rope(ctx0,
                                             ggml_reshape_3d(ctx0,
                                                 ggml_mul_mat(ctx0,
@@ -1065,7 +1065,7 @@ static struct ggml_tensor * forward_lora(
                                                         model->layers[il].wkb,
                                                         cur)),
                                                 n_embd/n_head, n_head, N),
-                                            KQ_pos, n_rot, 0, 0);
+                                            KQ_pos, n_rot, 0);
 
             // store key and value to memory
             {

examples/base-translate.sh

@@ -58,4 +58,4 @@ echo "$2
 model=$1
 
 # generate the most likely continuation until the string "===" is found
-./main -m $model -f $ftmp -n 64 --temp 0 --repeat-penalty 1.0 --no-penalize-nl -r "===" $eargs
+./llama-cli -m $model -f $ftmp -n 64 --temp 0 --repeat-penalty 1.0 --no-penalize-nl -r "===" $eargs

examples/batched-bench/CMakeLists.txt

@@ -1,4 +1,4 @@
-set(TARGET batched-bench)
+set(TARGET llama-batched-bench)
 add_executable(${TARGET} batched-bench.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})

examples/batched-bench/README.md

@@ -10,16 +10,16 @@ There are 2 modes of operation:
 - `prompt is shared` - there is a common prompt of size `PP` used by all batches (i.e. `N_KV = PP + B*TG`)
 
 ```bash
-./batched-bench MODEL_PATH [N_KV_MAX] [N_BATCH] [N_UBATCH] [IS_PP_SHARED] [NGL] [MMQ] <PP> <TG> <PL>
+./llama-batched-bench -m model.gguf -c 2048 -b 2048 -ub 512 -npp 128,256,512 -ntg 128,256 -npl 1,2,4,8,16,32 [-pps]
 
 # LLaMA 7B, F16, N_KV_MAX = 16384 (8GB), prompt not shared
-./batched-bench ./models/llama-7b/ggml-model-f16.gguf 16384 2048 512 0 99
+./llama-batched-bench -m ./models/llama-7b/ggml-model-f16.gguf -c 16384 -b 2048 -ub 512 -ngl 99
 
 # LLaMA 7B, Q8_0, N_KV_MAX = 16384 (8GB), prompt is shared
-./batched-bench ./models/llama-7b/ggml-model-q8_0.gguf 16384 2048 512 1 99
+./llama-batched-bench -m ./models/llama-7b/ggml-model-q8_0.gguf -c 16384 -b 2048 -ub 512 -ngl 99 -pps
 
 # custom set of batches
-./batched-bench ./models/llama-7b/ggml-model-q8_0.gguf 2048 512 512 0 999 0 128,256,512 128,256 1,2,4,8,16,32
+./llama-batched-bench -m ./models/llama-7b/ggml-model-q8_0.gguf -c 2048 -b 512 -ub 512 -ngl 999 -npp 128,256,512 -ntg 128,256 -npl 1,2,4,8,16,32
 ```
 
 ## Sample results

examples/batched-bench/batched-bench.cpp

@@ -28,67 +28,27 @@ static std::vector<int> parse_list(char * p) {
     return ret;
 }
 
+static void print_usage(int argc, char ** argv, const gpt_params & params) {
+    gpt_params_print_usage(argc, argv, params);
+
+    LOG_TEE("\nexample usage:\n");
+    LOG_TEE("\n    %s -m model.gguf -c 2048 -b 2048 -ub 512 -npp 128,256,512 -ntg 128,256 -npl 1,2,4,8,16,32 [-pps]\n", argv[0]);
+    LOG_TEE("\n");
+}
+
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    if (argc == 1 || argv[1][0] == '-') {
-        printf("usage: %s MODEL_PATH [N_KV_MAX] [N_BATCH] [N_UBATCH] [FATTN] [IS_PP_SHARED] [NGL] <PP> <TG> <PL>\n" , argv[0]);
-        printf("  <PP>, <TG> and PL are comma-separated lists of numbers without spaces\n\n");
-        printf("  example: %s ggml-model-f16.gguf 2048 2048 512 0 999 128,256,512 128,256 1,2,4,8,16,32\n\n", argv[0]);
-        return 1 ;
+    if (!gpt_params_parse(argc, argv, params)) {
+        print_usage(argc, argv, params);
+        return 1;
     }
 
-    int n_kv_max     = 2048;
-    int n_batch      = 2048;
-    int n_ubatch     = 512;
-    bool flash_attn  = false;
-    int is_pp_shared = 0;
-    int n_gpu_layers = 0;
+    int is_pp_shared = params.is_pp_shared;
 
-    std::vector<int> n_pp = { 128, 256, 512, 1024, 2048, 3584, 7680, };
-    std::vector<int> n_tg = { 128, 256, };
-    std::vector<int> n_pl = { 1, 2, 4, 8, 16, 32, };
-    //std::vector<int> n_pl = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 32, };
-
-    if (argc >= 2) {
-        params.model = argv[1];
-    }
-
-    if (argc >= 3) {
-        n_kv_max = std::atoi(argv[2]);
-    }
-
-    if (argc >= 4) {
-        n_batch = std::atoi(argv[3]);
-    }
-
-    if (argc >= 5) {
-        n_ubatch = std::atoi(argv[4]);
-    }
-
-    if (argc >= 6) {
-        flash_attn = std::atoi(argv[5]);
-    }
-
-    if (argc >= 7) {
-        is_pp_shared = std::atoi(argv[6]);
-    }
-
-    if (argc >= 8) {
-        n_gpu_layers = std::atoi(argv[7]);
-    }
-
-    if (argc >= 9) {
-        n_pp = parse_list(argv[8]);
-    }
-
-    if (argc >= 10) {
-        n_tg = parse_list(argv[9]);
-    }
-
-    if (argc >= 11) {
-        n_pl = parse_list(argv[10]);
-    }
+    std::vector<int> n_pp = params.n_pp;
+    std::vector<int> n_tg = params.n_tg;
+    std::vector<int> n_pl = params.n_pl;
 
     // init LLM
 
@@ -97,12 +57,7 @@ int main(int argc, char ** argv) {
 
     // initialize the model
 
-    llama_model_params model_params = llama_model_default_params();
-
-    const std::vector<float> t_split(llama_max_devices(), 0.0f);
-
-    model_params.n_gpu_layers = n_gpu_layers;
-    model_params.tensor_split = t_split.data();
+    llama_model_params model_params = llama_model_params_from_gpt_params(params);
 
     llama_model * model = llama_load_model_from_file(params.model.c_str(), model_params);
 
@@ -111,16 +66,7 @@ int main(int argc, char ** argv) {
         return 1;
     }
 
-    llama_context_params ctx_params = llama_context_default_params();
-
-    ctx_params.seed       = 1234;
-    ctx_params.n_ctx      = n_kv_max;
-    ctx_params.n_batch    = n_batch;
-    ctx_params.n_ubatch   = n_ubatch;
-    ctx_params.flash_attn = flash_attn;
-
-    ctx_params.n_threads       = params.n_threads;
-    ctx_params.n_threads_batch = params.n_threads_batch == -1 ? params.n_threads : params.n_threads_batch;
+    llama_context_params ctx_params = llama_context_params_from_gpt_params(params);
 
     // ensure enough sequences are available
     ctx_params.n_seq_max = *std::max_element(n_pl.begin(), n_pl.end());
@@ -132,6 +78,8 @@ int main(int argc, char ** argv) {
         return 1;
     }
 
+    const int32_t n_kv_max = llama_n_ctx(ctx);
+
     llama_batch batch = llama_batch_init(n_kv_max, 0, 1);
 
     // decode in batches of ctx_params.n_batch tokens
@@ -175,7 +123,7 @@ int main(int argc, char ** argv) {
     }
 
     LOG_TEE("\n");
-    LOG_TEE("%s: n_kv_max = %d, n_batch = %d, n_ubatch = %d, flash_attn = %d, is_pp_shared = %d, n_gpu_layers = %d, n_threads = %u, n_threads_batch = %u\n", __func__, n_kv_max, n_batch, n_ubatch, flash_attn, is_pp_shared, n_gpu_layers, ctx_params.n_threads, ctx_params.n_threads_batch);
+    LOG_TEE("%s: n_kv_max = %d, n_batch = %d, n_ubatch = %d, flash_attn = %d, is_pp_shared = %d, n_gpu_layers = %d, n_threads = %u, n_threads_batch = %u\n", __func__, n_kv_max, params.n_batch, params.n_ubatch, params.flash_attn, params.is_pp_shared, params.n_gpu_layers, ctx_params.n_threads, ctx_params.n_threads_batch);
     LOG_TEE("\n");
 
     LOG_TEE("|%6s | %6s | %4s | %6s | %8s | %8s | %8s | %8s | %8s | %8s |\n", "PP",     "TG",     "B",    "N_KV",     "T_PP s",   "S_PP t/s", "T_TG s",   "S_TG t/s", "T s",      "S t/s");

examples/batched.swift/Makefile

@@ -1,6 +1,6 @@
 .PHONY: build
 
 build:
-	xcodebuild -scheme batched_swift -destination "generic/platform=macOS" -derivedDataPath build
-	rm -f ./batched_swift
-	ln -s ./build/Build/Products/Debug/batched_swift ./batched_swift
+	xcodebuild -scheme llama-batched-swift -destination "generic/platform=macOS" -derivedDataPath build
+	rm -f ./llama-batched-swift
+	ln -s ./build/Build/Products/Debug/llama-batched-swift ./llama-batched-swift

examples/batched.swift/Package.swift

@@ -4,7 +4,7 @@
 import PackageDescription
 
 let package = Package(
-    name: "batched_swift",
+    name: "llama-batched-swift",
     platforms: [.macOS(.v12)],
     dependencies: [
         .package(name: "llama", path: "../../"),
@@ -13,7 +13,7 @@ let package = Package(
         // Targets are the basic building blocks of a package, defining a module or a test suite.
         // Targets can depend on other targets in this package and products from dependencies.
         .executableTarget(
-            name: "batched_swift",
+            name: "llama-batched-swift",
             dependencies: ["llama"],
             path: "Sources",
             linkerSettings: [.linkedFramework("Foundation"), .linkedFramework("AppKit")]

examples/batched.swift/README.md

@@ -1,4 +1,4 @@
 This is a swift clone of `examples/batched`.
 
 $ `make`
-$ `./batched_swift MODEL_PATH [PROMPT] [PARALLEL]`
+$ `./llama-batched-swift MODEL_PATH [PROMPT] [PARALLEL]`

examples/batched/CMakeLists.txt

@@ -1,4 +1,4 @@
-set(TARGET batched)
+set(TARGET llama-batched)
 add_executable(${TARGET} batched.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})

examples/batched/README.md

@@ -3,7 +3,7 @@
 The example demonstrates batched generation from a given prompt
 
 ```bash
-./batched ./models/llama-7b-v2/ggml-model-f16.gguf "Hello my name is" 4
+./llama-batched -m ./models/llama-7b-v2/ggml-model-f16.gguf -p "Hello my name is" -np 4
 
 ...
 

examples/batched/batched.cpp

@@ -7,48 +7,31 @@
 #include <string>
 #include <vector>
 
+static void print_usage(int argc, char ** argv, const gpt_params & params) {
+    gpt_params_print_usage(argc, argv, params);
+
+    LOG_TEE("\nexample usage:\n");
+    LOG_TEE("\n    %s -m model.gguf -p \"Hello my name is\" -n 32 -np 4\n", argv[0]);
+    LOG_TEE("\n");
+}
+
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    if (argc == 1 || argv[1][0] == '-') {
-        printf("usage: %s MODEL_PATH [PROMPT] [PARALLEL] [LEN] [NGL]\n" , argv[0]);
-        return 1 ;
+    params.prompt = "Hello my name is";
+    params.n_predict = 32;
+
+    if (!gpt_params_parse(argc, argv, params)) {
+        print_usage(argc, argv, params);
+        return 1;
     }
 
+
     // number of parallel batches
-    int n_parallel = 1;
+    int n_parallel = params.n_parallel;
 
     // total length of the sequences including the prompt
-    int n_len = 32;
-
-    // number of layers to offload to the GPU
-    int n_gpu_layers = 0;
-
-    if (argc >= 2) {
-        params.model = argv[1];
-    }
-
-    if (argc >= 3) {
-        params.prompt = argv[2];
-    }
-
-    if (argc >= 4) {
-        n_parallel = std::atoi(argv[3]);
-    }
-
-    if (argc >= 5) {
-        n_len = std::atoi(argv[4]);
-    }
-
-    if (argc >= 6) {
-        n_gpu_layers = std::atoi(argv[5]);
-    }
-
-    if (params.prompt.empty()) {
-        params.prompt = "Hello my name is";
-    }
-
-    string_process_escapes(params.prompt);
+    int n_predict = 32;
 
     // init LLM
 
@@ -57,9 +40,7 @@ int main(int argc, char ** argv) {
 
     // initialize the model
 
-    llama_model_params model_params = llama_model_default_params();
-
-    model_params.n_gpu_layers = n_gpu_layers;
+    llama_model_params model_params = llama_model_params_from_gpt_params(params);
 
     llama_model * model = llama_load_model_from_file(params.model.c_str(), model_params);
 
@@ -73,18 +54,14 @@ int main(int argc, char ** argv) {
     std::vector<llama_token> tokens_list;
     tokens_list = ::llama_tokenize(model, params.prompt, true);
 
-    const int n_kv_req = tokens_list.size() + (n_len - tokens_list.size())*n_parallel;
+    const int n_kv_req = tokens_list.size() + (n_predict - tokens_list.size())*n_parallel;
 
     // initialize the context
 
-    llama_context_params ctx_params = llama_context_default_params();
+    llama_context_params ctx_params = llama_context_params_from_gpt_params(params);
 
-    ctx_params.seed  = 1234;
     ctx_params.n_ctx   = n_kv_req;
-    ctx_params.n_batch = std::max(n_len, n_parallel);
-    ctx_params.n_seq_max       = n_parallel;
-    ctx_params.n_threads       = params.n_threads;
-    ctx_params.n_threads_batch = params.n_threads_batch == -1 ? params.n_threads : params.n_threads_batch;
+    ctx_params.n_batch = std::max(n_predict, n_parallel);
 
     llama_context * ctx = llama_new_context_with_model(model, ctx_params);
 
@@ -93,9 +70,9 @@ int main(int argc, char ** argv) {
         return 1;
     }
 
-    const int n_ctx    = llama_n_ctx(ctx);
+    const int n_ctx = llama_n_ctx(ctx);
 
-    LOG_TEE("\n%s: n_len = %d, n_ctx = %d, n_batch = %u, n_parallel = %d, n_kv_req = %d\n", __func__, n_len, n_ctx, ctx_params.n_batch, n_parallel, n_kv_req);
+    LOG_TEE("\n%s: n_predict = %d, n_ctx = %d, n_batch = %u, n_parallel = %d, n_kv_req = %d\n", __func__, n_predict, n_ctx, ctx_params.n_batch, n_parallel, n_kv_req);
 
     // make sure the KV cache is big enough to hold all the prompt and generated tokens
     if (n_kv_req > n_ctx) {
@@ -116,14 +93,34 @@ int main(int argc, char ** argv) {
 
     // create a llama_batch
     // we use this object to submit token data for decoding
-    llama_batch batch = llama_batch_init(std::max(tokens_list.size(), (size_t)n_parallel), 0, 1);
+    llama_batch batch = llama_batch_init(std::max(tokens_list.size(), (size_t) n_parallel), 0, n_parallel);
+
+    std::vector<llama_seq_id> seq_ids(n_parallel, 0);
+    for (int32_t i = 0; i < n_parallel; ++i) {
+        seq_ids[i] = i;
+    }
 
     // evaluate the initial prompt
     for (size_t i = 0; i < tokens_list.size(); ++i) {
-        llama_batch_add(batch, tokens_list[i], i, { 0 }, false);
+        llama_batch_add(batch, tokens_list[i], i, seq_ids, false);
     }
     GGML_ASSERT(batch.n_tokens == (int) tokens_list.size());
 
+    if (llama_model_has_encoder(model)) {
+        if (llama_encode(ctx, batch)) {
+            LOG_TEE("%s : failed to eval\n", __func__);
+            return 1;
+        }
+
+        llama_token decoder_start_token_id = llama_model_decoder_start_token(model);
+        if (decoder_start_token_id == -1) {
+            decoder_start_token_id = llama_token_bos(model);
+        }
+
+        llama_batch_clear(batch);
+        llama_batch_add(batch, decoder_start_token_id, 0, seq_ids, false);
+    }
+
     // llama_decode will output logits only for the last token of the prompt
     batch.logits[batch.n_tokens - 1] = true;
 
@@ -132,11 +129,11 @@ int main(int argc, char ** argv) {
         return 1;
     }
 
-    // assign the system KV cache to all parallel sequences
-    // this way, the parallel sequences will "reuse" the prompt tokens without having to copy them
-    for (int32_t i = 1; i < n_parallel; ++i) {
-        llama_kv_cache_seq_cp(ctx, 0, i, -1, -1);
-    }
+    //// assign the system KV cache to all parallel sequences
+    //// this way, the parallel sequences will "reuse" the prompt tokens without having to copy them
+    //for (int32_t i = 1; i < n_parallel; ++i) {
+    //    llama_kv_cache_seq_cp(ctx, 0, i, -1, -1);
+    //}
 
     if (n_parallel > 1) {
         LOG_TEE("\n\n%s: generating %d sequences ...\n", __func__, n_parallel);
@@ -156,7 +153,7 @@ int main(int argc, char ** argv) {
 
     const auto t_main_start = ggml_time_us();
 
-    while (n_cur <= n_len) {
+    while (n_cur <= n_predict) {
         // prepare the next batch
         llama_batch_clear(batch);
 
@@ -192,7 +189,7 @@ int main(int argc, char ** argv) {
             //const llama_token new_token_id = llama_sample_token_greedy(ctx, &candidates_p);
 
             // is it an end of generation? -> mark the stream as finished
-            if (llama_token_is_eog(model, new_token_id) || n_cur == n_len) {
+            if (llama_token_is_eog(model, new_token_id) || n_cur == n_predict) {
                 i_batch[i] = -1;
                 LOG_TEE("\n");
                 if (n_parallel > 1) {

examples/beam-search/beam-search.cpp

@@ -1,188 +0,0 @@
-#include "common.h"
-#include "llama.h"
-
-#include <cassert>
-#include <cinttypes>
-#include <cmath>
-#include <cstdio>
-#include <cstring>
-#include <ctime>
-#include <fstream>
-#include <iostream>
-#include <string>
-#include <vector>
-
-#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
-#include <signal.h>
-#include <unistd.h>
-#elif defined (_WIN32)
-#define WIN32_LEAN_AND_MEAN
-#ifndef NOMINMAX
-#   define NOMINMAX
-#endif
-#include <windows.h>
-#include <signal.h>
-#endif
-
-// Used for debugging to print out beam tokens.
-struct ostream_beam_view {
-    llama_context * ctx;
-    llama_beam_view beam_view;
-};
-
-static std::ostream & operator<<(std::ostream & os, const ostream_beam_view & obv) {
-    os << "p(" << obv.beam_view.p << ") eob(" << std::boolalpha << obv.beam_view.eob << ") tokens(";
-    for (size_t i = 0 ; i < obv.beam_view.n_tokens ; ++i) {
-        os << llama_token_to_piece(obv.ctx, obv.beam_view.tokens[i]);
-    }
-    return os << ')';
-}
-
-// Put here anything you want back in beam_search_callback().
-struct beam_search_callback_data {
-    llama_context * ctx;
-    std::vector<llama_token> response;
-};
-
-// In this case, end-of-beam (eob) is equivalent to end-of-sentence (eos) but this need not always be the same.
-// For example, eob can be flagged due to maximum token length, stop words, etc.
-static bool is_at_eob(const beam_search_callback_data & callback_data, const llama_token * tokens, size_t n_tokens) {
-    return n_tokens && llama_token_is_eog(llama_get_model(callback_data.ctx), tokens[n_tokens-1]);
-}
-
-// Function matching type llama_beam_search_callback_fn_t.
-// Custom callback example is called each time the beams lengths increase:
-//  * Show progress by printing ',' following by number of convergent beam tokens if any.
-//  * When all beams converge to a common prefix, they are made available in beams_state.beams[0].
-//    This is also called when the stop condition is met.
-//    Collect tokens into std::vector<llama_token> response which is pointed to by callback_data.
-static void beam_search_callback(void * callback_data_ptr, llama_beams_state beams_state) {
-    auto& callback_data = *static_cast<beam_search_callback_data*>(callback_data_ptr);
-    // Mark beams as EOS as needed.
-    for (size_t i = 0 ; i < beams_state.n_beams ; ++i) {
-        llama_beam_view& beam_view = beams_state.beam_views[i];
-        if (!beam_view.eob && is_at_eob(callback_data, beam_view.tokens, beam_view.n_tokens)) {
-            beam_view.eob = true;
-        }
-    }
-    printf(",");  // Show progress
-    if (const size_t n = beams_state.common_prefix_length) {
-        callback_data.response.resize(callback_data.response.size() + n);
-        assert(0u < beams_state.n_beams);
-        const llama_token * tokens = beams_state.beam_views[0].tokens;
-        std::copy(tokens, tokens + n, callback_data.response.end() - n);
-        printf("%zu", n);
-    }
-    fflush(stdout);
-#if 1 // DEBUG: print current beams for this iteration
-    std::cout << "\n\nCurrent beams (last_call=" << beams_state.last_call << "):\n";
-    for (size_t i = 0 ; i < beams_state.n_beams ; ++i) {
-        std::cout << "beams["<<i<<"]: " << ostream_beam_view{callback_data.ctx,beams_state.beam_views[i]} << std::endl;
-    }
-#endif
-}
-
-int main(int argc, char ** argv)
-{
-    gpt_params params;
-    //params.n_gpu_layers = 200;
-
-    //---------------------------------
-    // Print help :
-    //---------------------------------
-
-    if ( argc < 2 || argv[1][0] == '-' )
-    {
-        printf( "Usage: %s MODEL_PATH [BEAM_WIDTH=2] [PROMPT]\n" , argv[0] );
-        return 1 ;
-    }
-
-    //---------------------------------
-    // Load parameters :
-    //---------------------------------
-
-    params.model = argv[1];
-
-    params.n_beams = 2 < argc ? std::stoi(argv[2]) : 2;
-
-    if ( argc > 3 )
-    {
-        params.prompt = argv[3];
-    }
-
-    if ( params.prompt.empty() )
-    {
-        params.prompt = "### Request:\nHow many countries are there?\n\n### Response:\n";
-    }
-
-    //---------------------------------
-    // Init LLM :
-    //---------------------------------
-
-    llama_backend_init();
-    llama_numa_init(params.numa);
-
-    llama_model * model;
-    llama_context * ctx;
-
-    std::tie(model, ctx) = llama_init_from_gpt_params( params );
-
-    if ( model == NULL )
-    {
-        fprintf( stderr , "%s: error: unable to load model\n" , __func__ );
-        return 1;
-    }
-
-    //---------------------------------
-    // Tokenize the prompt :
-    //---------------------------------
-
-    std::vector<llama_token> tokens_list = llama_tokenize(ctx, params.prompt, true);
-
-    const size_t max_context_size     = llama_n_ctx( ctx );
-    const size_t max_tokens_list_size = max_context_size - 4 ;
-
-    if (tokens_list.size() > max_tokens_list_size)
-    {
-        fprintf( stderr , "%s: error: prompt too long (%zu tokens, max %zu)\n" ,
-             __func__ , tokens_list.size() , max_tokens_list_size );
-        return 1;
-    }
-
-    fprintf( stderr, "\n\n" );
-
-    // Print the tokens from the prompt :
-
-    for( auto id : tokens_list )
-    {
-        std::cout << llama_token_to_piece(ctx, id);
-    }
-    std::cout << std::flush;
-
-    int n_past = 0;
-
-    if (llama_decode(ctx, llama_batch_get_one(tokens_list.data(), tokens_list.size(), n_past, 0)))
-    {
-        fprintf(stderr, "%s : failed to eval prompt.\n" , __func__ );
-        return 1;
-    }
-    n_past += tokens_list.size();
-
-    beam_search_callback_data callback_data{ctx, {}};
-    size_t const beam_width = static_cast<size_t>(params.n_beams);
-    int const n_predict = 256;
-    llama_beam_search(ctx, beam_search_callback, &callback_data, beam_width, n_past, n_predict);
-
-    std::cout << "\n\n";
-    for (llama_token const token_id : callback_data.response) {
-        std::cout << llama_token_to_piece(ctx,token_id);
-    }
-    std::cout << std::endl;
-
-    llama_free( ctx );
-    llama_free_model( model );
-
-    llama_backend_free();
-
-    return 0;
-}

examples/benchmark/CMakeLists.txt

@@ -1,4 +1,4 @@
-set(TARGET benchmark)
+set(TARGET llama-bench-matmult)
 add_executable(${TARGET} benchmark-matmult.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE llama build_info ${CMAKE_THREAD_LIBS_INIT})

examples/chat-13B.sh

@@ -30,7 +30,7 @@ sed -e "s/\[\[USER_NAME\]\]/$USER_NAME/g" \
      $PROMPT_TEMPLATE > $PROMPT_FILE
 
 # shellcheck disable=SC2086 # Intended splitting of GEN_OPTIONS
-./main $GEN_OPTIONS \
+./llama-cli $GEN_OPTIONS \
   --model "$MODEL" \
   --threads "$N_THREAD" \
   --n_predict "$N_PREDICTS" \

examples/chat-persistent.sh

@@ -62,7 +62,7 @@ fi
 if [[ ! -e "$PROMPT_CACHE_FILE" ]]; then
     echo 'Prompt cache does not exist, building...'
     # Default batch_size to 64 here for better user feedback during initial prompt processing
-    ./main 2>>"$LOG" \
+    ./llama-cli 2>>"$LOG" \
         --batch_size 64 \
         "${OPTS[@]}" \
         --prompt-cache "$PROMPT_CACHE_FILE" \
@@ -109,13 +109,13 @@ while read -e line; do
 
     printf '%s: ' "$AI_NAME" >>"$CUR_PROMPT_FILE"
 
-    ./main 2>>"$LOG" "${OPTS[@]}" \
+    ./llama-cli 2>>"$LOG" "${OPTS[@]}" \
             --prompt-cache "$CUR_PROMPT_CACHE" \
             --prompt-cache-all \
             --file "$CUR_PROMPT_FILE" \
             --reverse-prompt "${USER_NAME}:" \
             --n_predict "$n_predict" |
-        skip_bytes 1 |                  # skip BOS token added by ./main
+        skip_bytes 1 |                  # skip BOS token added by ./llama-cli
         tee "$CUR_PROMPT_FILE.tmp" |    # save prompt + generation to tmp file
         skip_bytes "$n_prompt_len_pre"  # print generation
 
@@ -133,7 +133,7 @@ while read -e line; do
     # TODO get both messages in one go
     if  ! session_size_msg="$(tail -n30 "$LOG" | grep -oE "$SESSION_SIZE_MSG_PATTERN")" ||
         ! sample_time_msg="$(tail -n10 "$LOG" | grep -oE "$SAMPLE_TIME_MSG_PATTERN")"; then
-        echo >&2 "Couldn't get number of tokens from ./main output!"
+        echo >&2 "Couldn't get number of tokens from ./llama-cli output!"
         exit 1
     fi
 
@@ -144,7 +144,7 @@ while read -e line; do
     fi
 
     # Update cache for next prompt in background, ideally during user input
-    ./main >>"$LOG_BG" 2>&1 "${OPTS[@]}" \
+    ./llama-cli >>"$LOG_BG" 2>&1 "${OPTS[@]}" \
           --prompt-cache "$NEXT_PROMPT_CACHE" \
           --file "$NEXT_PROMPT_FILE" \
           --n_predict 1 &

examples/chat-vicuna.sh

@@ -30,7 +30,7 @@ sed -e "s/\[\[USER_NAME\]\]/$USER_NAME/g" \
      $PROMPT_TEMPLATE > $PROMPT_FILE
 
 # shellcheck disable=SC2086 # Intended splitting of GEN_OPTIONS
-./bin/main $GEN_OPTIONS \
+./bin/llama-cli $GEN_OPTIONS \
   --model "$MODEL" \
   --threads "$N_THREAD" \
   --n_predict "$N_PREDICTS" \

examples/chat.sh

@@ -11,6 +11,6 @@ cd ..
 #
 #   "--keep 48" is based on the contents of prompts/chat-with-bob.txt
 #
-./main -m ./models/llama-7b/ggml-model-q4_0.gguf -c 512 -b 1024 -n 256 --keep 48 \
+./llama-cli -m ./models/llama-7b/ggml-model-q4_0.gguf -c 512 -b 1024 -n 256 --keep 48 \
     --repeat_penalty 1.0 --color -i \
     -r "User:" -f prompts/chat-with-bob.txt

examples/convert-legacy-llama.py

@@ -24,14 +24,16 @@ from abc import ABC, abstractmethod
 from concurrent.futures import ProcessPoolExecutor, ThreadPoolExecutor
 from dataclasses import dataclass
 from pathlib import Path
-from typing import TYPE_CHECKING, Any, Callable, ClassVar, IO, Iterable, Literal, Protocol, TypeVar, runtime_checkable, Optional
+from typing import TYPE_CHECKING, Any, Callable, IO, Iterable, Literal, TypeVar, Optional
 
 import numpy as np
-from sentencepiece import SentencePieceProcessor
 
 if 'NO_LOCAL_GGUF' not in os.environ:
-    sys.path.insert(1, str(Path(__file__).parent / 'gguf-py'))
+    # use .parent.parent since we are in "examples" directory
+    sys.path.insert(1, str(Path(__file__).parent.parent / 'gguf-py'))
+
 import gguf
+from gguf import BaseVocab, Vocab, NoVocab, BpeVocab, SentencePieceVocab, LlamaHfVocab
 
 if TYPE_CHECKING:
     from typing_extensions import Self, TypeAlias
@@ -174,7 +176,7 @@ class Params:
     rope_scaling_type: gguf.RopeScalingType | None = None
     f_rope_freq_base: float | None = None
     f_rope_scale: float | None = None
-    n_orig_ctx: int | None = None
+    n_ctx_orig: int | None = None
     rope_finetuned: bool | None = None
 
     ftype: GGMLFileType | None = None
@@ -224,7 +226,7 @@ class Params:
         with open(config_path) as f:
             config = json.load(f)
 
-        rope_scaling_type = f_rope_scale = n_orig_ctx = rope_finetuned = None
+        rope_scaling_type = f_rope_scale = n_ctx_orig = rope_finetuned = None
         rope_scaling = config.get("rope_scaling")
 
         if rope_scaling is not None and (typ := rope_scaling.get("type")):
@@ -234,7 +236,7 @@ class Params:
                 rope_scaling_type = gguf.RopeScalingType.LINEAR
             elif typ == "yarn":
                 rope_scaling_type = gguf.RopeScalingType.YARN
-                n_orig_ctx = rope_scaling['original_max_position_embeddings']
+                n_ctx_orig = rope_scaling['original_max_position_embeddings']
                 rope_finetuned = rope_scaling['finetuned']
             else:
                 raise NotImplementedError(f'Unknown rope scaling type: {typ}')
@@ -270,7 +272,7 @@ class Params:
             f_rope_freq_base  = config.get("rope_theta"),
             rope_scaling_type = rope_scaling_type,
             f_rope_scale      = f_rope_scale,
-            n_orig_ctx        = n_orig_ctx,
+            n_ctx_orig        = n_ctx_orig,
             rope_finetuned    = rope_finetuned,
         )
 
@@ -380,306 +382,6 @@ class Metadata:
         return metadata
 
 
-#
-# vocab
-#
-
-
-@runtime_checkable
-class BaseVocab(Protocol):
-    tokenizer_model: ClassVar[str]
-    name: ClassVar[str]
-
-
-class NoVocab(BaseVocab):
-    tokenizer_model = "no_vocab"
-    name = "no_vocab"
-
-    def __repr__(self) -> str:
-        return "<NoVocab for a model without integrated vocabulary>"
-
-
-@runtime_checkable
-class Vocab(BaseVocab, Protocol):
-    vocab_size: int
-    added_tokens_dict: dict[str, int]
-    added_tokens_list: list[str]
-    fname_tokenizer: Path
-
-    def __init__(self, base_path: Path): ...
-    def all_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]: ...
-
-
-class BpeVocab(Vocab):
-    tokenizer_model = "gpt2"
-    name = "bpe"
-
-    def __init__(self, base_path: Path):
-        added_tokens: dict[str, int] = {}
-
-        if (fname_tokenizer := base_path / 'vocab.json').exists():
-            # "slow" tokenizer
-            with open(fname_tokenizer, encoding="utf-8") as f:
-                self.vocab = json.load(f)
-
-            try:
-                # FIXME: Verify that added tokens here _cannot_ overlap with the main vocab.
-                with open(base_path / ADDED_TOKENS_FILE, encoding="utf-8") as f:
-                    added_tokens = json.load(f)
-            except FileNotFoundError:
-                pass
-        else:
-            # "fast" tokenizer
-            fname_tokenizer = base_path / FAST_TOKENIZER_FILE
-
-            # if this fails, FileNotFoundError propagates to caller
-            with open(fname_tokenizer, encoding="utf-8") as f:
-                tokenizer_json = json.load(f)
-
-            tokenizer_model: dict[str, Any] = tokenizer_json['model']
-            if (
-                tokenizer_model['type'] != 'BPE' or tokenizer_model.get('byte_fallback', False)
-                or tokenizer_json['decoder']['type'] != 'ByteLevel'
-            ):
-                raise FileNotFoundError('Cannot find GPT-2 BPE tokenizer')
-
-            self.vocab = tokenizer_model["vocab"]
-
-            if (added := tokenizer_json.get('added_tokens')) is not None:
-                # Added tokens here can be duplicates of the main vocabulary.
-                added_tokens = {item['content']: item['id']
-                                for item in added
-                                if item['content'] not in self.vocab}
-
-        vocab_size   = len(self.vocab)
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids   = sorted(added_tokens.values())
-        if expected_ids != actual_ids:
-            expected_end_id = vocab_size + len(actual_ids) - 1
-            raise ValueError(f"Expected the {len(actual_ids)} added token ID(s) to be sequential in the range "
-                             f"{vocab_size} - {expected_end_id}; got {actual_ids}")
-
-        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_dict    = added_tokens
-        self.added_tokens_list    = [text for (text, idx) in items]
-        self.vocab_size_base      = vocab_size
-        self.vocab_size           = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer      = fname_tokenizer
-
-    def bpe_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        reverse_vocab = {id: encoded_tok for encoded_tok, id in self.vocab.items()}
-
-        for i, _ in enumerate(self.vocab):
-            yield reverse_vocab[i], 0.0, gguf.TokenType.NORMAL
-
-    def added_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score, gguf.TokenType.CONTROL
-
-    def all_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        yield from self.bpe_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"<BpeVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-
-class SentencePieceVocab(Vocab):
-    tokenizer_model = "llama"
-    name = "spm"
-
-    def __init__(self, base_path: Path):
-        added_tokens: dict[str, int] = {}
-        if (fname_tokenizer := base_path / 'tokenizer.model').exists():
-            # normal location
-            try:
-                with open(base_path / ADDED_TOKENS_FILE, encoding="utf-8") as f:
-                    added_tokens = json.load(f)
-            except FileNotFoundError:
-                pass
-        elif not (fname_tokenizer := base_path.parent / 'tokenizer.model').exists():
-            # not found in alternate location either
-            raise FileNotFoundError('Cannot find tokenizer.model')
-
-        self.sentencepiece_tokenizer = SentencePieceProcessor()
-        self.sentencepiece_tokenizer.LoadFromFile(str(fname_tokenizer))
-        vocab_size = self.sentencepiece_tokenizer.vocab_size()
-
-        new_tokens       = {id: piece for piece, id in added_tokens.items() if id >= vocab_size}
-        expected_new_ids = list(range(vocab_size, vocab_size + len(new_tokens)))
-        actual_new_ids   = sorted(new_tokens.keys())
-
-        if expected_new_ids != actual_new_ids:
-            raise ValueError(f"Expected new token IDs {expected_new_ids} to be sequential; got {actual_new_ids}")
-
-        # Token pieces that were added to the base vocabulary.
-        self.added_tokens_dict  = added_tokens
-        self.added_tokens_list  = [new_tokens[id] for id in actual_new_ids]
-        self.vocab_size_base    = vocab_size
-        self.vocab_size         = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer    = fname_tokenizer
-
-    def sentencepiece_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        tokenizer = self.sentencepiece_tokenizer
-        for i in range(tokenizer.vocab_size()):
-            piece = tokenizer.IdToPiece(i)
-            text         = piece.encode("utf-8")
-            score: float = tokenizer.GetScore(i)
-
-            toktype = gguf.TokenType.NORMAL
-            if tokenizer.IsUnknown(i):
-                toktype = gguf.TokenType.UNKNOWN
-            if tokenizer.IsControl(i):
-                toktype = gguf.TokenType.CONTROL
-
-            # NOTE: I think added_tokens are user defined.
-            # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
-            # if tokenizer.is_user_defined(i): toktype = gguf.TokenType.USER_DEFINED
-
-            if tokenizer.IsUnused(i):
-                toktype = gguf.TokenType.UNUSED
-            if tokenizer.IsByte(i):
-                toktype = gguf.TokenType.BYTE
-
-            yield text, score, toktype
-
-    def added_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score, gguf.TokenType.USER_DEFINED
-
-    def all_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        yield from self.sentencepiece_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-
-class LlamaHfVocab(Vocab):
-    tokenizer_model = "llama"
-    name = "hfft"
-
-    def __init__(self, base_path: Path):
-        fname_tokenizer = base_path / FAST_TOKENIZER_FILE
-        # if this fails, FileNotFoundError propagates to caller
-        with open(fname_tokenizer, encoding='utf-8') as f:
-            tokenizer_json = json.load(f)
-
-        # pre-check so we know if we need transformers
-        tokenizer_model: dict[str, Any] = tokenizer_json['model']
-        is_llama3 = (
-            tokenizer_model['type'] == 'BPE' and tokenizer_model.get('ignore_merges', False)
-            and not tokenizer_model.get('byte_fallback', True)
-        )
-        if is_llama3:
-            raise TypeError('Llama 3 must be converted with BpeVocab')
-
-        if not is_llama3 and (
-            tokenizer_model['type'] != 'BPE' or not tokenizer_model.get('byte_fallback', False)
-            or tokenizer_json['decoder']['type'] != 'Sequence'
-        ):
-            raise FileNotFoundError('Cannot find Llama BPE tokenizer')
-
-        try:
-            from transformers import AutoTokenizer
-        except ImportError as e:
-            raise ImportError(
-                "To use LlamaHfVocab, please install the `transformers` package. "
-                "You can install it with `pip install transformers`."
-            ) from e
-
-        # Allow the tokenizer to default to slow or fast versions.
-        # Explicitly set tokenizer to use local paths.
-        self.tokenizer = AutoTokenizer.from_pretrained(
-            base_path,
-            cache_dir=base_path,
-            local_files_only=True,
-        )
-        assert self.tokenizer.is_fast  # assume tokenizer.json is used
-
-        # Initialize lists and dictionaries for added tokens
-        self.added_tokens_list = []
-        self.added_tokens_dict = dict()
-        self.added_tokens_ids  = set()
-
-        # Process added tokens
-        for tok, tokidx in sorted(
-            self.tokenizer.get_added_vocab().items(), key=lambda x: x[1]
-        ):
-            # Only consider added tokens that are not in the base vocabulary
-            if tokidx >= self.tokenizer.vocab_size:
-                self.added_tokens_list.append(tok)
-                self.added_tokens_dict[tok] = tokidx
-                self.added_tokens_ids.add(tokidx)
-
-        # Store special tokens and their IDs
-        self.specials = {
-            tok: self.tokenizer.get_vocab()[tok]
-            for tok in self.tokenizer.all_special_tokens
-        }
-        self.special_ids = set(self.tokenizer.all_special_ids)
-
-        # Set vocabulary sizes
-        self.vocab_size_base = self.tokenizer.vocab_size
-        self.vocab_size      = self.vocab_size_base + len(self.added_tokens_list)
-
-        self.fname_tokenizer = fname_tokenizer
-
-    def hf_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        reverse_vocab = {
-            id: encoded_tok for encoded_tok, id in self.tokenizer.get_vocab().items()
-        }
-
-        for token_id in range(self.vocab_size_base):
-            # Skip processing added tokens here
-            if token_id in self.added_tokens_ids:
-                continue
-
-            # Convert token text to bytes
-            token_text = reverse_vocab[token_id].encode("utf-8")
-
-            # Yield token text, score, and type
-            yield token_text, self.get_token_score(token_id), self.get_token_type(
-                token_id, token_text, self.special_ids  # Reuse already stored special IDs
-            )
-
-    def get_token_type(self, token_id: int, token_text: bytes, special_ids: set[int]) -> gguf.TokenType:
-        # Special case for byte tokens
-        if re.fullmatch(br"<0x[0-9A-Fa-f]{2}>", token_text):
-            return gguf.TokenType.BYTE
-
-        # Determine token type based on whether it's a special token
-        return gguf.TokenType.CONTROL if token_id in special_ids else gguf.TokenType.NORMAL
-
-    def get_token_score(self, token_id: int) -> float:
-        # Placeholder for actual logic to determine the token's score
-        # This needs to be implemented based on specific requirements
-        return -1000.0  # Default score
-
-    def added_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        for text in self.added_tokens_list:
-            if text in self.specials:
-                toktype = self.get_token_type(self.specials[text], b'', self.special_ids)
-                score = self.get_token_score(self.specials[text])
-            else:
-                toktype = gguf.TokenType.USER_DEFINED
-                score = -1000.0
-
-            yield text.encode("utf-8"), score, toktype
-
-    def has_newline_token(self):
-        return "<0x0A>" in self.tokenizer.vocab or "\n" in self.tokenizer.vocab
-
-    def all_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
-        yield from self.hf_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"<LlamaHfVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-
 #
 # data loading
 # TODO: reuse (probably move to gguf.py?)
@@ -1162,8 +864,8 @@ class OutputFile:
             self.gguf.add_rope_scaling_type(params.rope_scaling_type)
             self.gguf.add_rope_scaling_factor(params.f_rope_scale)
 
-        if params.n_orig_ctx is not None:
-            self.gguf.add_rope_scaling_orig_ctx_len(params.n_orig_ctx)
+        if params.n_ctx_orig is not None:
+            self.gguf.add_rope_scaling_orig_ctx_len(params.n_ctx_orig)
 
         if params.rope_finetuned is not None:
             self.gguf.add_rope_scaling_finetuned(params.rope_finetuned)

examples/convert-llama2c-to-ggml/CMakeLists.txt

@@ -1,4 +1,4 @@
-set(TARGET convert-llama2c-to-ggml)
+set(TARGET llama-convert-llama2c-to-ggml)
 add_executable(${TARGET} convert-llama2c-to-ggml.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})

examples/convert-llama2c-to-ggml/README.md

@@ -8,7 +8,7 @@ To convert the model first download the models from the [llama2.c](https://githu
 
 After successful compilation, following usage options are available:
 ```
-usage: ./convert-llama2c-to-ggml [options]
+usage: ./llama-convert-llama2c-to-ggml [options]
 
 options:
   -h, --help                       show this help message and exit
@@ -19,10 +19,10 @@ options:
 
 An example command using a model from [karpathy/tinyllamas](https://huggingface.co/karpathy/tinyllamas) is as follows:
 
-`$ ./convert-llama2c-to-ggml --copy-vocab-from-model llama-2-7b-chat.gguf.q2_K.bin --llama2c-model stories42M.bin --llama2c-output-model stories42M.gguf.bin`
+`$ ./llama-convert-llama2c-to-ggml --copy-vocab-from-model llama-2-7b-chat.gguf.q2_K.bin --llama2c-model stories42M.bin --llama2c-output-model stories42M.gguf.bin`
 
 Note: The vocabulary for `stories260K.bin` should be its own tokenizer `tok512.bin` found in [karpathy/tinyllamas/stories260K](https://huggingface.co/karpathy/tinyllamas/tree/main/stories260K).
 
 Now you can use the model with a command like:
 
-`$ ./main -m stories42M.gguf.bin -p "One day, Lily met a Shoggoth" -n 500 -c 256`
+`$ ./llama-cli -m stories42M.gguf.bin -p "One day, Lily met a Shoggoth" -n 500 -c 256`

examples/cvector-generator/CMakeLists.txt

@@ -1,5 +1,5 @@
-set(TARGET beam-search)
-add_executable(${TARGET} beam-search.cpp)
+set(TARGET llama-cvector-generator)
+add_executable(${TARGET} cvector-generator.cpp pca.hpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
 target_compile_features(${TARGET} PRIVATE cxx_std_11)

examples/cvector-generator/README.md

@@ -0,0 +1,45 @@
+# cvector-generator
+
+This example demonstrates how to generate a control vector using gguf models.
+
+Related PRs:
+- [Add support for control vectors](https://github.com/ggerganov/llama.cpp/pull/5970)
+- (Issue) [Generate control vector using llama.cpp](https://github.com/ggerganov/llama.cpp/issues/6880)
+- [Add cvector-generator example](https://github.com/ggerganov/llama.cpp/pull/7514)
+
+## Examples
+
+```sh
+# CPU only
+./cvector-generator -m ./llama-3.Q4_K_M.gguf
+
+# With GPU
+./cvector-generator -m ./llama-3.Q4_K_M.gguf -ngl 99
+
+# With advanced options
+./cvector-generator -m ./llama-3.Q4_K_M.gguf -ngl 99 --pca-iter 2000 --pca-batch 100
+
+# Using mean value instead of PCA
+./cvector-generator -m ./llama-3.Q4_K_M.gguf --method mean
+
+# To see help message
+./cvector-generator -h
+# Then, have a look at "cvector" section
+```
+
+## Tips and tricks
+
+If you have multiple lines per prompt, you can escape the newline character (change it to `\n`). For example:
+
+```
+<|im_start|>system\nAct like a person who is extremely happy.<|im_end|>
+<|im_start|>system\nYou are in a very good mood today<|im_end|>
+```
+
+Example to use output file with `llama-cli`:
+
+(Tips: The control vector works better when apply to layers higher than 10)
+
+```sh
+./llama-cli -m ./llama-3.Q4_K_M.gguf -p "<|start_header_id|>system<|end_header_id|>\n\nYou are a helpful assistant<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nSing a song<|im_end|><|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n" --special --control-vector-scaled ./control_vector.gguf 0.8 --control-vector-layer-range 10 31
+```

examples/cvector-generator/completions.txt

@@ -0,0 +1,582 @@
+
+That game
+I can see
+Hmm, this
+I can relate to
+Who is
+I understand the
+Ugh,
+What the hell was
+Hey, did anyone
+Although
+Thank you for choosing
+What are you
+Oh w
+How dare you open
+It was my pleasure
+I'm hon
+I appreciate that you
+Are you k
+Whoever left this
+It's always
+Ew,
+Hey, I l
+Hello? Is someone
+I understand that
+That poem
+Aww, poor
+Hey, it
+Alright, who
+I didn't
+Well, life
+The document
+Oh no, this
+I'm concerned
+Hello, this is
+This art
+Hmm, this drink
+Hi there!
+It seems
+Is
+Good
+I can't
+Ex
+Who are
+I can see that
+Wow,
+Today is a
+Hey friend
+Sometimes friends
+Oh, this old
+The weather outside
+This place is sur
+I appreciate your input
+Thank you for the
+Look at
+I'm disappoint
+To my
+How dare you
+That's an
+This piece of art
+Eww
+This park is
+This is incredible
+Oh no, someone
+Exc
+Well, it'
+I warned
+Hey, I understand
+Hey, I saw
+How dare you go
+What the he
+Hey
+It's
+Hello? Hello?
+It
+Oh no!
+This is the perfect
+Good morning,
+Oh no, there
+It's so
+Yeah
+Uh,
+Hello everyone
+Who turned off
+The weather
+Who'
+Hey, this
+Wait,
+Eww, gross
+Excuse
+It seems like you
+Thank you so
+What happened?
+Oh my g
+I am deeply sad
+I war
+Okay, let'
+Hey, that
+That was a beautiful
+Oh no! That
+What happened
+Hey there
+The artist'
+What?!
+Hey, it'
+I am disappoint
+It seems like
+Oh no! The
+This park is a
+If you
+Yes! I did
+It sounds
+What
+Who is it
+Hmm, that
+That's strange
+Yeah, that was
+That's interesting
+This park
+What the hell
+Who is that
+I feel like my
+Oh well
+What the hell is
+Hello? Hello
+To my dearest
+Bless you!\"
+Thank you for
+Oh, looks like
+Can you please
+This place is
+Eww, what
+Bless you
+Is everything
+Hey, I just
+Whoever left these
+Well, that'
+I feel
+Hey, do you
+It's sad
+Oh no, it
+Hey, that'
+Oh my god,
+Thank you,
+Hello little one,
+I apolog
+Hey team, I
+How dare you read
+Who is this and
+Whoever left
+Hi there! W
+A
+If you have
+I was
+U
+Bless
+Well, this
+Oh, I'
+It's a
+Eww,
+Is everything okay?
+Oh, I
+Hello, can you
+Al
+That was a great
+What are
+I understand that not
+Oh no, not
+Who is it?\"
+Hey, can we
+Whoever is taking
+I would love to
+Hey, I noticed
+Hey, could
+I understand that there
+Hello?
+D
+Oh man, I
+Thank you so much
+Oh no, my
+Dear [Name
+Uh
+I remember
+Hey, who
+Well, it
+Are you
+I understand that it
+Hey, is
+I would
+Who is this
+Excuse me
+Alright
+I am thrilled
+Sometimes friends have
+Who the
+It's interesting
+I would love
+E
+Hello? Is anyone
+Well, this is
+This place
+Well,
+I warned you
+Hey, watch where
+Oh my
+That'
+Sometimes friends have different
+I understand that everyone
+What?
+What do these notes
+I can relate
+I'm not
+I understand
+To my dear
+Guys
+Well
+Hey, I appreciate
+Wow, what
+Dear
+That melody
+Who the hell
+Today is
+Hello little
+Wow, look
+That's great
+Love is never wrong
+I'm having
+Whoa, did
+Ugh
+Can you please provide
+I miss you,
+I feel uncom
+I know
+Ugh, this
+Hey, watch
+Oh great, a
+I didn
+Okay
+That game of char
+Oh
+I appreciate
+Who's there
+I am so
+Oh great, someone
+Hey, could you
+I remember wondering
+Wait, what?
+What do
+Hello? Can
+Hey there,
+That game of
+This is incred
+Oh my gosh
+Oh great, f
+I appreciate your
+It sounds like
+What the heck
+Okay, I understand
+Ew
+I understand that this
+Uh, hi
+Hi everyone!
+What the hell?
+Thank you for your
+Oh no, the
+Wow, I
+Who turned
+Dear [
+Whoever
+This is a
+Whoa, he
+What in the world
+Although the physical
+Hello, who is
+That's amaz
+Hey, I know
+Okay, that
+Hi everyone
+Hey, is everything
+I understand your fr
+Oh no, poor
+Oh, look
+Good morning
+Ew, gross
+Oh no, did
+Look at the family
+Hey team
+Yes!
+Hey, can I
+Okay, that'
+It's great
+Love is
+Hey, what
+Good morning, world
+Who is it?
+That poem really reson
+I
+That's
+I understand the task
+Gu
+Hello? Who'
+This postcard is
+Whoa,
+Oh, that
+I understand that I
+Whoever is
+Hello? Who is
+I'm really
+Wow, this
+Can
+This artwork really
+This is a shame
+I miss you too
+Who are you?
+Today is a difficult
+Hey, just
+Are you okay
+I am
+Hi,
+Wow, that
+Hey there! Can
+Okay, stay
+Oh great, just
+Yeah,
+Hello? Can you
+Oh, looks
+Thank you for sharing
+I'm glad
+Hey, is that
+Hmm
+It was my
+It sounds like you
+Wow, your
+I was promised certain
+That was such a
+Thank
+Excuse you
+That was
+Hey team,
+I feel un
+It was
+What'
+Hey friend, I
+How
+Saying goodbye
+That
+It's heart
+How dare
+Oh,
+Hello, may
+What's this
+Thank you for recogn
+Aww, that
+Oh, I remember
+Hmm, that'
+I miss
+I know this
+Wait
+Is everything okay
+Who is that person
+Wow, you
+Oh great
+I'm sad
+Wow, the
+I am very disappoint
+Who turned off the
+I understand that things
+I'm very
+Hi
+That's very
+Okay, I
+Oh no,
+Wow, there
+What's wrong
+I apologize for
+Hey, I
+Can I help you
+Oh, I didn
+Alright,
+Oh wow,
+Oh my goodness
+I know this event
+What in the
+Saying
+Yeah, that
+Guys, I
+Hey, this v
+This post
+Are
+Hey, can
+Hello? Is
+I can only imagine
+Oh, that sounds
+Hey, is anyone
+I am disappointed
+Hello,
+Hey everyone, I
+That was such
+It's okay
+The artist
+Whoa
+I understand that mistakes
+Can I help
+Who
+Hi everyone! I
+Hey, can you
+Wow, how
+Today
+Oh no, I
+Oh well, I
+Well, that
+This is the
+Yes! I finally
+Hey there little
+Hello everyone!
+Love is never
+Look at the
+This postcard
+Oh great,
+Can I
+Hmm, this is
+I understand your
+Oh, look at
+B
+I'm so
+Whoa, this
+W
+Oh, this
+Sometimes
+This piece of
+What the
+That was a
+Hey, do
+Oh no
+Whoa, what
+I feel like I
+The documentary
+Hello
+Hello little one
+I understand that my
+Eww, that
+Wow, an
+Yes! Finally,
+Although the physical location
+Whoever is watching
+That movie
+I remember wondering about
+Hey there, little
+Who's
+Hello, who
+Hello everyone! Thank
+Hello, can
+That's too
+Hey, just wanted
+Hey there, I
+Saying good
+Hey there!
+Who is there?
+Oh my good
+I am very
+Oh no, what
+Wow, thank
+I was promised
+Hi, is
+Hey, I'
+Guys, the
+Oh no, that
+Who is there
+Hello, this
+That movie really touched
+If you have something
+The documentary was
+I'm starting
+Are you kidd
+That movie really
+Hey everyone,
+Thank you for considering
+I didn'
+Yes! I
+Can you
+Oh my god
+Hey, whoever
+That melody really
+Thank you, little
+Hello, may I
+Look
+Wow, we
+It looks
+What do these
+Oh wow
+I apologize
+What are you all
+It's such
+It's clear
+Hey, I was
+Hey friend,
+I can only
+The weather outside is
+Eww, this
+I miss you
+Wow
+Aww,
+Hi, is there
+This artwork
+Okay,
+Oh well,
+This
+I'
+Say
+Hey there little gu
+Hmm,
+Whoa, who
+I am thr
+Oh man
+Okay, stay calm
+I'm happy
+Oh, this cur
+Oh man,
+I'm sorry
+Hello? Who
+What?! That
+This piece
+Hey everyone
+That's so
+Are you okay?
+What happened? Where
+Hi there
+The
+Who the hell entered
+I can
+Guys,
+What's
+What in
+It's important
+I'm
+I'm coming
+It'
+Yes! Finally
+Wait, what
+Wow, reading
+I'm surprised
+Hey, did
+Hey,
+Okay, let
+I understand that you
+Who the hell threw
+Eww, who
+Thank you for thinking
+Who is this?\"
+I am deeply
+Thank you for including
+Oh no, an
+It looks like you
+Aww
+I'm confused
+Wow, it
+That poem really
+Yes
+Hey there, is
+Hey, what'
+Thank you for remember
+To
+This is
+Thank you for making
+I can'
+That mel
+Wow, they
+I feel like
+Although the
+Who are you
+Love
+If
+What the hell are
+I am so sad
+Oh, I found
+Thank you
+It looks like
+Well, life is
+I appreciate that
+The artist's
+Whoa, that
+It's never

examples/cvector-generator/cvector-generator.cpp

@@ -0,0 +1,503 @@
+#include "common.h"
+#include "llama.h"
+#include "ggml.h"
+#include "pca.hpp"
+#include "mean.hpp"
+
+#ifdef GGML_USE_CUDA
+#include "ggml-cuda.h"
+#endif
+
+#ifdef GGML_USE_METAL
+#include "ggml-metal.h"
+#endif
+
+#include <cstdio>
+#include <string>
+#include <tuple>
+#include <vector>
+#include <algorithm>
+#include <iostream>
+#include <fstream>
+#include <climits>
+
+
+//////////////////////////////////////////////////
+// utils
+
+template <class Iter>
+static std::string tokens_to_str(llama_context * ctx, Iter begin, Iter end) {
+    std::string ret;
+    for (; begin != end; ++begin) {
+        ret += llama_token_to_piece(ctx, *begin);
+    }
+
+    return ret;
+}
+
+static void print_usage(int argc, char ** argv, const gpt_params & params) {
+    gpt_params_print_usage(argc, argv, params);
+
+    printf("\nexample usage:\n");
+    printf("\n    CPU only:   %s -m ./llama-3.Q4_K_M.gguf\n", argv[0]);
+    printf("\n    with GPU:   %s -m ./llama-3.Q4_K_M.gguf -ngl 99\n", argv[0]);
+    printf("\n    advanced:   %s -m ./llama-3.Q4_K_M.gguf -ngl 99 --pca-iter 2000 --pca-batch 100\n", argv[0]);
+    printf("\n    using mean: %s -m ./llama-3.Q4_K_M.gguf --method mean\n", argv[0]);
+    printf("\n");
+}
+
+//////////////////////////////////////////////////
+
+
+// cb_eval is reused for each pair of positive - negative prompt
+struct callback_data {
+    ggml_context * ctx_ggml = nullptr;   // holds v_pos, v_neg, v_diff_filtered
+
+    int n_layers = 0;
+    int n_tokens = 0;
+    bool is_eval_pos = true;
+
+    // each element of the vector correspond to one layer
+    std::vector<struct ggml_tensor *> v_pos; // vector of matrices of size [n_embd, n_tokens]
+    std::vector<struct ggml_tensor *> v_neg; // vector of matrices of size [n_embd, n_tokens]
+    std::vector<struct ggml_tensor *> v_diff_filtered;   // vector of matrices of size [n_embd, n_nonzero_rows]. NOTE: n_nonzero_rows maybe different for each layer
+
+    // save a tensor into either v_pos or v_neg (decided by is_eval_pos)
+    void save_tensor_for_layer(struct ggml_tensor * t) {
+        GGML_ASSERT(t->type == GGML_TYPE_F32);
+
+        if (ctx_ggml == nullptr) {
+            // alloc a new ctx_ggml if needed
+            struct ggml_init_params params_ggml = {
+                /*.mem_size   =*/ ggml_tensor_overhead() * n_layers * 3u,
+                /*.mem_buffer =*/ NULL,
+                /*.no_alloc   =*/ true,
+            };
+            ctx_ggml = ggml_init(params_ggml);
+        }
+
+        // copy tensor data
+        auto n_bytes = ggml_nbytes(t);
+        struct ggml_tensor * t_layer = ggml_new_tensor_2d(ctx_ggml, t->type, t->ne[0], t->ne[1]);
+        t_layer->data = malloc(n_bytes); // TODO @ngxson : get rid of this malloc somehow
+        ggml_backend_tensor_get(t, t_layer->data, 0, n_bytes);
+        ggml_set_name(t_layer, ggml_get_name(t));
+        //print_debug_tensor(t_layer);
+
+        if (is_eval_pos) {
+            v_pos.push_back(t_layer);
+        } else {
+            v_neg.push_back(t_layer);
+        }
+    }
+
+    // calculate diff (v_pos - v_neg) and place the result back to v_pos
+    // all zero rows in the diff tensor will also be removed
+    // NOTE: final layer is ignored. we only have (n_layers - 1) to process
+    std::vector<struct ggml_tensor *> calc_diff() {
+        for (float il = 0; il < v_pos.size(); il++) {
+            float * a = (float *) v_pos[il]->data;
+            float * b = (float *) v_neg[il]->data;
+            size_t n_elem = ggml_nelements(v_pos[il]);
+            for (size_t j = 0; j < n_elem; j++) {
+                a[j] -= b[j];
+            }
+            //print_debug_tensor(v_pos[i]);
+            auto diff_filtered = filter_nonzero_rows(v_pos[il]);
+            v_diff_filtered.push_back(diff_filtered);
+        }
+        return v_diff_filtered; // for convinient, we return the result std::vector
+    }
+
+    // delete zero rows from a given 2D tensor
+    struct ggml_tensor * filter_nonzero_rows(struct ggml_tensor * a) {
+        //printf("filter_nonzero_rows\n");
+        auto is_row_all_zeros = [](struct ggml_tensor * t, int row, float eps) -> bool {
+            // check if given row containing all zero elements
+            int n_cols = t->ne[0]; // hint: should be equal to n_embd
+            for (int col = 0; col < n_cols; ++col) {
+                if (ggml_get_f32_nd(t, col, row, 0, 0) > eps) {
+                    return false;
+                }
+            }
+            return true;
+        };
+        std::vector<int> rows_to_copy; // the idx of non-zero cols (to be copied to row of diff_filtered)
+        for (int i_row = 0; i_row < a->ne[1]; i_row++) {
+            if (!is_row_all_zeros(a, i_row, 1e-6)) {
+                rows_to_copy.push_back(i_row);
+            }
+        }
+
+        // get "n_nonzero_rows" for the output "diff_filtered"
+        int n_nonzero_rows = rows_to_copy.size();
+        //printf("n_nonzero_rows: %d\n", n_nonzero_rows);
+        int n_embd = a->ne[0];
+        GGML_ASSERT(n_nonzero_rows > 0);
+
+        // diff_filtered: [n_embd, n_nonzero_rows]
+        struct ggml_tensor * diff_filtered = ggml_new_tensor_2d(
+            ctx_ggml, GGML_TYPE_F32, n_embd, n_nonzero_rows);
+        ggml_format_name(diff_filtered, "diff_filtered_%s", a->name);
+        diff_filtered->data = malloc(ggml_nbytes(diff_filtered));
+
+        // copy non-zero rows
+        for (int dest_row = 0; dest_row < n_nonzero_rows; dest_row++) {
+            int src_row = rows_to_copy[dest_row];
+            for (int i = 0; i < n_embd; i++) {
+                float src_elem = ggml_get_f32_nd(a, i, src_row, 0, 0);
+                ggml_set_f32_nd(diff_filtered, i, dest_row, 0, 0, src_elem);
+            }
+        }
+
+        //print_debug_tensor(diff_filtered);
+
+        return diff_filtered;
+    }
+
+    // we don't implement destructor, because we want to reuse callback_data. we just want to free the tensors
+    void reset() {
+        for (auto ptr : v_pos) free(ptr->data);
+        for (auto ptr : v_neg) free(ptr->data);
+        for (auto ptr : v_diff_filtered) free(ptr->data);
+        v_pos.clear();
+        v_neg.clear();
+        v_diff_filtered.clear();
+        if (ctx_ggml) {
+            ggml_free(ctx_ggml);
+        }
+        ctx_ggml = nullptr;
+    }
+};
+
+/**
+ * process_ctx is used to store the ggml context for pre-post processing the diff vectors
+ * in short, input => v_diff and output => v_final
+ */
+struct train_context {
+    ggml_context * ctx_ggml;
+    int n_embd;
+    int n_layers;
+
+    /* pair of prompts to be used for generating final vector */
+    std::vector<std::string> positive_entries;
+    std::vector<std::string> negative_entries;
+
+    // each element of the vector correspond to one layer
+    // NOTE: the last layer is discard. therefore, we will have (n_layers - 1) elements here
+    // NOTE (2): v_diff is transposed from v_diff_tmp
+    std::vector<struct ggml_tensor *> v_diff;  // vector of matrices of size [m, n_embd] where m ~ n_tokens * n_completions (v_diff contains no zero-rows)
+    std::vector<struct ggml_tensor *> v_final; // vector of vectors of size [n_embd] to be written to file
+
+    // to easily re-alloc when concat v_diff, we temporary store v_diff in a vector instead of a tensor
+    // v_diff_tmp will get converted unto v_diff later on
+    std::vector<std::vector<uint8_t>> v_diff_tmp;
+
+    train_context(int n_embd_, int n_layers_) {
+        n_embd = n_embd_;
+        n_layers = n_layers_;
+        struct ggml_init_params params_ggml = {
+            /*.mem_size   =*/ ggml_tensor_overhead() * (n_layers - 1) * 2u,
+            /*.mem_buffer =*/ NULL,
+            /*.no_alloc   =*/ true,
+        };
+        ctx_ggml = ggml_init(params_ggml);
+        for (int il = 0; il < n_layers - 1; il++) {
+            std::vector<uint8_t> empty;
+            v_diff_tmp.push_back(empty);
+            auto t = ggml_new_tensor_1d(ctx_ggml, GGML_TYPE_F32, n_embd);
+            t->data = malloc(ggml_nbytes(t)); // TODO: get rid of malloc if possible
+            v_final.push_back(t);
+        }
+    }
+
+    // add new rows into existing tensor in v_diff_tmp
+    void concat_diff_tmp(const std::vector<struct ggml_tensor *> & diff_filtered) {
+        GGML_ASSERT((int) diff_filtered.size() == n_layers - 1);
+        for (int il = 0; il < n_layers - 1; il++) {
+            auto t = diff_filtered[il];
+            auto & diff_tmp = v_diff_tmp[il];
+            size_t curr_size = diff_tmp.size();
+            diff_tmp.resize(curr_size + ggml_nbytes(t));
+            memcpy(diff_tmp.data() + curr_size, t->data, ggml_nbytes(t));
+        }
+    }
+
+    // build the v_diff tensors from v_diff_tmp (v_diff need to be transposed)
+    // TODO @ngxson : maybe add option NOT to transpose v_diff; will be useful for "mean" method
+    void build_v_diff(bool transpose) {
+        printf("build_v_diff\n");
+        for (int il = 0; il < n_layers - 1; il++) {
+            auto & diff_tmp = v_diff_tmp[il];
+            int n_elem = diff_tmp.size() / sizeof(float);
+            GGML_ASSERT(n_elem % n_embd == 0);
+            int n_rows = n_elem / n_embd;
+            struct ggml_tensor * diff = transpose
+                ? ggml_new_tensor_2d(ctx_ggml, GGML_TYPE_F32, n_rows, n_embd)
+                : ggml_new_tensor_2d(ctx_ggml, GGML_TYPE_F32, n_embd, n_rows);
+            ggml_set_name(diff, (std::string("diff_") + std::to_string(il)).c_str());
+            diff->data = malloc(ggml_nbytes(diff)); // TODO: get rid of this malloc if possible
+            if (transpose) {
+                // copy data & transpose
+                float * arr = (float *) diff_tmp.data();
+                for (int ir = 0; ir < n_rows; ++ir) {
+                    for (int ic = 0; ic < n_embd; ++ic) {
+                        float f = arr[ir*n_embd + ic];
+                        ggml_set_f32_nd(diff, ir, ic, 0, 0, f);
+                    }
+                }
+            } else {
+                // only copy
+                memcpy(diff->data, diff_tmp.data(), ggml_nbytes(diff));
+            }
+            v_diff.push_back(diff);
+            print_debug_tensor(diff);
+            // free memory of diff_tmp
+            diff_tmp.resize(0);
+        }
+    }
+
+    ~train_context() {
+        for (auto ptr : v_final) free(ptr->data);
+        for (auto ptr : v_diff) free(ptr->data);
+        // no need to free v_diff_tmp, since we didn't use malloc
+        ggml_free(ctx_ggml);
+    }
+};
+
+struct tokenized_prompt {
+    std::vector<llama_token> tokens_pos;
+    std::vector<llama_token> tokens_neg;
+    size_t max_seq_len;
+
+    tokenized_prompt(llama_context * ctx, std::string pos, std::string neg) {
+        const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
+        tokens_pos = ::llama_tokenize(ctx, pos, add_bos, true);
+        tokens_neg = ::llama_tokenize(ctx, neg, add_bos, true);
+        max_seq_len = std::max(tokens_pos.size(), tokens_neg.size());
+        padding_seq(ctx, tokens_pos, max_seq_len);
+        padding_seq(ctx, tokens_neg, max_seq_len);
+    }
+
+    void padding_seq(llama_context * ctx, std::vector<llama_token> & tokens, size_t len) {
+        // TODO: customize padding token
+        std::vector<llama_token> pad_tokens = ::llama_tokenize(ctx, " ", false);
+        llama_token pad_tok = pad_tokens.back();
+        while (tokens.size() < len) {
+            tokens.push_back(pad_tok);
+        }
+    }
+};
+
+//////////////////////////////////////////////////
+
+template <typename T>
+static std::string to_string(const T & val) {
+    std::stringstream ss;
+    ss << val;
+    return ss.str();
+}
+
+static std::vector<std::string> ctrlvec_load_prompt_file(std::string path, bool skip_empty_lines) {
+    std::vector<std::string> output;
+    std::ifstream file(path);
+    if (!file.is_open()) {
+        fprintf(stderr, "error: unable to open file: %s\n", path.c_str());
+        exit(1);
+    }
+    std::string line;
+    while (std::getline(file, line)) {
+        bool is_skip = skip_empty_lines && line.empty();
+        if (!is_skip) {
+            string_process_escapes(line);
+            output.push_back(line);
+        }
+    }
+    file.close();
+    return output;
+}
+
+//////////////////////////////////////////////////
+
+static bool cb_eval(struct ggml_tensor * t, bool ask, void * user_data) {
+    auto * cb_data = (callback_data *) user_data;
+    static const char * l_out_name = "l_out";
+    const bool is_l_out = strncmp(t->name, l_out_name, strlen(l_out_name)) == 0;
+
+    if (ask) {
+        return is_l_out;
+    }
+
+    if (!is_l_out || t->ne[1] != cb_data->n_tokens) {
+        return true;
+    }
+
+    // save the tensor to current context
+    cb_data->save_tensor_for_layer(t);
+    return true;
+}
+
+static bool get_hidden_layers(llama_context * ctx, std::vector<llama_token> & tokens) {
+    llama_kv_cache_clear(ctx);
+    if (llama_decode(ctx, llama_batch_get_one(tokens.data(), tokens.size(), 0, 0))) {
+        fprintf(stderr, "%s : failed to eval\n", __func__);
+        return false;
+    }
+    return true;
+}
+
+static void export_gguf(const std::vector<struct ggml_tensor *> & v_ctrl, const std::string fname, const std::string model_hint) {
+    struct gguf_context * ctx = gguf_init_empty();
+
+    const std::string arch = "controlvector";
+    gguf_set_val_str(ctx, "general.architecture", arch.c_str());
+    gguf_set_val_str(ctx, (arch + ".model_hint").c_str(), model_hint.c_str());
+    gguf_set_val_i32(ctx, (arch + ".layer_count").c_str(), v_ctrl.size());
+
+    for (size_t i = 0; i < v_ctrl.size(); ++i) {
+        gguf_add_tensor(ctx, v_ctrl[i]);
+        print_debug_tensor(v_ctrl[i]);
+        printf("Added tensor: %s\n", v_ctrl[i]->name);
+    }
+
+    printf("%s: writing file...\n", __func__);
+    gguf_write_to_file(ctx, fname.c_str(), false);
+    printf("%s: wrote file '%s'\n", __func__, fname.c_str());
+    gguf_free(ctx);
+}
+
+/**
+ * Load prompt files and completion file.
+ * Then format each pair of prompt + completion to make an entry.
+ */
+static int prepare_entries(gpt_params & params, train_context & ctx_train) {
+    // load prompts
+    std::vector<std::string> positive_prompts = ctrlvec_load_prompt_file(params.cvector_positive_file, true);
+    std::vector<std::string> negative_prompts = ctrlvec_load_prompt_file(params.cvector_negative_file, true);
+    if (positive_prompts.size() != negative_prompts.size()) {
+        fprintf(stderr, "number of positive and negative prompts must be equal\n");
+        return 1;
+    }
+    if (positive_prompts.empty()) {
+        fprintf(stderr, "must provide at least one prompt pair\n");
+        return 1;
+    }
+    ctx_train.positive_entries = positive_prompts;
+    ctx_train.negative_entries = negative_prompts;
+    return 0;
+}
+
+int main(int argc, char ** argv) {
+    gpt_params params;
+
+    if (!gpt_params_parse(argc, argv, params)) {
+        print_usage(argc, argv, params);
+        return 1;
+    }
+
+    if (params.n_pca_iterations % params.n_pca_batch != 0) {
+        fprintf(stderr, "PCA iterations must by multiply of PCA batch size\n");
+        return 1;
+    }
+
+
+    callback_data cb_data;
+
+    // pass the callback to the backend scheduler
+    // it will be executed for each node during the graph computation
+    params.cb_eval = cb_eval;
+    params.cb_eval_user_data = &cb_data;
+    params.warmup = false;
+
+    print_build_info();
+    llama_backend_init();
+    llama_numa_init(params.numa);
+
+    // load the model to get hparams
+    llama_model * model;
+    llama_context * ctx;
+    std::tie(model, ctx) = llama_init_from_gpt_params(params);
+
+    // int n_ctx = llama_n_ctx(ctx);
+    int n_layers = llama_n_layer(model);
+    int n_embd = llama_n_embd(model);
+    // get model hint param (a.k.a model arch name)
+    char model_hint[128];
+    llama_model_meta_val_str(model, "general.architecture", model_hint, 128);
+
+    // init train_context
+    train_context ctx_train(n_embd, n_layers);
+
+    // load and prepare entries for training
+    prepare_entries(params, ctx_train);
+
+    // we have to pretokenize everything because otherwise we don't know how much overhead to allocate ctx_diffs_wrapped
+    std::vector<tokenized_prompt> tokenized_prompts;
+    size_t n_total_tokens = 0;
+    for (size_t i = 0; i < ctx_train.positive_entries.size(); ++i) {
+        tokenized_prompt t(ctx, ctx_train.positive_entries[i], ctx_train.negative_entries[i]);
+        n_total_tokens += 2 * t.max_seq_len;
+        tokenized_prompts.push_back(std::move(t));
+    }
+
+    std::cout << "n_total_tokens: " << n_total_tokens << std::endl;
+
+    for(size_t i = 0; i < ctx_train.positive_entries.size(); ++i) {
+        bool success = false;
+        tokenized_prompt t = tokenized_prompts[i];
+        cb_data.n_layers = n_layers;
+        cb_data.n_tokens = t.max_seq_len;
+
+        printf("Evaluating prompt[%d/%d]: \"%s\" - \"%s\" (%d tokens)\n",
+            (int) i+1, (int) ctx_train.positive_entries.size(),
+            tokens_to_str(ctx, t.tokens_pos.cbegin(), t.tokens_pos.cend()).c_str(),
+            tokens_to_str(ctx, t.tokens_neg.cbegin(), t.tokens_neg.cend()).c_str(),
+            (int) t.max_seq_len);
+
+        cb_data.is_eval_pos = true;
+        success = get_hidden_layers(ctx, t.tokens_pos);
+        if (!success) break;
+
+        cb_data.is_eval_pos = false;
+        success = get_hidden_layers(ctx, t.tokens_neg);
+        if (!success) break;
+
+        // calculate diff and remove all zero rows
+        auto v_diff_filtered = cb_data.calc_diff();
+
+        // save & concat the filtered v_diff to ctx_train
+        ctx_train.concat_diff_tmp(v_diff_filtered);
+
+        // reset for next iteration
+        cb_data.reset();
+    }
+
+    // done with the model, we can now free it to make gain some memory
+    printf("Done evaluate prompts, unload model...\n");
+    llama_free(ctx);
+    llama_free_model(model);
+
+    bool use_pca = params.cvector_dimre_method == DIMRE_METHOD_PCA;
+
+    // prepare ctx_train for PCA
+    ctx_train.build_v_diff(use_pca);
+
+    if (use_pca) {
+        // run PCA
+        PCA::pca_params pca_params;
+        pca_params.n_threads = params.n_threads;
+        pca_params.n_batch = params.n_pca_batch;
+        pca_params.n_iterations = params.n_pca_iterations;
+        PCA::run_pca(pca_params, ctx_train.v_diff, ctx_train.v_final);
+    } else {
+        // run mean
+        mean::run(ctx_train.v_diff, ctx_train.v_final);
+    }
+
+    // write output vectors to gguf
+    export_gguf(ctx_train.v_final, params.cvector_outfile, model_hint);
+
+    llama_backend_free();
+
+    return 0;
+}