Avoid using __fp16 on ARM with old nvcc (#10616)

* Add notes for a static build

* Update docs/build.md

---------

Co-authored-by: Diego Devesa <slarengh@gmail.com>

.clang-tidy

@@ -17,8 +17,10 @@ Checks: >
     -clang-analyzer-security.insecureAPI.DeprecatedOrUnsafeBufferHandling,
     performance-*,
     portability-*,
+    -portability-simd-intrinsics,
     misc-*,
     -misc-const-correctness,
     -misc-non-private-member-variables-in-classes,
     -misc-no-recursion,
+    -misc-use-anonymous-namespace,
 FormatStyle: none

.devops/full-musa.Dockerfile

@@ -6,6 +6,9 @@ ARG BASE_MUSA_DEV_CONTAINER=mthreads/musa:${MUSA_VERSION}-devel-ubuntu${UBUNTU_V
 
 FROM ${BASE_MUSA_DEV_CONTAINER} AS build
 
+# MUSA architecture to build for (defaults to all supported archs)
+ARG MUSA_DOCKER_ARCH=default
+
 RUN apt-get update && \
     apt-get install -y build-essential cmake python3 python3-pip git libcurl4-openssl-dev libgomp1
 
@@ -19,7 +22,11 @@ WORKDIR /app
 
 COPY . .
 
-RUN cmake -B build -DGGML_NATIVE=OFF -DGGML_MUSA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
+# Use the default MUSA archs if not specified
+RUN if [ "${MUSA_DOCKER_ARCH}" != "default" ]; then \
+        export CMAKE_ARGS="-DMUSA_ARCHITECTURES=${MUSA_DOCKER_ARCH}"; \
+    fi && \
+    cmake -B build -DGGML_NATIVE=OFF -DGGML_MUSA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
     cmake --build build --config Release -j$(nproc) && \
     cp build/bin/* .
 

.devops/llama-cli-musa.Dockerfile

@@ -8,6 +8,9 @@ ARG BASE_MUSA_RUN_CONTAINER=mthreads/musa:${MUSA_VERSION}-runtime-ubuntu${UBUNTU
 
 FROM ${BASE_MUSA_DEV_CONTAINER} AS build
 
+# MUSA architecture to build for (defaults to all supported archs)
+ARG MUSA_DOCKER_ARCH=default
+
 RUN apt-get update && \
     apt-get install -y build-essential git cmake
 
@@ -15,7 +18,11 @@ WORKDIR /app
 
 COPY . .
 
-RUN cmake -B build -DGGML_NATIVE=OFF -DGGML_MUSA=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
+# Use the default MUSA archs if not specified
+RUN if [ "${MUSA_DOCKER_ARCH}" != "default" ]; then \
+        export CMAKE_ARGS="-DMUSA_ARCHITECTURES=${MUSA_DOCKER_ARCH}"; \
+    fi && \
+    cmake -B build -DGGML_NATIVE=OFF -DGGML_MUSA=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
     cmake --build build --config Release --target llama-cli -j$(nproc) && \
     mkdir -p /app/lib && \
     find build -name "*.so" -exec cp {} /app/lib \;

.devops/llama-server-musa.Dockerfile

@@ -8,6 +8,9 @@ ARG BASE_MUSA_RUN_CONTAINER=mthreads/musa:${MUSA_VERSION}-runtime-ubuntu${UBUNTU
 
 FROM ${BASE_MUSA_DEV_CONTAINER} AS build
 
+# MUSA architecture to build for (defaults to all supported archs)
+ARG MUSA_DOCKER_ARCH=default
+
 RUN apt-get update && \
     apt-get install -y build-essential git cmake libcurl4-openssl-dev
 
@@ -15,7 +18,11 @@ WORKDIR /app
 
 COPY . .
 
-RUN cmake -B build -DGGML_NATIVE=OFF -DGGML_MUSA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
+# Use the default MUSA archs if not specified
+RUN if [ "${MUSA_DOCKER_ARCH}" != "default" ]; then \
+        export CMAKE_ARGS="-DMUSA_ARCHITECTURES=${MUSA_DOCKER_ARCH}"; \
+    fi && \
+    cmake -B build -DGGML_NATIVE=OFF -DGGML_MUSA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
     cmake --build build --config Release --target llama-server -j$(nproc) && \
     mkdir -p /app/lib && \
     find build -name "*.so" -exec cp {} /app/lib \;

.devops/llama-server.Dockerfile

@@ -3,22 +3,34 @@ 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
+    apt-get install -y build-essential git cmake libcurl4-openssl-dev
 
 WORKDIR /app
 
 COPY . .
 
-ENV LLAMA_CURL=1
 
-RUN make -j$(nproc) llama-server
+RUN \
+    # Build multiple versions of the CPU backend
+    scripts/build-cpu.sh avx         -DGGML_AVX=ON -DGGML_AVX2=OFF && \
+    scripts/build-cpu.sh avx2        -DGGML_AVX=ON -DGGML_AVX2=ON && \
+    scripts/build-cpu.sh avx512      -DGGML_AVX=ON -DGGML_AVX2=ON -DGGML_AVX512=ON && \
+    scripts/build-cpu.sh amx         -DGGML_AVX=ON -DGGML_AVX2=ON -DGGML_AVX512=ON -DGGML_AVX_VNNI=ON -DGGML_AVX512_VNNI=ON -DGGML_AMX_TILE=ON -DGGML_AMX_INT8=ON && \
+    # Build llama-server
+    cmake -S . -B build -DGGML_BACKEND_DL=ON -DGGML_NATIVE=OFF -DLLAMA_CURL=ON -DCMAKE_BUILD_TYPE=Release && \
+    cmake --build build --target llama-server -j $(nproc) && \
+    # Copy the built libraries to /app/lib
+    mkdir -p /app/lib && \
+    mv libggml-cpu* /app/lib/ && \
+    find build -name "*.so" -exec cp {} /app/lib/ \;
 
 FROM ubuntu:$UBUNTU_VERSION AS runtime
 
 RUN apt-get update && \
     apt-get install -y libcurl4-openssl-dev libgomp1 curl
 
-COPY --from=build /app/llama-server /llama-server
+COPY --from=build /app/build/bin/llama-server /llama-server
+COPY --from=build /app/lib/ /
 
 ENV LC_ALL=C.utf8
 # Must be set to 0.0.0.0 so it can listen to requests from host machine

.devops/nix/python-scripts.nix

@@ -34,7 +34,7 @@ let
 
     # server tests
     openai
-    behave
+    pytest
     prometheus-client
   ];
 in

.github/labeler.yml

@@ -3,19 +3,18 @@ Kompute:
     - changed-files:
         - any-glob-to-any-file:
             - ggml/include/ggml-kompute.h
-            - ggml/src/ggml-kompute.cpp
+            - ggml/src/ggml-kompute/**
             - README-kompute.md
 Apple Metal:
     - changed-files:
         - any-glob-to-any-file:
             - ggml/include/ggml-metal.h
-            - ggml/src/ggml-metal.cpp
+            - ggml/src/ggml-metal/**
             - README-metal.md
 SYCL:
     - changed-files:
         - any-glob-to-any-file:
             - ggml/include/ggml-sycl.h
-            - ggml/src/ggml-sycl.cpp
             - ggml/src/ggml-sycl/**
             - docs/backend/SYCL.md
             - examples/sycl/**
@@ -27,8 +26,8 @@ Nvidia GPU:
 Vulkan:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml/ggml_vk_generate_shaders.py
-            - ggml/src/ggml-vulkan*
+            - ggml/include/ggml-vulkan.h
+            - ggml/src/ggml-vulkan/**
 documentation:
     - changed-files:
         - any-glob-to-any-file:
@@ -75,11 +74,7 @@ server:
 ggml:
     - changed-files:
         - any-glob-to-any-file:
-            - ggml/include/ggml*.h
-            - ggml/src/ggml*.c
-            - ggml/src/ggml*.cpp
-            - ggml/src/ggml*.h
-            - ggml-cuda/**
+            - ggml/**
 nix:
     - changed-files:
         - any-glob-to-any-file:

.github/pull_request_template.md

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

.github/workflows/build.yml

@@ -160,66 +160,6 @@ jobs:
           path: llama-${{ steps.tag.outputs.name }}-bin-macos-x64.zip
           name: llama-bin-macos-x64.zip
 
-  ubuntu-focal-make:
-    runs-on: ubuntu-20.04
-    env:
-      LLAMA_NODE_AVAILABLE: true
-      LLAMA_PYTHON_AVAILABLE: true
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v4
-
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential gcc-8
-
-      - uses: actions/setup-node@v4
-        with:
-          node-version: "20"
-
-      - uses: actions/setup-python@v5
-        with:
-          python-version: "3.11"
-
-      - name: Build
-        id: make_build
-        env:
-            LLAMA_FATAL_WARNINGS: 1
-        run: |
-          CC=gcc-8 make -j $(nproc)
-
-      - name: Test
-        id: make_test
-        run: |
-          CC=gcc-8 make tests -j $(nproc)
-          make test -j $(nproc)
-
-  ubuntu-focal-make-curl:
-    runs-on: ubuntu-20.04
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v4
-
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential gcc-8 libcurl4-openssl-dev
-
-      - name: Build
-        id: make_build
-        env:
-          LLAMA_FATAL_WARNINGS: 1
-          LLAMA_CURL: 1
-        run: |
-          CC=gcc-8 make -j $(nproc)
-
   ubuntu-latest-cmake:
     runs-on: ubuntu-latest
 
@@ -517,36 +457,6 @@ jobs:
           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 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:
-    runs-on: macos-latest
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v4
-
-      - name: Dependencies
-        id: depends
-        continue-on-error: true
-        run: |
-          brew update
-
-      - name: Build
-        id: make_build
-        env:
-            LLAMA_FATAL_WARNINGS: 1
-        run: |
-          GGML_NO_METAL=1 make -j $(sysctl -n hw.logicalcpu)
-
-      - name: Test
-        id: make_test
-        run: |
-          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 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
@@ -642,33 +552,35 @@ jobs:
             -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
-
-    strategy:
-      matrix:
-        destination: ['generic/platform=macOS', 'generic/platform=iOS', 'generic/platform=tvOS']
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v4
-
-      - name: Dependencies
-        id: depends
-        continue-on-error: true
-        run: |
-          brew update
-
-      - name: xcodebuild for swift package
-        id: xcodebuild
-        run: |
-          xcodebuild -scheme llama -destination "${{ matrix.destination }}"
-
-      - name: Build Swift Example
-        id: make_build_swift_example
-        run: |
-            make swift
+# TODO: tmp disabled. see for possible re-enable:
+#       https://github.com/ggerganov/llama.cpp/pull/10525
+#  macOS-latest-swift:
+#    runs-on: macos-latest
+#
+#    strategy:
+#      matrix:
+#        destination: ['generic/platform=macOS', 'generic/platform=iOS', 'generic/platform=tvOS']
+#
+#    steps:
+#      - name: Clone
+#        id: checkout
+#        uses: actions/checkout@v4
+#
+#      - name: Dependencies
+#        id: depends
+#        continue-on-error: true
+#        run: |
+#          brew update
+#
+#      - name: xcodebuild for swift package
+#        id: xcodebuild
+#        run: |
+#          xcodebuild -scheme llama -destination "${{ matrix.destination }}"
+#
+#      - name: Build Swift Example
+#        id: make_build_swift_example
+#        run: |
+#            make swift
 
   windows-msys2:
     runs-on: windows-latest
@@ -695,21 +607,6 @@ jobs:
             mingw-w64-${{matrix.env}}-cmake
             mingw-w64-${{matrix.env}}-openblas
 
-      - name: Build using make
-        shell: msys2 {0}
-        run: |
-            make -j $(nproc)
-
-      - name: Clean after building using make
-        shell: msys2 {0}
-        run: |
-            make clean
-
-      - name: Build using make w/ OpenBLAS
-        shell: msys2 {0}
-        run: |
-            make GGML_OPENBLAS=1 -j $(nproc)
-
       - name: Build using CMake
         shell: msys2 {0}
         run: |
@@ -728,7 +625,7 @@ jobs:
             cmake --build build --config ${{ matrix.build }} -j $(nproc)
 
   windows-latest-cmake:
-    runs-on: windows-2019
+    runs-on: windows-latest
 
     env:
       OPENBLAS_VERSION: 0.3.23
@@ -871,12 +768,33 @@ jobs:
           path: llama-${{ steps.tag.outputs.name }}-bin-win-${{ matrix.build }}.zip
           name: llama-bin-win-${{ matrix.build }}.zip
 
-  windows-latest-cmake-cuda:
+  ubuntu-latest-cmake-cuda:
+    runs-on: ubuntu-latest
+    container: nvidia/cuda:12.6.2-devel-ubuntu24.04
+
+    steps:
+        - name: Clone
+          id: checkout
+          uses: actions/checkout@v4
+
+        - name: Install dependencies
+          env:
+            DEBIAN_FRONTEND: noninteractive
+          run: |
+              apt update
+              apt install -y cmake build-essential ninja-build libgomp1 git
+
+        - name: Build with CMake
+          run: |
+            cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release -DGGML_NATIVE=OFF -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=89-real -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined -DLLAMA_FATAL_WARNINGS=ON
+            cmake --build build
+
+  windows-2019-cmake-cuda:
     runs-on: windows-2019
 
     strategy:
       matrix:
-        cuda: ['12.2.0', '11.7.1']
+        cuda: ['12.4', '11.7']
         build: ['cuda']
 
     steps:
@@ -884,24 +802,83 @@ jobs:
         id: checkout
         uses: actions/checkout@v4
         with:
-          fetch-depth: 0
+            fetch-depth: 0
 
-      - name: Install CUDA toolkit
-        id: cuda-toolkit
-        uses: Jimver/cuda-toolkit@v0.2.15
+      - name: Install Cuda Toolkit 11.7
+        if: ${{ matrix.cuda == '11.7' }}
+        run: |
+          mkdir -p "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7"
+          choco install unzip -y
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_cudart/windows-x86_64/cuda_cudart-windows-x86_64-11.7.99-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvcc/windows-x86_64/cuda_nvcc-windows-x86_64-11.7.99-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvrtc/windows-x86_64/cuda_nvrtc-windows-x86_64-11.7.99-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/libcublas/windows-x86_64/libcublas-windows-x86_64-11.7.4.6-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvtx/windows-x86_64/cuda_nvtx-windows-x86_64-11.7.91-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/visual_studio_integration/windows-x86_64/visual_studio_integration-windows-x86_64-11.7.91-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvprof/windows-x86_64/cuda_nvprof-windows-x86_64-11.7.101-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_cccl/windows-x86_64/cuda_cccl-windows-x86_64-11.7.91-archive.zip"
+          unzip '*.zip' -d "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7"
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\cuda_cudart-windows-x86_64-11.7.99-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\cuda_nvcc-windows-x86_64-11.7.99-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\cuda_nvrtc-windows-x86_64-11.7.99-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\libcublas-windows-x86_64-11.7.4.6-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\cuda_nvtx-windows-x86_64-11.7.91-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\visual_studio_integration-windows-x86_64-11.7.91-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\cuda_nvprof-windows-x86_64-11.7.101-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\cuda_cccl-windows-x86_64-11.7.91-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" /E /I /H /Y
+          echo "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+          echo "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\libnvvp" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+          echo "CUDA_PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
+          echo "CUDA_PATH_V11_7=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
+
+      - name: Install Cuda Toolkit 12.4
+        if: ${{ matrix.cuda == '12.4' }}
+        run: |
+          mkdir -p "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4"
+          choco install unzip -y
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_cudart/windows-x86_64/cuda_cudart-windows-x86_64-12.4.127-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvcc/windows-x86_64/cuda_nvcc-windows-x86_64-12.4.131-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvrtc/windows-x86_64/cuda_nvrtc-windows-x86_64-12.4.127-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/libcublas/windows-x86_64/libcublas-windows-x86_64-12.4.5.8-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvtx/windows-x86_64/cuda_nvtx-windows-x86_64-12.4.127-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_profiler_api/windows-x86_64/cuda_profiler_api-windows-x86_64-12.4.127-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/visual_studio_integration/windows-x86_64/visual_studio_integration-windows-x86_64-12.4.127-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvprof/windows-x86_64/cuda_nvprof-windows-x86_64-12.4.127-archive.zip"
+          curl -O "https://developer.download.nvidia.com/compute/cuda/redist/cuda_cccl/windows-x86_64/cuda_cccl-windows-x86_64-12.4.127-archive.zip"
+          unzip '*.zip' -d "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4"
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_cudart-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_nvcc-windows-x86_64-12.4.131-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_nvrtc-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\libcublas-windows-x86_64-12.4.5.8-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_nvtx-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_profiler_api-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\visual_studio_integration-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_nvprof-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          xcopy "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\cuda_cccl-windows-x86_64-12.4.127-archive\*" "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" /E /I /H /Y
+          echo "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+          echo "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4\libnvvp" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+          echo "CUDA_PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
+          echo "CUDA_PATH_V12_4=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.4" | Out-File -FilePath $env:GITHUB_ENV -Append -Encoding utf8
+
+      - name: Install ccache
+        uses: hendrikmuhs/ccache-action@v1.2
         with:
-          cuda: ${{ matrix.cuda }}
-          method: 'network'
-          sub-packages: '["nvcc", "cudart", "cublas", "cublas_dev", "thrust", "visual_studio_integration"]'
+          key: ${{ github.job }}-${{ matrix.cuda }}-${{ matrix.build }}
+
+      - name: Install Ninja
+        id: install_ninja
+        run: |
+          choco install ninja
 
       - name: Build
         id: cmake_build
+        shell: cmd
         run: |
-          mkdir build
-          cd build
-          cmake .. -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_CUDA=ON -DBUILD_SHARED_LIBS=ON -DGGML_RPC=ON
-          cmake --build . --config Release -j $((${env:NUMBER_OF_PROCESSORS} - 1)) -t ggml
-          cmake --build . --config Release -j ${env:NUMBER_OF_PROCESSORS}
+          call "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Auxiliary\Build\vcvars64.bat"
+          cmake -S . -B build -G "Ninja Multi-Config" -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_CUDA=ON -DBUILD_SHARED_LIBS=ON -DGGML_RPC=ON
+          set /A NINJA_JOBS=%NUMBER_OF_PROCESSORS%-1
+          cmake --build build --config Release -j %NINJA_JOBS% -t ggml
+          cmake --build build --config Release
 
       - name: Determine tag name
         id: tag
@@ -930,10 +907,12 @@ jobs:
           name: llama-bin-win-cu${{ matrix.cuda }}-x64.zip
 
       - name: Copy and pack Cuda runtime
+        if: ${{ github.event_name == 'push' && github.ref == 'refs/heads/master' }}
         run: |
-          echo "Cuda install location: ${{steps.cuda-toolkit.outputs.CUDA_PATH}}"
+          echo "Cuda install location: ${{ env.CUDA_PATH }}"
           $dst='.\build\bin\cudart\'
-          robocopy "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
+          robocopy "${{env.CUDA_PATH}}\bin" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
+          robocopy "${{env.CUDA_PATH}}\lib" $dst cudart64_*.dll cublas64_*.dll cublasLt64_*.dll
           7z a cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip $dst\*
 
       - name: Upload Cuda runtime
@@ -984,7 +963,7 @@ jobs:
 
       - name: Build the release package
         id: pack_artifacts
-        if: ${{ ( github.event_name == 'pull_request' && github.base_ref == 'master' ) }}
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
         run: |
           echo "cp oneAPI running time dll files in ${{ env.ONEAPI_ROOT }} to ./build/bin"
 
@@ -1009,7 +988,7 @@ jobs:
           7z a llama-${{ steps.tag.outputs.name }}-bin-win-sycl-x64.zip ./build/bin/*
 
       - name: Upload the release package
-        if: ${{ ( github.event_name == 'pull_request' && github.base_ref == 'master' ) }}
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
         uses: actions/upload-artifact@v4
         with:
           path: llama-${{ steps.tag.outputs.name }}-bin-win-sycl-x64.zip
@@ -1039,6 +1018,11 @@ jobs:
         run: |
           & 'C:\Program Files\AMD\ROCm\*\bin\clang.exe' --version
 
+      - name: Install ccache
+        uses: hendrikmuhs/ccache-action@v1.2
+        with:
+          key: ${{ github.job }}
+
       - name: Build
         id: cmake_build
         run: |
@@ -1059,6 +1043,8 @@ jobs:
       - name: Clone
         id: checkout
         uses: actions/checkout@v4
+        with:
+            fetch-depth: 0
 
       - name: Install
         id: depends
@@ -1168,12 +1154,10 @@ jobs:
     runs-on: ubuntu-latest
 
     needs:
-      - ubuntu-focal-make
       - ubuntu-latest-cmake
-      - macOS-latest-make
       - macOS-latest-cmake
       - windows-latest-cmake
-      - windows-latest-cmake-cuda
+      - windows-2019-cmake-cuda
       - windows-latest-cmake-hip-release
       - macOS-latest-cmake-arm64
       - macOS-latest-cmake-x64

.github/workflows/docker.yml

@@ -114,7 +114,7 @@ jobs:
           swap-storage: true
 
       - name: Build and push Docker image (tagged + versioned)
-        if: github.event_name == 'push'
+        if: ${{ github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' }}
         uses: docker/build-push-action@v6
         with:
           context: .

.github/workflows/nix-ci-aarch64.yml

@@ -1,72 +0,0 @@
-name: Nix aarch64 builds
-
-on:
-  workflow_dispatch: # allows manual triggering
-  schedule:
-    # Rebuild daily rather than on every push because QEMU is expensive (e.g.
-    # 1.5h instead of minutes with the cold cache).
-    #
-    # randint(0, 59), randint(0, 23)
-    - cron: '26 12 * * *'
-  # But also rebuild if we touched any of the Nix expressions:
-  push:
-    branches:
-      - master
-    paths: ['**/*.nix', 'flake.lock']
-  pull_request:
-    types: [opened, synchronize, reopened]
-    paths: ['**/*.nix', 'flake.lock']
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
-  cancel-in-progress: true
-
-# Fine-grant permission
-# https://docs.github.com/en/actions/security-for-github-actions/security-guides/automatic-token-authentication#modifying-the-permissions-for-the-github_token
-permissions:
-  # https://github.com/DeterminateSystems/nix-installer-action?tab=readme-ov-file#with-flakehub
-  id-token: write
-  contents: read
-
-jobs:
-  nix-build-aarch64:
-    runs-on: ubuntu-latest
-    steps:
-    - name: Checkout repository
-      uses: actions/checkout@v4
-    - name: Install QEMU
-      # Copy-paste from https://github.com/orgs/community/discussions/8305#discussioncomment-5888654
-      run: |
-        sudo apt-get update
-        sudo apt-get install -y qemu-user-static qemu-system-aarch64
-        sudo usermod -a -G kvm $USER
-    - name: Install Nix
-      uses: DeterminateSystems/nix-installer-action@v9
-      with:
-        github-token: ${{ secrets.GITHUB_TOKEN }}
-        extra-conf: |
-          extra-platforms = aarch64-linux
-          extra-system-features = nixos-test kvm
-          extra-substituters = https://llama-cpp.cachix.org https://cuda-maintainers.cachix.org
-          extra-trusted-public-keys = llama-cpp.cachix.org-1:H75X+w83wUKTIPSO1KWy9ADUrzThyGs8P5tmAbkWhQc= cuda-maintainers.cachix.org-1:0dq3bujKpuEPMCX6U4WylrUDZ9JyUG0VpVZa7CNfq5E=
-    - uses: DeterminateSystems/magic-nix-cache-action@v2
-      with:
-        upstream-cache: https://${{ matrix.cachixName }}.cachix.org
-    - name: Set-up cachix to push the results to
-      uses: cachix/cachix-action@v13
-      with:
-        authToken: '${{ secrets.CACHIX_AUTH_TOKEN }}'
-        name: llama-cpp
-    - name: Show all output paths
-      run: >
-          nix run github:nix-community/nix-eval-jobs
-          -- --gc-roots-dir gcroot
-          --flake
-          ".#packages.aarch64-linux"
-    - name: Build
-      run: >
-          nix run github:Mic92/nix-fast-build
-          -- --skip-cached --no-nom
-          --systems aarch64-linux
-          --flake
-          ".#checks.aarch64-linux"

.github/workflows/nix-ci.yml

@@ -1,79 +0,0 @@
-name: Nix CI
-
-on:
-  workflow_dispatch: # allows manual triggering
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [opened, synchronize, reopened]
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
-  cancel-in-progress: true
-
-# Fine-grant permission
-# https://docs.github.com/en/actions/security-for-github-actions/security-guides/automatic-token-authentication#modifying-the-permissions-for-the-github_token
-permissions:
-  # https://github.com/DeterminateSystems/nix-installer-action?tab=readme-ov-file#with-flakehub
-  id-token: write
-  contents: read
-
-jobs:
-  nix-eval:
-    strategy:
-      fail-fast: false
-      matrix:
-        os: [ ubuntu-latest, macos-latest ]
-    runs-on: ${{ matrix.os }}
-    steps:
-    - name: Checkout repository
-      uses: actions/checkout@v4
-    - name: Install Nix
-      uses: DeterminateSystems/nix-installer-action@v9
-      with:
-        github-token: ${{ secrets.GITHUB_TOKEN }}
-        extra-conf: |
-          extra-substituters = https://llama-cpp.cachix.org https://cuda-maintainers.cachix.org
-          extra-trusted-public-keys = llama-cpp.cachix.org-1:H75X+w83wUKTIPSO1KWy9ADUrzThyGs8P5tmAbkWhQc= cuda-maintainers.cachix.org-1:0dq3bujKpuEPMCX6U4WylrUDZ9JyUG0VpVZa7CNfq5E=
-    - uses: DeterminateSystems/magic-nix-cache-action@v2
-      with:
-        upstream-cache: https://${{ matrix.cachixName }}.cachix.org
-    - name: List all flake outputs
-      run: nix flake show --all-systems
-    - name: Show all output paths
-      run: >
-          nix run github:nix-community/nix-eval-jobs
-          -- --gc-roots-dir gcroot
-          --flake
-          ".#packages.$(nix eval --raw --impure --expr builtins.currentSystem)"
-  nix-build:
-    strategy:
-      fail-fast: false
-      matrix:
-        os: [ ubuntu-latest, macos-latest ]
-    runs-on: ${{ matrix.os }}
-    steps:
-    - name: Checkout repository
-      uses: actions/checkout@v4
-    - name: Install Nix
-      uses: DeterminateSystems/nix-installer-action@v9
-      with:
-        github-token: ${{ secrets.GITHUB_TOKEN }}
-        extra-conf: |
-          extra-substituters = https://llama-cpp.cachix.org https://cuda-maintainers.cachix.org
-          extra-trusted-public-keys = llama-cpp.cachix.org-1:H75X+w83wUKTIPSO1KWy9ADUrzThyGs8P5tmAbkWhQc= cuda-maintainers.cachix.org-1:0dq3bujKpuEPMCX6U4WylrUDZ9JyUG0VpVZa7CNfq5E=
-    - uses: DeterminateSystems/magic-nix-cache-action@v2
-      with:
-        upstream-cache: https://${{ matrix.cachixName }}.cachix.org
-    - name: Set-up cachix to push the results to
-      uses: cachix/cachix-action@v13
-      with:
-        authToken: '${{ secrets.CACHIX_AUTH_TOKEN }}'
-        name: llama-cpp
-    - name: Build
-      run: >
-          nix run github:Mic92/nix-fast-build
-          -- --skip-cached --no-nom
-          --flake
-          ".#checks.$(nix eval --raw --impure --expr builtins.currentSystem)"

.github/workflows/nix-flake-update.yml

@@ -1,22 +0,0 @@
-name: update-flake-lock
-on:
-  workflow_dispatch:
-  schedule:
-    - cron: '0 0 * * 0' # runs weekly on Sunday at 00:00
-
-jobs:
-  lockfile:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@v4
-      - name: Install Nix
-        uses: DeterminateSystems/nix-installer-action@main
-      - name: Update flake.lock
-        uses: DeterminateSystems/update-flake-lock@main
-        with:
-          pr-title: "nix: update flake.lock"
-          pr-labels: |
-            nix
-          pr-reviewers: philiptaron,SomeoneSerge
-          token: ${{ secrets.FLAKE_TOKEN }}

.github/workflows/nix-publish-flake.yml

@@ -1,36 +0,0 @@
-# Make the flake discoverable on https://flakestry.dev and https://flakehub.com/flakes
-name: "Publish a flake to flakestry & flakehub"
-on:
-    push:
-        tags:
-        - "*"
-    workflow_dispatch:
-        inputs:
-            tag:
-                description: "The existing tag to publish"
-                type: "string"
-                required: true
-jobs:
-    flakestry-publish:
-        runs-on: ubuntu-latest
-        permissions:
-            id-token: "write"
-            contents: "read"
-        steps:
-            - uses: flakestry/flakestry-publish@main
-              with:
-                version: "${{ inputs.tag || github.ref_name }}"
-    flakehub-publish:
-      runs-on: "ubuntu-latest"
-      permissions:
-        id-token: "write"
-        contents: "read"
-      steps:
-        - uses: "actions/checkout@v4"
-          with:
-            ref: "${{ (inputs.tag != null) && format('refs/tags/{0}', inputs.tag) || '' }}"
-        - uses: "DeterminateSystems/nix-installer-action@main"
-        - uses: "DeterminateSystems/flakehub-push@main"
-          with:
-            visibility: "public"
-            tag: "${{ inputs.tag }}"

.github/workflows/python-lint.yml

@@ -1,6 +1,13 @@
 name: flake8 Lint
 
-on: [push, pull_request]
+on:
+  push:
+    branches:
+      - master
+    paths: ['.github/workflows/python-lint.yml', '**/*.py']
+  pull_request:
+    types: [opened, synchronize, reopened]
+    paths: ['.github/workflows/python-lint.yml', '**/*.py']
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}

.github/workflows/server.yml

@@ -76,20 +76,26 @@ jobs:
         run: |
           pip install -r examples/server/tests/requirements.txt
 
-      - name: Verify server deps
-        id: verify_server_deps
+      # Setup nodejs (to be used for verifying bundled index.html)
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 22
+
+      - name: Verify bundled index.html
+        id: verify_server_index_html
         run: |
           git config --global --add safe.directory $(realpath .)
-          cd examples/server
-          git ls-files --others --modified
+          cd examples/server/webui
           git status
-          ./deps.sh
+          npm ci
+          npm run build
           git status
-          not_ignored_files="$(git ls-files --others --modified)"
-          echo "Modified files: ${not_ignored_files}"
-          if [ -n "${not_ignored_files}" ]; then
-            echo "Repository is dirty or server deps are not built as expected"
-            echo "${not_ignored_files}"
+          modified_files="$(git status -s)"
+          echo "Modified files: ${modified_files}"
+          if [ -n "${modified_files}" ]; then
+            echo "Repository is dirty or server/webui is not built as expected"
+            echo "Hint: You may need to follow Web UI build guide in server/README.md"
+            echo "${modified_files}"
             exit 1
           fi
 
@@ -122,14 +128,14 @@ jobs:
         id: server_integration_tests
         run: |
           cd examples/server/tests
-          PORT=8888 ./tests.sh
+          ./tests.sh
 
       - name: Slow tests
         id: server_integration_tests_slow
         if: ${{ (github.event.schedule || github.event.inputs.slow_tests == 'true') && matrix.build_type == 'Release' }}
         run: |
           cd examples/server/tests
-          PORT=8888 ./tests.sh --stop --no-skipped --no-capture --tags slow
+          SLOW_TESTS=1 ./tests.sh
 
 
   server-windows:
@@ -180,11 +186,12 @@ jobs:
         run: |
           cd examples/server/tests
           $env:PYTHONIOENCODING = ":replace"
-          behave.exe --summary --stop --no-capture --exclude 'issues|wrong_usages|passkey' --tags llama.cpp
+          pytest -v -x
 
       - name: Slow tests
         id: server_integration_tests_slow
         if: ${{ (github.event.schedule || github.event.inputs.slow_tests == 'true') && matrix.build_type == 'Release' }}
         run: |
           cd examples/server/tests
-          behave.exe --stop --no-skipped --no-capture --tags slow
+          $env:SLOW_TESTS = "1"
+          pytest -v -x

.gitignore

@@ -104,6 +104,10 @@ examples/server/*.mjs.hpp
 !examples/sycl/*.bat
 !examples/sycl/*.sh
 
+# Server Web UI temporary files
+node_modules
+examples/server/webui/dist
+
 # Python
 
 /.venv

AUTHORS

@@ -1,4 +1,4 @@
-# date: Wed Jun 26 19:36:34 EEST 2024
+# date: Thu Nov 28 20:46:15 EET 2024
 # this file is auto-generated by scripts/gen-authors.sh
 
 0cc4m <picard12@live.de>
@@ -7,6 +7,7 @@
 2f38b454 <dxf@protonmail.com>
 3ooabkhxtn <31479382+3ooabkhxtn@users.noreply.github.com>
 44670 <44670@users.noreply.github.com>
+65a <10104049+65a@users.noreply.github.com>
 AN Long <aisk@users.noreply.github.com>
 AT <manyoso@users.noreply.github.com>
 Aarni Koskela <akx@iki.fi>
@@ -19,20 +20,28 @@ Adithya Balaji <adithya.b94@gmail.com>
 AdithyanI <adithyan.i4internet@gmail.com>
 Adrian <smith.adriane@gmail.com>
 Adrian Hesketh <a-h@users.noreply.github.com>
+Ahmad Tameem <113388789+Tameem-10xE@users.noreply.github.com>
 Ahmet Zeer <ahmed.zeer@std.yildiz.edu.tr>
 AidanBeltonS <87009434+AidanBeltonS@users.noreply.github.com>
+AidanBeltonS <aidan.belton@codeplay.com>
 Aisuko <urakiny@gmail.com>
+Akarshan Biswas <akarshan.biswas@gmail.com>
 Akarshan Biswas <akarshanbiswas@fedoraproject.org>
+Al Mochkin <14274697+amochkin@users.noreply.github.com>
 Albert Jin <albert.jin@gmail.com>
 Alberto <57916483+albbus-stack@users.noreply.github.com>
+Alberto Cabrera Pérez <alberto.cabrera@codeplay.com>
+Alberto Cabrera Pérez <alberto.cabrera@intel.com>
 Alex <awhill19@icloud.com>
 Alex Azarov <alex@azarov.by>
 Alex Azarov <alexander.azarov@mapbox.com>
 Alex Klinkhamer <from.github.com.917@grencez.dev>
 Alex Klinkhamer <git@grencez.dev>
 Alex Nguyen <tiendung@users.noreply.github.com>
+Alex O'Connell <35843486+acon96@users.noreply.github.com>
 Alex Petenchea <alex.petenchea@gmail.com>
 Alex Renda <alexrenda@users.noreply.github.com>
+Alex Tuddenham <61622354+AlexsCode@users.noreply.github.com>
 Alex von Gluck IV <kallisti5@unixzen.com>
 Alexey Parfenov <zxed@alkatrazstudio.net>
 Ali Chraghi <63465728+alichraghi@users.noreply.github.com>
@@ -45,18 +54,25 @@ 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>
+Andreas (Andi) Kunar <andreask@msn.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>
+Andrew Minh Nguyen <40281306+amqdn@users.noreply.github.com>
+Andy Salerno <andysalerno@gmail.com>
 Andy Tai <andy-tai@users.noreply.github.com>
+Anthony Van de Gejuchte <anthonyvdgent@gmail.com>
+Antonis Makropoulos <benuix@gmail.com>
 Arik Poznanski <arikpoz@users.noreply.github.com>
+Armen Kaleshian <kriation@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>
+Asghar Ghorbani <a-ghorbani@users.noreply.github.com>
 Ashish <1856117+ashishdatta@users.noreply.github.com>
 Ashok Gelal <401055+ashokgelal@users.noreply.github.com>
 Ashraful Islam <ashraful.meche@gmail.com>
@@ -76,12 +92,16 @@ 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>
+Bert Wagner <github@bertwagner.com>
 Bingan <70050083+binganao@users.noreply.github.com>
+Bjarke Viksøe <164612031+bviksoe@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>
+Brandon Squizzato <35474886+bsquizz@users.noreply.github.com>
 Brian <mofosyne@gmail.com>
+Brian Cunnie <brian.cunnie@gmail.com>
 Bruce MacDonald <brucewmacdonald@gmail.com>
 Bryan Honof <bryanhonof@gmail.com>
 CJ Pais <cj@cjpais.com>
@@ -90,32 +110,47 @@ Calvin Laurenson <calvin@laurenson.dev>
 Cameron <csteele@steelecameron.com>
 Cameron Kaiser <classilla@users.noreply.github.com>
 Carolinabanana <140120812+Carolinabanana@users.noreply.github.com>
+CarryFun <76023481+CarryFun@users.noreply.github.com>
+Carsten Kragelund Jørgensen <carsten@kragelund.me>
+CarterLi999 <664681047@qq.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>
+Changyeon Kim <cyzero.kim@samsung.com>
 Chao Jiang <jc19chaoj@zoho.com>
+Charles Xu <63788048+chaxu01@users.noreply.github.com>
+Charles Xu <charles.xu@arm.com>
+Chen Xi <xi2.chen@intel.com>
+Chen Xi <xixichen08@foxmail.com>
 Cheng Shao <terrorjack@type.dance>
+Chenguang Li <87689256+noemotiovon@users.noreply.github.com>
 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 Köhnenkamp <cvk5@me.com>
 Christian Zhou-Zheng <59622928+christianazinn@users.noreply.github.com>
 Clark Saben <76020733+csaben@users.noreply.github.com>
 Clint Herron <hanclinto@gmail.com>
+Conrad Kramer <conrad@conradkramer.com>
 CrispStrobe <154636388+CrispStrobe@users.noreply.github.com>
+Csaba Kecskemeti <csaba.kecskemeti@gmail.com>
 Cuong Trinh Manh <nguoithichkhampha@gmail.com>
 DAN™ <dranger003@gmail.com>
 Damian Stewart <d@damianstewart.com>
+Dan Johansson <164997844+eddnjjn@users.noreply.github.com>
+Dan Johansson <dan.johansson@arm.com>
 Dane Madsen <dane_madsen@hotmail.com>
 DaniAndTheWeb <57776841+DaniAndTheWeb@users.noreply.github.com>
 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>
+Daniel Kleine <53251018+d-kleine@users.noreply.github.com>
 Daniele <57776841+daniandtheweb@users.noreply.github.com>
 DannyDaemonic <DannyDaemonic@gmail.com>
 Dat Quoc Nguyen <2412555+datquocnguyen@users.noreply.github.com>
@@ -129,19 +164,28 @@ David Pflug <david@pflug.email>
 David Renshaw <dwrenshaw@gmail.com>
 David Sommers <12738+databyte@users.noreply.github.com>
 David Yang <davidyang6us@gmail.com>
+DavidKorczynski <david@adalogics.com>
 Dawid Potocki <github@dawidpotocki.com>
 Dawid Wysocki <62249621+TortillaZHawaii@users.noreply.github.com>
 Dean <Dean.Sinaean@gmail.com>
 Deins <deinsegle@gmail.com>
+Denis Spasyuk <34203011+dspasyuk@users.noreply.github.com>
+Derrick T. Woolworth <dwoolworth@gmail.com>
 Deven Mistry <31466137+deven367@users.noreply.github.com>
+Dibakar Gope <dibakar.gope@arm.com>
 Didzis Gosko <didzis@users.noreply.github.com>
+Diego Devesa <slarengh@gmail.com>
+Diogo Teles Sant'Anna <diogoteles@google.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>
+Dou Xinpeng <15529241576@163.com>
+Dou Xinpeng <81913537+Dou-Git@users.noreply.github.com>
 Douglas Hanley <thesecretaryofwar@gmail.com>
 Dr. Tom Murphy VII Ph.D <499244+tom7@users.noreply.github.com>
 Ebey Abraham <ebey97@gmail.com>
+Echo Nolan <echo@echonolan.net>
 Ed Lee <edilee@mozilla.com>
 Ed Lepedus <ed.lepedus@googlemail.com>
 Eddie-Wang <wangjinheng1120@163.com>
@@ -151,10 +195,13 @@ Elbios <141279586+Elbios@users.noreply.github.com>
 Elton Kola <eltonkola@gmail.com>
 Engininja2 <139037756+Engininja2@users.noreply.github.com>
 Equim <sayaka@ekyu.moe>
+Eric Curtin <ecurtin@redhat.com>
+Eric Curtin <ericcurtin17@gmail.com>
 Eric Sommerlade <es0m@users.noreply.github.com>
 Eric Zhang <34133756+EZForever@users.noreply.github.com>
 Erik Garrison <erik.garrison@gmail.com>
 Erik Scholz <Green-Sky@users.noreply.github.com>
+Esko Toivonen <eskot98@gmail.com>
 Ettore Di Giacinto <mudler@users.noreply.github.com>
 Evan Jones <evan.q.jones@gmail.com>
 Evan Miller <emmiller@gmail.com>
@@ -166,19 +213,26 @@ FK <sozforex@gmail.com>
 Fabian <cmdrf@users.noreply.github.com>
 Fabio R. Sluzala <Fabio3rs@users.noreply.github.com>
 Faez Shakil <faez.shakil@gmail.com>
+Faisal Zaghloul <faisal.zaghloul@gmail.com>
+Faisal Zaghloul <quic_fzaghlou@quicinc.com>
+Fan Shupei <dymarkfan@outlook.com>
 FantasyGmm <16450052+FantasyGmm@users.noreply.github.com>
+Farbod Bijary <110523279+farbodbj@users.noreply.github.com>
 Fattire <528174+fat-tire@users.noreply.github.com>
 Felix <stenbackfelix@gmail.com>
 Finn Voorhees <finnvoorhees@gmail.com>
 Firat <firatkiral@gmail.com>
+FirstTimeEZ <179362031+FirstTimeEZ@users.noreply.github.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>
+Frankie Robertson <frankier@users.noreply.github.com>
 Fred Douglas <43351173+fredlas@users.noreply.github.com>
 Frederik Vogel <Schaltfehler@users.noreply.github.com>
 Gabe Goodhart <gabe.l.hart@gmail.com>
+Gabe Goodhart <ghart@us.ibm.com>
 GainLee <perfecter.gen@gmail.com>
 Galunid <karolek1231456@gmail.com>
 Gary Linscott <glinscott@gmail.com>
@@ -187,11 +241,13 @@ Gavin Zhao <gavinzhaojw@protonmail.com>
 Genkagaku.GPT <hlhr202@163.com>
 Georgi Gerganov <ggerganov@gmail.com>
 Gilad S <giladgd@users.noreply.github.com>
+Gilad S. <7817232+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>
+Guoliang Hua <32868157+nbcsm@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>
@@ -213,11 +269,14 @@ 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>
+Huang Qi <huangqi3@xiaomi.com>
 Huawei Lin <huaweilin.cs@gmail.com>
 Hugo Roussel <hugo.rous@gmail.com>
+Huifeng Ou <79071290+ho2103@users.noreply.github.com>
 Ian Bull <irbull@eclipsesource.com>
 Ian Bull <irbull@gmail.com>
 Ian Scrivener <github@zilogy.asia>
+Icecream95 <the.real.icecream95@gmail.com>
 Ido S <ido.pluto@gmail.com>
 IgnacioFDM <ignaciofdm@gmail.com>
 Igor Okulist <okigan@gmail.com>
@@ -226,11 +285,15 @@ Ilya Kurdyukov <59548320+ilyakurdyukov@users.noreply.github.com>
 Ionoclast Laboratories <brigham@ionoclast.com>
 Isaac McFadyen <isaac@imcf.me>
 IsaacDynamo <61521674+IsaacDynamo@users.noreply.github.com>
+Ivan <nekotekina@gmail.com>
+Ivan Filipov <159561759+vanaka11@users.noreply.github.com>
 Ivan Komarov <Ivan.Komarov@dfyz.info>
 Ivan Stepanov <ivanstepanovftw@gmail.com>
 JH23X <165871467+JH23X@users.noreply.github.com>
+Jack Mousseau <jack@software.inc>
 Jack Mousseau <jmousseau@users.noreply.github.com>
 JackJollimore <130917767+JackJollimore@users.noreply.github.com>
+Jaeden Amero <jaeden@patater.com>
 Jaemin Son <woalsdnd@gmail.com>
 Jag Chadha <jagtesh@gmail.com>
 Jakub N <jakubniemczyk97@gmail.com>
@@ -243,10 +306,14 @@ Jannis Schönleber <joennlae@gmail.com>
 Jared Van Bortel <cebtenzzre@gmail.com>
 Jared Van Bortel <jared@nomic.ai>
 Jason McCartney <jmac@theroot.org>
+Jason Stillerman <jason.t.stillerman@gmail.com>
 Jean-Christophe Hoelt <hoelt@fovea.cc>
 Jean-Michaël Celerier <jeanmichael.celerier+github@gmail.com>
 Jed Fox <git@jedfox.com>
+Jeff Bolz <jbolz@nvidia.com>
+Jeffrey Morgan <jmorganca@gmail.com>
 Jeffrey Quesnelle <emozilla@nousresearch.com>
+Jeroen Mostert <jeroen.mostert@cm.com>
 Jesse Jojo Johnson <williamsaintgeorge@gmail.com>
 Jeximo <jeximo@gmail.com>
 Jhen-Jie Hong <iainst0409@gmail.com>
@@ -258,6 +325,9 @@ 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>
+João Dinis Ferreira <hello@joaof.eu>
+Joe Eli McIlvain <joe.eli.mac@gmail.com>
+Joe Todd <joe.todd@codeplay.com>
 Johan <JohanAR@users.noreply.github.com>
 Johannes Gäßler <johannesg@5d6.de>
 Johannes Rudolph <johannes.rudolph@gmail.com>
@@ -274,7 +344,9 @@ 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 Hee Yoo <contact.jhyoo@gmail.com>
 Jun Jie <71215065+junnjiee16@users.noreply.github.com>
+Junil Kim <logyourself@gmail.com>
 Junyang Lin <justinlin930319@hotmail.com>
 Juraj Bednar <juraj@bednar.io>
 Justin Parker <jparkerweb@gmail.com>
@@ -292,12 +364,14 @@ Karthik Sethuraman <k.seth1993@gmail.com>
 Kasumi <90275229+kasumi-1@users.noreply.github.com>
 Kawrakow <48489457+ikawrakow@users.noreply.github.com>
 Keiichi Tabata <keiichi.tabata@outlook.com>
+Keke Han <hankeke303@163.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>
+Kevin Wang <kevmo314@gmail.com>
 Kolen Cheung <ickc@users.noreply.github.com>
 Konstantin Herud <konstantin.herud@denkbares.com>
 Konstantin Zhuravlyov <konstantin.zhuravlyov@amd.com>
@@ -315,22 +389,29 @@ LeonEricsson <70749762+LeonEricsson@users.noreply.github.com>
 Leonardo Neumann <leonardo@neumann.dev.br>
 Li Tan <tanliboy@gmail.com>
 Linwei Wang <wanix1988@gmail.com>
+Liu Jia <109258120+Septa2112@users.noreply.github.com>
+Liu Jia <jia3.liu@intel.com>
 LoganDark <github@logandark.mozmail.com>
+Loïc Carrère <loic.carrere@gmail.com>
 LostRuins <39025047+LostRuins@users.noreply.github.com>
 Luciano <lucianostrika44@gmail.com>
 Luo Tian <lt@basecity.com>
 Lyle Dean <dean@lyle.dev>
+M-A <maruel@gmail.com>
 M. Yusuf Sarıgöz <yusufsarigoz@gmail.com>
+Ma Mingfei <mingfei.ma@intel.com>
 Maarten ter Huurne <maarten@treewalker.org>
 Mack Straight <eiz@users.noreply.github.com>
 Maël Kerbiriou <m431.kerbiriou@gmail.com>
 MaggotHATE <clay1326@gmail.com>
+Mahesh Madhav <67384846+heshpdx@users.noreply.github.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>
+Mark Zhuang <zhuangqiubin@gmail.com>
 Marko Tasic <mtasic85@gmail.com>
 Markus Tavenrath <mtavenrath@users.noreply.github.com>
 Martin Delille <martin@delille.org>
@@ -342,11 +423,15 @@ 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>
+Mathieu Geli <mathieu.geli@gmail.com>
 Mathieu Nayrolles <MathieuNls@users.noreply.github.com>
+Mathijs Henquet <mathijs.henquet@gmail.com>
 Mathijs de Bruin <mathijs@mathijsfietst.nl>
 Matt Clayton <156335168+mattjcly@users.noreply.github.com>
 Matt Pulver <matt.pulver@heavy.ai>
+Matt Stephenson <mstephenson6@users.noreply.github.com>
 Matteo Boschini <12133566+mbosc@users.noreply.github.com>
+Matteo Mortari <matteo.mortari@gmail.com>
 Mattheus Chediak <shammcity00@gmail.com>
 Matthew Tejo <matthew.tejo@gmail.com>
 Matvey Soloviev <blackhole89@gmail.com>
@@ -356,8 +441,10 @@ Maxime <672982+maximegmd@users.noreply.github.com>
 Maximilian Winter <maximilian.winter.91@gmail.com>
 Meng Zhang <meng@tabbyml.com>
 Meng, Hengyu <hengyu.meng@intel.com>
+Mengqing Cao <cmq0113@163.com>
 Merrick Christensen <merrick.christensen@gmail.com>
 Michael Coppola <m18coppola@gmail.com>
+Michael Francis <edude03@gmail.com>
 Michael Hueschen <m@mhueschen.dev>
 Michael Kesper <mkesper@schokokeks.org>
 Michael Klimenko <mklimenko29@gmail.com>
@@ -365,41 +452,57 @@ Michael Podvitskiy <podvitskiymichael@gmail.com>
 Michael Potter <NanoTekGuy@Gmail.com>
 Michael de Gans <michael.john.degans@gmail.com>
 Michaël de Vries <vriesdemichael@gmail.com>
+Michał Tuszyński <srgtuszy@gmail.com>
 Mihai <mihai.chirculescu@yahoo.com>
 Mike <ytianhui2004@gmail.com>
 Mikko Juola <mikjuo@gmail.com>
 Minsoo Cheong <54794500+mscheong01@users.noreply.github.com>
+Minsoo Cheong <icycle0409@snu.ac.kr>
 Mirko185 <mirkosig@gmail.com>
 Mirror Azure <54669636+MirrorAzure@users.noreply.github.com>
+MistApproach <98988043+MistApproach@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>
+Molly Sophia <mollysophia379@gmail.com>
+MorganRO8 <47795945+MorganRO8@users.noreply.github.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>
+Natsu <chino@hotococoa.moe>
 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>
+Nexes the Old <124105151+Nexesenex@users.noreply.github.com>
 Nexesenex <124105151+Nexesenex@users.noreply.github.com>
 Niall Coates <1349685+Niall-@users.noreply.github.com>
+Nicholai Tukanov <nicholaitukanov@gmail.com>
+Nico Bosshard <nico@bosshome.ch>
 Nicolai Weitkemper <kontakt@nicolaiweitkemper.de>
 Nicolás Pérez <nicolas_perez@brown.edu>
 Nigel Bosch <pnigelb@gmail.com>
 Niklas Korz <niklas@niklaskorz.de>
+NikolaiLyssogor <59844691+NikolaiLyssogor@users.noreply.github.com>
 Nikolas <127742645+nneubacher@users.noreply.github.com>
 Nindaleth <Nindaleth@users.noreply.github.com>
+OSecret <135510162+OLSecret@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>
+PAB <pierreantoine.bannier@gmail.com>
+Pablo Duboue <pablo.duboue@gmail.com>
+Pascal Patry <ppatry@mtacitlabs.com>
 Patrice Ferlet <metal3d@gmail.com>
 Paul Tsochantaris <ptsochantaris@icloud.com>
+Pavel Zloi <github.com@drteam.rocks>
 Pavol Rusnak <pavol@rusnak.io>
+Paweł Wodnicki <151604+32bitmicro@users.noreply.github.com>
 Pedro Cuenca <pedro@huggingface.co>
 Peter Sugihara <peter@campsh.com>
 Phil H <5756783+phiharri@users.noreply.github.com>
@@ -407,10 +510,15 @@ Philip Taron <philip.taron@gmail.com>
 Phillip Kravtsov <phillip@kravtsov.net>
 Pierre Alexandre SCHEMBRI <pa.schembri@gmail.com>
 Pierrick Hymbert <pierrick.hymbert@gmail.com>
+Pieter Ouwerkerk <pieter.ouwerkerk@gmail.com>
+Plamen Minev <pacominev@gmail.com>
+Prashant Vithule <119530321+Vithulep@users.noreply.github.com>
 Przemysław Pawełczyk <przemoc@gmail.com>
 Qin Yue Chen <71813199+chenqiny@users.noreply.github.com>
 Qingyou Meng <meng.qingyou@gmail.com>
 Qu Zongfu <43257352+yancaoweidaode@users.noreply.github.com>
+R0CKSTAR <xiaodong.ye@mthreads.com>
+R0CKSTAR <yeahdongcn@gmail.com>
 RJ Adriaansen <adriaansen@eshcc.eur.nl>
 Radoslav Gerganov <rgerganov@gmail.com>
 Radosław Gryta <radek.gryta@gmail.com>
@@ -419,11 +527,13 @@ Raj Hammeer Singh Hada <hammeerraj@gmail.com>
 Ralph Soika <ralph.soika@imixs.com>
 Rand Xie <randxiexyy29@gmail.com>
 Randall Fitzgerald <randall@dasaku.net>
+Random Fly <renfei8@live.cn>
 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>
+Rich Dougherty <rich@rd.nz>
 Richard Kiss <him@richardkiss.com>
 Richard Roberson <richardr1126@gmail.com>
 Rick G <26732651+TheFlipbook@users.noreply.github.com>
@@ -439,21 +549,30 @@ Robey Holderith <robey@flaminglunchbox.net>
 Robyn <robyngraf@users.noreply.github.com>
 Roger Meier <r.meier@siemens.com>
 Roland <14355895+rbur0425@users.noreply.github.com>
+Romain Biessy <romain.biessy@codeplay.com>
 Romain D <90720+Artefact2@users.noreply.github.com>
 Romain Neutron <romain@neutron.io>
 Roman Parykin <donderom@gmail.com>
 Ron Evans <ron@hybridgroup.com>
 Ron Jailall <rojailal@gmail.com>
+Roni <sulpher@gmx.net>
 Ronny Brendel <ronnybrendel@gmail.com>
 Ronsor <ronsor@ronsor.pw>
 Rowan Hart <rowanbhart@gmail.com>
+Ruchira Hasaranga <ruchira66@gmail.com>
+Ruixin Huang <18860020911@163.com>
 Rune <43761327+Rune-AI@users.noreply.github.com>
+RunningLeon <maningsheng@sensetime.com>
+RunningLeon <mnsheng@yeah.net>
 Ryan Landay <rlanday@gmail.com>
 Ryder Wishart <ryderwishart@gmail.com>
 Ryuei <louixs@users.noreply.github.com>
 Rőczey Barnabás <31726601+An0nie@users.noreply.github.com>
+SRHMorris <69468379+SRHMorris@users.noreply.github.com>
+SXX <sxx1136965276@gmail.com>
 SakuraUmi <yukinon244@gmail.com>
 Salvador E. Tropea <stropea@inti.gob.ar>
+Salvatore Mesoraca <s.mesoraca16@gmail.com>
 Sam Spilsbury <smspillaz@gmail.com>
 Sami Farin <3876865+Safari77@users.noreply.github.com>
 Samuel Maynard <samwmaynard@gmail.com>
@@ -463,23 +582,29 @@ Sebastián A <sebastian.aedo29@gmail.com>
 SebastianApel <13675545+SebastianApel@users.noreply.github.com>
 Senemu <10880819+Senemu@users.noreply.github.com>
 Sergey Alirzaev <zl29ah@gmail.com>
+Sergio López <slp@redhat.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>
+Shane A <shanea@allenai.org>
 Shangning Xu <32517059+xushangning@users.noreply.github.com>
+Shankar <gshankar.87@gmail.com>
+Shanshan Shen <467638484@qq.com>
 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>
+Shupei Fan <dymarkfan@outlook.com>
 Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
 Simon Willison <swillison@gmail.com>
 Siwen Yu <yusiwen@gmail.com>
 Sky Yan <skyan83@gmail.com>
 Slaren <2141330+slaren@users.noreply.github.com>
 Slava Primenko <primenko.s@gmail.com>
+Small Grass Forest <zixuanxcl@gmail.com>
 SoftwareRenderer <138734813+SoftwareRenderer@users.noreply.github.com>
 Someone <sergei.kozlukov@aalto.fi>
 Someone Serge <sergei.kozlukov@aalto.fi>
@@ -491,12 +616,15 @@ Stefan Sydow <stefan@sydow.email>
 Steffen Röcker <sroecker@gmail.com>
 Stephan Walter <stephan@walter.name>
 Stephen Nichols <snichols@users.noreply.github.com>
+Steve Bonds <sbonds@gmail.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>
+StrangeBytesDev <141275258+StrangeBytesDev@users.noreply.github.com>
 Suaj Carrot <72162667+SuajCarrot@users.noreply.github.com>
 SuperUserNameMan <yoann@terminajones.com>
+Sutou Kouhei <kou@cozmixng.org>
 Tai Duc Nguyen <taiducnguyen.drexel@gmail.com>
 Taikono-Himazin <kazu@po.harenet.ne.jp>
 Tameem <113388789+AhmadTameem@users.noreply.github.com>
@@ -507,7 +635,9 @@ Theia Vogel <theia@vgel.me>
 Thérence <13496987+Royalphax@users.noreply.github.com>
 Thibault Terrasson <thibault.terrasson@gmail.com>
 Thomas Klausner <wiz@gatalith.at>
+Thorsten Sommer <SommerEngineering@users.noreply.github.com>
 Tim Miller <drasticactions@users.noreply.github.com>
+Tim Wang <overocean@gmail.com>
 Timmy Knight <r2d2fish@gmail.com>
 Timothy Cronin <40186632+4imothy@users.noreply.github.com>
 Ting Lou <ting.lou@gmail.com>
@@ -517,24 +647,31 @@ 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>
+Tony Wasserka <4840017+neobrain@users.noreply.github.com>
 Tristan Druyen <tristan@vault81.mozmail.com>
 Tristan Ross <rosscomputerguy@protonmail.com>
+Trivikram Kamat <16024985+trivikr@users.noreply.github.com>
 Tungsten842 <886724vf@anonaddy.me>
 Tungsten842 <quantmint@protonmail.com>
 Tushar <ditsuke@protonmail.com>
 UEXTM.com <84163508+uextm@users.noreply.github.com>
+Ujjawal Panchal <31011628+Ujjawal-K-Panchal@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>
+Vali Malinoiu <0x4139@gmail.com>
 Victor Nogueira <felladrin@gmail.com>
 Victor Z. Peng <ziliangdotme@gmail.com>
+Viet-Anh NGUYEN (Andrew) <vietanh.dev@gmail.com>
+Vinesh Janarthanan <36610342+VJHack@users.noreply.github.com>
 Vlad <spitfireage@gmail.com>
 Vladimir <bogdad@gmail.com>
 Vladimir Malyutin <first-leon@yandex.ru>
 Vladimir Zorin <vladimir@deviant.guru>
+VoidIsVoid <343750470@qq.com>
 Volodymyr Vitvitskyi <72226+signalpillar@users.noreply.github.com>
 WangHaoranRobin <56047610+WangHaoranRobin@users.noreply.github.com>
 Weird Constructor <weirdconstructor@gmail.com>
@@ -551,15 +688,22 @@ Xiang (Kevin) Li <kevinli020508@gmail.com>
 Xiao-Yong Jin <jinxiaoyong@gmail.com>
 XiaotaoChen <chenxiaotao1234@gmail.com>
 Xiaoyi Chen <cxychina@gmail.com>
+Xie Yanbo <xieyanbo@gmail.com>
 Xingchen Song(宋星辰) <xingchensong1996@163.com>
+Xinpeng Dou <81913537+Dou-Git@users.noreply.github.com>
 Xuan Son Nguyen <thichthat@gmail.com>
+Yaiko <elyaiko@hotmail.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>
+Yoshi Suhara <y.suhara@gmail.com>
+Yoshi Suhara <ysuhara@nvidia.com>
+Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
 Yueh-Po Peng <94939112+y10ab1@users.noreply.github.com>
 Yui <dev@sleepyyui.com>
+Yuri Khrustalev <ykhrustalev@users.noreply.github.com>
 Yusuf Kağan Hanoğlu <hanoglu@yahoo.com>
 Yuval Peled <31162840+Yuval-Peled@users.noreply.github.com>
 ZHAOKAI WANG <sanxianwei@163.com>
@@ -568,6 +712,8 @@ 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>
+Zhenwei Jin <109658203+kylo5aby@users.noreply.github.com>
+Zhiyuan Li <lizhiyuan@uniartisan.com>
 ZhouYuChen <zhouyuchen@naver.com>
 Ziad Ben Hadj-Alouane <zied.benhadjalouane@gmail.com>
 Ziang Wu <97337387+ZiangWu-77@users.noreply.github.com>
@@ -581,6 +727,7 @@ alexpinel <93524949+alexpinel@users.noreply.github.com>
 alonfaraj <alonfaraj@gmail.com>
 alwqx <kenan3015@gmail.com>
 amd-lalithnc <lalithnc@amd.com>
+amritahs-ibm <amritahs@linux.vnet.ibm.com>
 andrijdavid <david@geek.mg>
 anon998 <131767832+anon998@users.noreply.github.com>
 anzz1 <anzz1@live.com>
@@ -588,14 +735,18 @@ apaz <aarpazdera@gmail.com>
 apcameron <37645737+apcameron@users.noreply.github.com>
 arch-btw <57669023+arch-btw@users.noreply.github.com>
 arcrank <arcrank@gmail.com>
+ardfork <134447697+ardfork@users.noreply.github.com>
 arlo-phoenix <140345165+arlo-phoenix@users.noreply.github.com>
 at8u <129688334+at8u@users.noreply.github.com>
 automaticcat <daogiatuank54@gmail.com>
+awatuna <23447591+awatuna@users.noreply.github.com>
+b4b4o <zwbao@foxmail.com>
 bandoti <141645996+bandoti@users.noreply.github.com>
 beiller <beiller@gmail.com>
 bhubbb <79117352+bhubbb@users.noreply.github.com>
 bmwl <brian.marshall@tolko.com>
 bobqianic <129547291+bobqianic@users.noreply.github.com>
+brucepro <git@brucepro.net>
 bryanSwk <93190252+bryanSwk@users.noreply.github.com>
 bsilvereagle <bsilvereagle@users.noreply.github.com>
 bssrdf <merlintiger@hotmail.com>
@@ -614,10 +765,14 @@ cpumaxx <163466046+cpumaxx@users.noreply.github.com>
 crasm <crasm@git.vczf.net>
 crasm <crasm@git.vczf.us>
 daboe01 <daboe01@googlemail.com>
+daghanerdonmez <44506702+daghanerdonmez@users.noreply.github.com>
+daminho <37615795+daminho@users.noreply.github.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>
+devojony <61173062+devojony@users.noreply.github.com>
+ditsuke <ditsuke@protonmail.com>
 divinity76 <divinity76@gmail.com>
 dm4 <sunrisedm4@gmail.com>
 dotpy314 <33351922+dotpy314@users.noreply.github.com>
@@ -629,14 +784,18 @@ ebraminio <ebraminio@gmail.com>
 eiery <19350831+eiery@users.noreply.github.com>
 eric8607242 <e0928021388@gmail.com>
 fairydreaming <166155368+fairydreaming@users.noreply.github.com>
+fengerhu1 <2748250768@qq.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>
 goerch <jhr.walter@t-online.de>
 grahameth <96447521+grahameth@users.noreply.github.com>
+gtygo <gtydoit@gmail.com>
 gwjr <502526+gwjr@users.noreply.github.com>
 h-h-h-h <13482553+h-h-h-h@users.noreply.github.com>
 hankcs <cnhankmc@gmail.com>
+haopeng <657407891@qq.com>
+hipudding <huafengchun@gmail.com>
 hoangmit <hoangmit@users.noreply.github.com>
 hongbo.mo <352280764@qq.com>
 hopkins385 <98618192+hopkins385@users.noreply.github.com>
@@ -649,12 +808,14 @@ hxer7963 <hxer7963@gmail.com>
 hydai <z54981220@gmail.com>
 iSma <ismail.senhaji@gmail.com>
 iacore <74560659+iacore@users.noreply.github.com>
+icppWorld <124377669+icppWorld@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>
+jdomke <28772296+jdomke@users.noreply.github.com>
 jiez <373447296@qq.com>
 jneem <joeneeman@gmail.com>
 joecryptotoo <80373433+joecryptotoo@users.noreply.github.com>
@@ -677,28 +838,35 @@ klosax <131523366+klosax@users.noreply.github.com>
 kunal-vaishnavi <115581922+kunal-vaishnavi@users.noreply.github.com>
 kunnis <kunnis@users.noreply.github.com>
 kuronekosaiko <EvanChanJ@163.com>
+kustaaya <58045274+kustaaya@users.noreply.github.com>
 kuvaus <22169537+kuvaus@users.noreply.github.com>
 kwin1412 <42286931+kwin1412@users.noreply.github.com>
 l3utterfly <gc.pthzfoldr@gmail.com>
+laik <laik.lj@me.com>
 ldwang <ftgreat@163.com>
 le.chang <cljs118@126.com>
 leejet <leejet714@gmail.com>
+leo-pony <nengjunma@outlook.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>
+ltoniazzi <61414566+ltoniazzi@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>
+matiaslin <45382001+matiaslin@users.noreply.github.com>
+matteo <matteogeniaccio@yahoo.it>
 mdrokz <mohammadmunshi@gmail.com>
 mgroeber9110 <45620825+mgroeber9110@users.noreply.github.com>
 minarchist <minarchist@users.noreply.github.com>
 mj-shifu <77107165+mj-shifu@users.noreply.github.com>
 mmyjona <jonathan.gonse@gmail.com>
 momonga <115213907+mmnga@users.noreply.github.com>
+momonga <146910567+mmngays@users.noreply.github.com>
 moritzbrantner <31051084+moritzbrantner@users.noreply.github.com>
 mzcu <milos.cubrilo@gmail.com>
 nanahi <130121847+na-na-hi@users.noreply.github.com>
@@ -716,8 +884,10 @@ 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>
+pculliton <phillipculliton@gmail.com>
 pengxin99 <pengxin.yuan@intel.com>
 perserk <perserk@gmail.com>
+piDack <104877312+piDack@users.noreply.github.com>
 pmysl <piotr.myslinski@outlook.com>
 postmasters <namnguyen@google.com>
 pudepiedj <pudepiedj@gmail.com>
@@ -733,6 +903,7 @@ runfuture <runfuture@users.noreply.github.com>
 sandyiscool <sandyiscool@gmail.com>
 sasha0552 <admin@sasha0552.org>
 semidark <me@semidark.net>
+serhii-nakon <57632032+serhii-nakon@users.noreply.github.com>
 sharpHL <132747147+sharpHL@users.noreply.github.com>
 shibe2 <shibe@tuta.io>
 singularity <12184989+singularity-s0@users.noreply.github.com>
@@ -741,42 +912,55 @@ 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>
+standby24x7 <standby24x7@gmail.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>
+tc-mb <157115220+tc-mb@users.noreply.github.com>
 texmex76 <40733439+texmex76@users.noreply.github.com>
 thement <40525767+thement@users.noreply.github.com>
+thewh1teagle <61390950+thewh1teagle@users.noreply.github.com>
 tjohnman <tjohnman@users.noreply.github.com>
+toyer <2042519524@qq.com>
 tslmy <tslmy@users.noreply.github.com>
 ubik2 <ubik2@users.noreply.github.com>
 uint256_t <konndennsa@gmail.com>
 uint256_t <maekawatoshiki1017@gmail.com>
 unbounded <haakon@likedan.net>
+uvos <devnull@uvos.xyz>
 valiray <133289098+valiray@users.noreply.github.com>
+vb <vaibhavs10@gmail.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>
+wangshuai09 <391746016@qq.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>
+wwoodsTM <104587230+wwoodsTM@users.noreply.github.com>
 wzy <32936898+Freed-Wu@users.noreply.github.com>
 xaedes <xaedes@gmail.com>
 xaedes <xaedes@googlemail.com>
+xctan <axunlei@gmail.com>
 xloem <0xloem@gmail.com>
 yangli2 <yangli2@gmail.com>
 yuiseki <yuiseki@gmail.com>
+yuri@FreeBSD <yurivict@users.noreply.github.com>
 zakkor <edward.partenie@gmail.com>
 zhangkaihuo <zhangkaihuo@gmail.com>
+zhentaoyu <zhentao.yu@intel.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>
+杨朱 · Kiki <baofa.fan@daocloud.io>
 源文雨 <41315874+fumiama@users.noreply.github.com>
+蕭澧邦 <45505768+shou692199@users.noreply.github.com>
 Нияз Гарифзянов <112617865+garrnizon@users.noreply.github.com>

CMakeLists.txt

@@ -82,6 +82,7 @@ option(LLAMA_CURL "llama: use libcurl to download model from an URL" OFF)
 
 # Required for relocatable CMake package
 include(${CMAKE_CURRENT_SOURCE_DIR}/cmake/build-info.cmake)
+include(${CMAKE_CURRENT_SOURCE_DIR}/cmake/common.cmake)
 
 # override ggml options
 set(GGML_SANITIZE_THREAD    ${LLAMA_SANITIZE_THREAD})
@@ -95,10 +96,6 @@ if (NOT DEFINED GGML_LLAMAFILE)
     set(GGML_LLAMAFILE_DEFAULT ON)
 endif()
 
-if (NOT DEFINED GGML_AMX)
-    set(GGML_AMX ON)
-endif()
-
 if (NOT DEFINED GGML_CUDA_GRAPHS)
     set(GGML_CUDA_GRAPHS_DEFAULT ON)
 endif()

CODEOWNERS

@@ -0,0 +1,3 @@
+# collaborators can optionally add themselves here to indicate their availability for reviewing related PRs
+
+ci/ @ggerganov

CONTRIBUTING.md

@@ -1,9 +1,10 @@
 # Pull requests (for contributors)
 
 - 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
-- Optionally 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
+  - Verify that the perplexity and the performance are not affected negatively by your changes (use `llama-perplexity` and `llama-bench`)
+  - If you modified the `ggml` source, run the `test-backend-ops` tool to check whether different backend implementations of the `ggml` operators produce consistent results (this requires access to at least two different `ggml` backends)
+  - If you modified a `ggml` operator or added a new one, add the corresponding test cases to `test-backend-ops`
 - Consider allowing write access to your branch for faster reviews, as reviewers can push commits directly
 - If your PR becomes stale, don't hesitate to ping the maintainers in the comments
 
@@ -12,6 +13,7 @@
 - Squash-merge PRs
 - Use the following format for the squashed commit title: `<module> : <commit title> (#<issue_number>)`. For example: `utils : fix typo in utils.py (#1234)`
 - Optionally pick a `<module>` from here: https://github.com/ggerganov/llama.cpp/wiki/Modules
+- Consider adding yourself to [CODEOWNERS](CODEOWNERS)
 
 # Coding guidelines
 

Makefile

@@ -1,3 +1,7 @@
+ifndef LLAMA_MAKEFILE
+$(error The Makefile build is deprecated. Use the CMake build instead. For more details, see https://github.com/ggerganov/llama.cpp/blob/master/docs/build.md)
+endif
+
 # Define the default target now so that it is always the first target
 BUILD_TARGETS = \
 	libllava.a \
@@ -251,11 +255,11 @@ endif
 # Compile flags
 #
 
-# keep standard at C11 and C++11
+# keep standard at C11 and C++17
 MK_CPPFLAGS  = -Iggml/include -Iggml/src -Iinclude -Isrc -Icommon -DGGML_USE_CPU
 MK_CFLAGS    = -std=c11   -fPIC
-MK_CXXFLAGS  = -std=c++11 -fPIC
-MK_NVCCFLAGS = -std=c++11
+MK_CXXFLAGS  = -std=c++17 -fPIC
+MK_NVCCFLAGS = -std=c++17
 
 ifdef LLAMA_NO_CCACHE
 GGML_NO_CCACHE := 1
@@ -575,9 +579,12 @@ endif
 
 ifndef GGML_NO_AMX
 	MK_CPPFLAGS += -DGGML_USE_AMX
-	OBJ_GGML_EXT += ggml/src/ggml-amx/ggml-amx.o ggml/src/ggml-amx/mmq.o
+	OBJ_GGML_EXT += ggml/src/ggml-cpu/amx/amx.o ggml/src/ggml-cpu/amx/mmq.o
 endif
 
+# only necessary for the CPU backend files
+MK_CPPFLAGS += -Iggml/src/ggml-cpu
+
 ifdef GGML_RPC
 	MK_CPPFLAGS  += -DGGML_USE_RPC
 	OBJ_GGML_EXT += ggml/src/ggml-rpc.o
@@ -752,7 +759,7 @@ vulkan-shaders-gen: ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp
 
 endif # GGML_VULKAN
 
-ifdef GGML_HIPBLAS
+ifdef GGML_HIP
 	ifeq ($(wildcard /opt/rocm),)
 		ROCM_PATH      ?= /usr
 		AMDGPU_TARGETS ?= $(shell $(shell which amdgpu-arch))
@@ -807,7 +814,7 @@ ggml/src/ggml-cuda/%.o: \
 	ggml/src/ggml-common.h \
 	ggml/src/ggml-cuda/common.cuh
 	$(HIPCC) $(CXXFLAGS) $(HIPFLAGS) -x hip -c -o $@ $<
-endif # GGML_HIPBLAS
+endif # GGML_HIP
 
 ifdef GGML_MUSA
 	ifeq ($(wildcard /opt/musa),)
@@ -815,7 +822,7 @@ ifdef GGML_MUSA
 	else
 		MUSA_PATH ?= /opt/musa
 	endif
-	MTGPU_TARGETS ?= mp_21 mp_22
+	MUSA_ARCHITECTURES ?= 21;22
 
 	MK_CPPFLAGS += -DGGML_USE_MUSA -DGGML_USE_CUDA
 	MK_LDFLAGS += -L$(MUSA_PATH)/lib -Wl,-rpath=$(MUSA_PATH)/lib
@@ -834,7 +841,8 @@ ifdef GGML_MUSA
 	CXX := $(MUSA_PATH)/bin/clang++
 	MCC := $(CCACHE) $(MUSA_PATH)/bin/mcc
 
-	MUSAFLAGS += $(addprefix --cuda-gpu-arch=, $(MTGPU_TARGETS))
+	MUSAFLAGS  = -x musa -mtgpu
+	MUSAFLAGS += $(foreach arch,$(subst ;, ,$(MUSA_ARCHITECTURES)),--cuda-gpu-arch=mp_$(arch))
 
 ifdef GGML_CUDA_FORCE_MMQ
 	MUSAFLAGS += -DGGML_CUDA_FORCE_MMQ
@@ -878,14 +886,14 @@ ggml/src/ggml-cuda/ggml-cuda.o: \
 	ggml/src/ggml-backend-impl.h \
 	ggml/src/ggml-common.h \
 	$(wildcard ggml/src/ggml-cuda/*.cuh)
-	$(MCC) $(CXXFLAGS) $(MUSAFLAGS) -x musa -mtgpu -c -o $@ $<
+	$(MCC) $(CXXFLAGS) $(MUSAFLAGS) -c -o $@ $<
 
 ggml/src/ggml-cuda/%.o: \
 	ggml/src/ggml-cuda/%.cu \
 	ggml/include/ggml.h \
 	ggml/src/ggml-common.h \
 	ggml/src/ggml-cuda/common.cuh
-	$(MCC) $(CXXFLAGS) $(MUSAFLAGS) -x musa -mtgpu -c -o $@ $<
+	$(MCC) $(CXXFLAGS) $(MUSAFLAGS) -c -o $@ $<
 endif # GGML_MUSA
 
 ifdef GGML_METAL
@@ -1137,8 +1145,15 @@ $(LIB_COMMON_S): $(OBJ_COMMON)
 # Include dependency files
 -include $(DEP_FILES)
 
+# Clean generated server assets
+clean-server-assets:
+	find examples/server -type f -name "*.js.hpp"   -delete
+	find examples/server -type f -name "*.mjs.hpp"  -delete
+	find examples/server -type f -name "*.css.hpp"  -delete
+	find examples/server -type f -name "*.html.hpp" -delete
+
 # Clean rule
-clean:
+clean: clean-server-assets
 	rm -vrf $(BUILD_TARGETS) $(TEST_TARGETS)
 	rm -rvf *.a *.dll *.so *.dot
 	find ggml src common tests examples pocs -type f -name "*.o" -delete
@@ -1346,20 +1361,14 @@ llama-server: \
 	examples/server/utils.hpp \
 	examples/server/httplib.h \
 	examples/server/index.html.hpp \
-	examples/server/completion.js.hpp \
 	examples/server/loading.html.hpp \
-	examples/server/deps_daisyui.min.css.hpp \
-	examples/server/deps_markdown-it.js.hpp \
-	examples/server/deps_tailwindcss.js.hpp \
-	examples/server/deps_vue.esm-browser.js.hpp \
 	common/json.hpp \
-	common/stb_image.h \
 	$(OBJ_ALL)
 	$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h %.hpp $<,$^) -Iexamples/server $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS) $(LWINSOCK2)
 
 # Portable equivalent of `cd examples/server/public && xxd -i $(notdir $<) ../$(notdir $<).hpp`:
-examples/server/%.hpp: examples/server/public/% Makefile
+examples/server/%.hpp: examples/server/public/% FORCE Makefile
 	@( export NAME=$(subst .,_,$(subst -,_,$(notdir $<))) && \
 		echo "unsigned char $${NAME}[] = {" && \
 		cat $< | od -v -t x1 -An | sed -E 's/([0-9a-fA-F]+)/0x\1, /g' && \
@@ -1534,7 +1543,7 @@ llama-q8dot: pocs/vdot/q8dot.cpp ggml/src/ggml.o \
 # Deprecated binaries that we want to keep around long enough for people to migrate to the new filenames, then these can be removed.
 #
 # Mark legacy binary targets as .PHONY so that they are always checked.
-.PHONY: main quantize perplexity embedding server
+.PHONY: FORCE main quantize perplexity embedding server
 
 # Define the object file target
 examples/deprecation-warning/deprecation-warning.o: examples/deprecation-warning/deprecation-warning.cpp

Package.swift

@@ -28,13 +28,16 @@ var cSettings: [CSetting] =  [
     .unsafeFlags(["-Wno-shorten-64-to-32", "-O3", "-DNDEBUG"]),
     .unsafeFlags(["-fno-objc-arc"]),
     .headerSearchPath("ggml/src"),
+    .headerSearchPath("ggml/src/ggml-cpu"),
     // NOTE: NEW_LAPACK will required iOS version 16.4+
     // We should consider add this in the future when we drop support for iOS 14
     // (ref: ref: https://developer.apple.com/documentation/accelerate/1513264-cblas_sgemm?language=objc)
     // .define("ACCELERATE_NEW_LAPACK"),
     // .define("ACCELERATE_LAPACK_ILP64")
+    .define("GGML_USE_CPU"),
 ]
 
+
 #if canImport(Darwin)
 sources.append("ggml/src/ggml-common.h")
 sources.append("ggml/src/ggml-metal/ggml-metal.m")
@@ -44,7 +47,6 @@ cSettings.append(
     contentsOf: [
         .define("GGML_USE_ACCELERATE"),
         .define("GGML_USE_METAL"),
-        .define("GGML_USE_CPU")
     ]
 )
 #endif
@@ -86,5 +88,5 @@ let package = Package(
             linkerSettings: linkerSettings
         )
     ],
-    cxxLanguageStandard: .cxx11
+    cxxLanguageStandard: .cxx17
 )

README.md

@@ -4,7 +4,6 @@
 
 [![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)](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)
 
@@ -26,7 +25,7 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others)
 ## Description
 
 The main goal of `llama.cpp` is to enable LLM inference with minimal setup and state-of-the-art performance on a wide
-variety of hardware - locally and in the cloud.
+range of hardware - locally and in the cloud.
 
 - Plain C/C++ implementation without any dependencies
 - Apple silicon is a first-class citizen - optimized via ARM NEON, Accelerate and Metal frameworks
@@ -36,14 +35,17 @@ variety of hardware - locally and in the cloud.
 - 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
-improved significantly thanks to many contributions. It is the main playground for developing new features for the
-[ggml](https://github.com/ggerganov/ggml) library.
+The `llama.cpp` project is the main playground for developing new features for the [ggml](https://github.com/ggerganov/ggml) library.
 
-**Supported models:**
+<details>
+<summary>Models</summary>
 
 Typically finetunes of the base models below are supported as well.
 
+Instructions for adding support for new models: [HOWTO-add-model.md](docs/development/HOWTO-add-model.md)
+
+#### Text-only
+
 - [X] LLaMA 🦙
 - [x] LLaMA 2 🦙🦙
 - [x] LLaMA 3 🦙🦙🦙
@@ -79,6 +81,7 @@ Typically finetunes of the base models below are supported as well.
 - [x] [SEA-LION](https://huggingface.co/models?search=sea-lion)
 - [x] [GritLM-7B](https://huggingface.co/GritLM/GritLM-7B) + [GritLM-8x7B](https://huggingface.co/GritLM/GritLM-8x7B)
 - [x] [OLMo](https://allenai.org/olmo)
+- [x] [OLMo 2](https://allenai.org/olmo)
 - [x] [OLMoE](https://huggingface.co/allenai/OLMoE-1B-7B-0924)
 - [x] [Granite models](https://huggingface.co/collections/ibm-granite/granite-code-models-6624c5cec322e4c148c8b330)
 - [x] [GPT-NeoX](https://github.com/EleutherAI/gpt-neox) + [Pythia](https://github.com/EleutherAI/pythia)
@@ -96,9 +99,7 @@ Typically finetunes of the base models below are supported as well.
 - [x] [Bielik-11B-v2.3](https://huggingface.co/collections/speakleash/bielik-11b-v23-66ee813238d9b526a072408a)
 - [x] [RWKV-6](https://github.com/BlinkDL/RWKV-LM)
 
-(instructions for supporting more models: [HOWTO-add-model.md](./docs/development/HOWTO-add-model.md))
-
-**Multimodal models:**
+#### Multimodal
 
 - [x] [LLaVA 1.5 models](https://huggingface.co/collections/liuhaotian/llava-15-653aac15d994e992e2677a7e), [LLaVA 1.6 models](https://huggingface.co/collections/liuhaotian/llava-16-65b9e40155f60fd046a5ccf2)
 - [x] [BakLLaVA](https://huggingface.co/models?search=SkunkworksAI/Bakllava)
@@ -110,7 +111,10 @@ Typically finetunes of the base models below are supported as well.
 - [x] [Moondream](https://huggingface.co/vikhyatk/moondream2)
 - [x] [Bunny](https://github.com/BAAI-DCAI/Bunny)
 
-**Bindings:**
+</details>
+
+<details>
+<summary>Bindings</summary>
 
 - Python: [abetlen/llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
 - Go: [go-skynet/go-llama.cpp](https://github.com/go-skynet/go-llama.cpp)
@@ -137,316 +141,314 @@ Typically finetunes of the base models below are supported as well.
 - Swift [srgtuszy/llama-cpp-swift](https://github.com/srgtuszy/llama-cpp-swift)
 - Swift [ShenghaiWang/SwiftLlama](https://github.com/ShenghaiWang/SwiftLlama)
 
-**UI:**
+</details>
 
-Unless otherwise noted these projects are open-source with permissive licensing:
-
-- [MindWorkAI/AI-Studio](https://github.com/MindWorkAI/AI-Studio) (FSL-1.1-MIT)
-- [iohub/collama](https://github.com/iohub/coLLaMA)
-- [janhq/jan](https://github.com/janhq/jan) (AGPL)
-- [nat/openplayground](https://github.com/nat/openplayground)
-- [Faraday](https://faraday.dev/) (proprietary)
-- [LMStudio](https://lmstudio.ai/) (proprietary)
-- [Layla](https://play.google.com/store/apps/details?id=com.laylalite) (proprietary)
-- [ramalama](https://github.com/containers/ramalama) (MIT)
-- [LocalAI](https://github.com/mudler/LocalAI) (MIT)
-- [LostRuins/koboldcpp](https://github.com/LostRuins/koboldcpp) (AGPL)
-- [Mozilla-Ocho/llamafile](https://github.com/Mozilla-Ocho/llamafile)
-- [nomic-ai/gpt4all](https://github.com/nomic-ai/gpt4all)
-- [ollama/ollama](https://github.com/ollama/ollama)
-- [oobabooga/text-generation-webui](https://github.com/oobabooga/text-generation-webui) (AGPL)
-- [psugihara/FreeChat](https://github.com/psugihara/FreeChat)
-- [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)
-- [Mobile-Artificial-Intelligence/maid](https://github.com/Mobile-Artificial-Intelligence/maid) (MIT)
-- [Msty](https://msty.app) (proprietary)
-- [LLMFarm](https://github.com/guinmoon/LLMFarm?tab=readme-ov-file) (MIT)
-- [KanTV](https://github.com/zhouwg/kantv?tab=readme-ov-file)(Apachev2.0 or later)
-- [Dot](https://github.com/alexpinel/Dot) (GPL)
-- [MindMac](https://mindmac.app) (proprietary)
-- [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)
-- [LLMUnity](https://github.com/undreamai/LLMUnity) (MIT)
-- [Llama Assistant](https://github.com/vietanhdev/llama-assistant) (GPL)
-- [PocketPal AI - An iOS and Android App](https://github.com/a-ghorbani/pocketpal-ai) (MIT)
+<details>
+<summary>UIs</summary>
 
 *(to have a project listed here, it should clearly state that it depends on `llama.cpp`)*
 
-**Tools:**
+- [AI Sublime Text plugin](https://github.com/yaroslavyaroslav/OpenAI-sublime-text) (MIT)
+- [cztomsik/ava](https://github.com/cztomsik/ava) (MIT)
+- [Dot](https://github.com/alexpinel/Dot) (GPL)
+- [eva](https://github.com/ylsdamxssjxxdd/eva) (MIT)
+- [iohub/collama](https://github.com/iohub/coLLaMA) (Apache-2.0)
+- [janhq/jan](https://github.com/janhq/jan) (AGPL)
+- [KanTV](https://github.com/zhouwg/kantv?tab=readme-ov-file) (Apache-2.0)
+- [KodiBot](https://github.com/firatkiral/kodibot) (GPL)
+- [llama.vim](https://github.com/ggml-org/llama.vim) (MIT)
+- [LARS](https://github.com/abgulati/LARS) (AGPL)
+- [Llama Assistant](https://github.com/vietanhdev/llama-assistant) (GPL)
+- [LLMFarm](https://github.com/guinmoon/LLMFarm?tab=readme-ov-file) (MIT)
+- [LLMUnity](https://github.com/undreamai/LLMUnity) (MIT)
+- [LMStudio](https://lmstudio.ai/) (proprietary)
+- [LocalAI](https://github.com/mudler/LocalAI) (MIT)
+- [LostRuins/koboldcpp](https://github.com/LostRuins/koboldcpp) (AGPL)
+- [MindMac](https://mindmac.app) (proprietary)
+- [MindWorkAI/AI-Studio](https://github.com/MindWorkAI/AI-Studio) (FSL-1.1-MIT)
+- [Mobile-Artificial-Intelligence/maid](https://github.com/Mobile-Artificial-Intelligence/maid) (MIT)
+- [Mozilla-Ocho/llamafile](https://github.com/Mozilla-Ocho/llamafile) (Apache-2.0)
+- [nat/openplayground](https://github.com/nat/openplayground) (MIT)
+- [nomic-ai/gpt4all](https://github.com/nomic-ai/gpt4all) (MIT)
+- [ollama/ollama](https://github.com/ollama/ollama) (MIT)
+- [oobabooga/text-generation-webui](https://github.com/oobabooga/text-generation-webui) (AGPL)
+- [PocketPal AI](https://github.com/a-ghorbani/pocketpal-ai) (MIT)
+- [psugihara/FreeChat](https://github.com/psugihara/FreeChat) (MIT)
+- [ptsochantaris/emeltal](https://github.com/ptsochantaris/emeltal) (MIT)
+- [pythops/tenere](https://github.com/pythops/tenere) (AGPL)
+- [ramalama](https://github.com/containers/ramalama) (MIT)
+- [semperai/amica](https://github.com/semperai/amica) (MIT)
+- [withcatai/catai](https://github.com/withcatai/catai) (MIT)
+
+</details>
+
+<details>
+<summary>Tools</summary>
 
 - [akx/ggify](https://github.com/akx/ggify) – download PyTorch models from HuggingFace Hub and convert them to GGML
 - [akx/ollama-dl](https://github.com/akx/ollama-dl) – download models from the Ollama library to be used directly with llama.cpp
 - [crashr/gppm](https://github.com/crashr/gppm) – launch llama.cpp instances utilizing NVIDIA Tesla P40 or P100 GPUs with reduced idle power consumption
 - [gpustack/gguf-parser](https://github.com/gpustack/gguf-parser-go/tree/main/cmd/gguf-parser) - review/check the GGUF file and estimate the memory usage
-- [Styled Lines](https://marketplace.unity.com/packages/tools/generative-ai/styled-lines-llama-cpp-model-292902) (proprietary licensed, async wrapper of inference part for game development in Unity3d with prebuild Mobile and Web platform wrappers and a model example)
+- [Styled Lines](https://marketplace.unity.com/packages/tools/generative-ai/styled-lines-llama-cpp-model-292902) (proprietary licensed, async wrapper of inference part for game development in Unity3d with pre-built Mobile and Web platform wrappers and a model example)
 
-**Infrastructure:**
+</details>
+
+<details>
+<summary>Infrastructure</summary>
 
 - [Paddler](https://github.com/distantmagic/paddler) - Stateful load balancer custom-tailored for llama.cpp
 - [GPUStack](https://github.com/gpustack/gpustack) - Manage GPU clusters for running LLMs
 - [llama_cpp_canister](https://github.com/onicai/llama_cpp_canister) - llama.cpp as a smart contract on the Internet Computer, using WebAssembly
 
-**Games:**
+</details>
+
+<details>
+<summary>Games</summary>
+
 - [Lucy's Labyrinth](https://github.com/MorganRO8/Lucys_Labyrinth) - A simple maze game where agents controlled by an AI model will try to trick you.
 
-## Demo
-
-<details>
-<summary>Typical run using LLaMA v2 13B on M2 Ultra</summary>
-
-```
-$ 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
-I UNAME_M:  arm64
-I CFLAGS:   -I.            -O3 -std=c11   -fPIC -DNDEBUG -Wall -Wextra -Wpedantic -Wcast-qual -Wdouble-promotion -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -pthread -DGGML_USE_K_QUANTS -DGGML_USE_ACCELERATE
-I CXXFLAGS: -I. -I./common -O3 -std=c++11 -fPIC -DNDEBUG -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-multichar -pthread -DGGML_USE_K_QUANTS
-I LDFLAGS:   -framework Accelerate
-I CC:       Apple clang version 14.0.3 (clang-1403.0.22.14.1)
-I CXX:      Apple clang version 14.0.3 (clang-1403.0.22.14.1)
-
-make: Nothing to be done for `default'.
-main: build = 1041 (cf658ad)
-main: seed  = 1692823051
-llama_model_loader: loaded meta data with 16 key-value pairs and 363 tensors from models/llama-13b-v2/ggml-model-q4_0.gguf (version GGUF V1 (latest))
-llama_model_loader: - type  f32:   81 tensors
-llama_model_loader: - type q4_0:  281 tensors
-llama_model_loader: - type q6_K:    1 tensors
-llm_load_print_meta: format         = GGUF V1 (latest)
-llm_load_print_meta: arch           = llama
-llm_load_print_meta: vocab type     = SPM
-llm_load_print_meta: n_vocab        = 32000
-llm_load_print_meta: n_merges       = 0
-llm_load_print_meta: n_ctx_train    = 4096
-llm_load_print_meta: n_ctx          = 512
-llm_load_print_meta: n_embd         = 5120
-llm_load_print_meta: n_head         = 40
-llm_load_print_meta: n_head_kv      = 40
-llm_load_print_meta: n_layer        = 40
-llm_load_print_meta: n_rot          = 128
-llm_load_print_meta: n_gqa          = 1
-llm_load_print_meta: f_norm_eps     = 1.0e-05
-llm_load_print_meta: f_norm_rms_eps = 1.0e-05
-llm_load_print_meta: n_ff           = 13824
-llm_load_print_meta: freq_base      = 10000.0
-llm_load_print_meta: freq_scale     = 1
-llm_load_print_meta: model type     = 13B
-llm_load_print_meta: model ftype    = mostly Q4_0
-llm_load_print_meta: model size     = 13.02 B
-llm_load_print_meta: general.name   = LLaMA v2
-llm_load_print_meta: BOS token = 1 '<s>'
-llm_load_print_meta: EOS token = 2 '</s>'
-llm_load_print_meta: UNK token = 0 '<unk>'
-llm_load_print_meta: LF token  = 13 '<0x0A>'
-llm_load_tensors: ggml ctx size =    0.11 MB
-llm_load_tensors: mem required  = 7024.01 MB (+  400.00 MB per state)
-...................................................................................................
-llama_new_context_with_model: kv self size  =  400.00 MB
-llama_new_context_with_model: compute buffer total size =   75.41 MB
-
-system_info: n_threads = 16 / 24 | AVX = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | VSX = 0 |
-sampling: repeat_last_n = 64, repeat_penalty = 1.100000, presence_penalty = 0.000000, frequency_penalty = 0.000000, top_k = 40, tfs_z = 1.000000, top_p = 0.950000, typical_p = 1.000000, temp = 0.800000, mirostat = 0, mirostat_lr = 0.100000, mirostat_ent = 5.000000
-generate: n_ctx = 512, n_batch = 512, n_predict = 400, n_keep = 0
-
-
- Building a website can be done in 10 simple steps:
-Step 1: Find the right website platform.
-Step 2: Choose your domain name and hosting plan.
-Step 3: Design your website layout.
-Step 4: Write your website content and add images.
-Step 5: Install security features to protect your site from hackers or spammers
-Step 6: Test your website on multiple browsers, mobile devices, operating systems etc…
-Step 7: Test it again with people who are not related to you personally – friends or family members will work just fine!
-Step 8: Start marketing and promoting the website via social media channels or paid ads
-Step 9: Analyze how many visitors have come to your site so far, what type of people visit more often than others (e.g., men vs women) etc…
-Step 10: Continue to improve upon all aspects mentioned above by following trends in web design and staying up-to-date on new technologies that can enhance user experience even further!
-How does a Website Work?
-A website works by having pages, which are made of HTML code. This code tells your computer how to display the content on each page you visit – whether it’s an image or text file (like PDFs). In order for someone else’s browser not only be able but also want those same results when accessing any given URL; some additional steps need taken by way of programming scripts that will add functionality such as making links clickable!
-The most common type is called static HTML pages because they remain unchanged over time unless modified manually (either through editing files directly or using an interface such as WordPress). They are usually served up via HTTP protocols – this means anyone can access them without having any special privileges like being part of a group who is allowed into restricted areas online; however, there may still exist some limitations depending upon where one lives geographically speaking.
-How to
-llama_print_timings:        load time =   576.45 ms
-llama_print_timings:      sample time =   283.10 ms /   400 runs   (    0.71 ms per token,  1412.91 tokens per second)
-llama_print_timings: prompt eval time =   599.83 ms /    19 tokens (   31.57 ms per token,    31.68 tokens per second)
-llama_print_timings:        eval time = 24513.59 ms /   399 runs   (   61.44 ms per token,    16.28 tokens per second)
-llama_print_timings:       total time = 25431.49 ms
-```
-
 </details>
 
-<details>
-<summary>Demo of running both LLaMA-7B and whisper.cpp on a single M1 Pro MacBook</summary>
-
-And here is another demo of running both LLaMA-7B and [whisper.cpp](https://github.com/ggerganov/whisper.cpp) on a single M1 Pro MacBook:
-
-https://user-images.githubusercontent.com/1991296/224442907-7693d4be-acaa-4e01-8b4f-add84093ffff.mp4
-
-</details>
-
-## Usage
-
-Here are the end-to-end binary build and model conversion steps for most supported models.
-
-### Basic usage
-
-Firstly, you need to get the binary. There are different methods that you can follow:
-- Method 1: Clone this repository and build locally, see [how to build](./docs/build.md)
-- Method 2: If you are using MacOS or Linux, you can install llama.cpp via [brew, flox or nix](./docs/install.md)
-- Method 3: Use a Docker image, see [documentation for Docker](./docs/docker.md)
-- Method 4: Download pre-built binary from [releases](https://github.com/ggerganov/llama.cpp/releases)
-
-You can run a basic completion using this command:
-
-```bash
-llama-cli -m your_model.gguf -p "I believe the meaning of life is" -n 128
-
-# Output:
-# I believe the meaning of life is to find your own truth and to live in accordance with it. For me, this means being true to myself and following my passions, even if they don't align with societal expectations. I think that's what I love about yoga – it's not just a physical practice, but a spiritual one too. It's about connecting with yourself, listening to your inner voice, and honoring your own unique journey.
-```
-
-See [this page](./examples/main/README.md) for a full list of parameters.
-
-### Conversation mode
-
-If you want a more ChatGPT-like experience, you can run in conversation mode by passing `-cnv` as a parameter:
-
-```bash
-llama-cli -m your_model.gguf -p "You are a helpful assistant" -cnv
-
-# Output:
-# > hi, who are you?
-# Hi there! I'm your helpful assistant! I'm an AI-powered chatbot designed to assist and provide information to users like you. I'm here to help answer your questions, provide guidance, and offer support on a wide range of topics. I'm a friendly and knowledgeable AI, and I'm always happy to help with anything you need. What's on your mind, and how can I assist you today?
-#
-# > what is 1+1?
-# Easy peasy! The answer to 1+1 is... 2!
-```
-
-By default, the chat template will be taken from the input model. If you want to use another chat template, pass `--chat-template NAME` as a parameter. See the list of [supported templates](https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template)
-
-```bash
-./llama-cli -m your_model.gguf -p "You are a helpful assistant" -cnv --chat-template chatml
-```
-
-You can also use your own template via in-prefix, in-suffix and reverse-prompt parameters:
-
-```bash
-./llama-cli -m your_model.gguf -p "You are a helpful assistant" -cnv --in-prefix 'User: ' --reverse-prompt 'User:'
-```
-
-### Web server
-
-[llama.cpp web server](./examples/server/README.md) 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.
-
-Example usage:
-
-```bash
-./llama-server -m your_model.gguf --port 8080
-
-# Basic web UI can be accessed via browser: http://localhost:8080
-# Chat completion endpoint: http://localhost:8080/v1/chat/completions
-```
-
-### Interactive mode
-
-> [!NOTE]
-> If you prefer basic usage, please consider using conversation mode instead of interactive mode
-
-In this mode, you can always interrupt generation by pressing Ctrl+C and entering one or more lines of text, which will be converted into tokens and appended to the current context. You can also specify a *reverse prompt* with the parameter `-r "reverse prompt string"`. This will result in user input being prompted whenever the exact tokens of the reverse prompt string are encountered in the generation. A typical use is to use a prompt that makes LLaMA emulate a chat between multiple users, say Alice and Bob, and pass `-r "Alice:"`.
-
-Here is an example of a few-shot interaction, invoked with the command
-
-```bash
-# default arguments using a 7B model
-./examples/chat.sh
-
-# advanced chat with a 13B model
-./examples/chat-13B.sh
-
-# custom arguments using a 13B model
-./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 `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 `./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
-PROMPT_CACHE_FILE=chat.prompt.bin CHAT_SAVE_DIR=./chat/default ./examples/chat-persistent.sh
-
-# Resume that chat
-PROMPT_CACHE_FILE=chat.prompt.bin CHAT_SAVE_DIR=./chat/default ./examples/chat-persistent.sh
-
-# Start a different chat with the same prompt/model
-PROMPT_CACHE_FILE=chat.prompt.bin CHAT_SAVE_DIR=./chat/another ./examples/chat-persistent.sh
-
-# Different prompt cache for different prompt/model
-PROMPT_TEMPLATE=./prompts/chat-with-bob.txt PROMPT_CACHE_FILE=bob.prompt.bin \
-    CHAT_SAVE_DIR=./chat/bob ./examples/chat-persistent.sh
-```
-
-### Constrained output with grammars
-
-`llama.cpp` supports grammars to constrain model output. For example, you can force the model to output JSON only:
-
-```bash
-./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.
-
-## Build
-
-Please refer to [Build llama.cpp locally](./docs/build.md)
-
 ## Supported backends
 
 | Backend | Target devices |
 | --- | --- |
-| [Metal](./docs/build.md#metal-build) | Apple Silicon |
-| [BLAS](./docs/build.md#blas-build) | All |
-| [BLIS](./docs/backend/BLIS.md) | All |
-| [SYCL](./docs/backend/SYCL.md) | Intel and Nvidia GPU |
-| [MUSA](./docs/build.md#musa) | Moore Threads MTT GPU |
-| [CUDA](./docs/build.md#cuda) | Nvidia GPU |
-| [hipBLAS](./docs/build.md#hipblas) | AMD GPU |
-| [Vulkan](./docs/build.md#vulkan) | GPU |
-| [CANN](./docs/build.md#cann) | Ascend NPU |
+| [Metal](docs/build.md#metal-build) | Apple Silicon |
+| [BLAS](docs/build.md#blas-build) | All |
+| [BLIS](docs/backend/BLIS.md) | All |
+| [SYCL](docs/backend/SYCL.md) | Intel and Nvidia GPU |
+| [MUSA](docs/build.md#musa) | Moore Threads MTT GPU |
+| [CUDA](docs/build.md#cuda) | Nvidia GPU |
+| [hipBLAS](docs/build.md#hipblas) | AMD GPU |
+| [Vulkan](docs/build.md#vulkan) | GPU |
+| [CANN](docs/build.md#cann) | Ascend NPU |
 
-## Tools
+## Building the project
 
-### Prepare and Quantize
+The main product of this project is the `llama` library. Its C-style interface can be found in [include/llama.h](include/llama.h).
+The project also includes many example programs and tools using the `llama` library. The examples range from simple, minimal code snippets to sophisticated sub-projects such as an OpenAI-compatible HTTP server. Possible methods for obtaining the binaries:
 
-> [!NOTE]
-> You can use the [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space on Hugging Face to quantise your model weights without any setup too. It is synced from `llama.cpp` main every 6 hours.
+- Clone this repository and build locally, see [how to build](docs/build.md)
+- On MacOS or Linux, install `llama.cpp` via [brew, flox or nix](docs/install.md)
+- Use a Docker image, see [documentation for Docker](docs/docker.md)
+- Download pre-built binaries from [releases](https://github.com/ggerganov/llama.cpp/releases)
 
-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.
+## Obtaining and quantizing models
 
-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.
+The [Hugging Face](https://huggingface.co) platform hosts a [number of LLMs](https://huggingface.co/models?library=gguf&sort=trending) compatible with `llama.cpp`:
 
-To learn more about quantizing model, [read this documentation](./examples/quantize/README.md)
+- [Trending](https://huggingface.co/models?library=gguf&sort=trending)
+- [LLaMA](https://huggingface.co/models?sort=trending&search=llama+gguf)
 
-### Perplexity (measuring model quality)
+After downloading a model, use the CLI tools to run it locally - see below.
 
-You can use the `perplexity` example to measure perplexity over a given prompt (lower perplexity is better).
-For more information, see [https://huggingface.co/docs/transformers/perplexity](https://huggingface.co/docs/transformers/perplexity).
+`llama.cpp` requires the model to be stored in the [GGUF](https://github.com/ggerganov/ggml/blob/master/docs/gguf.md) file format. Models in other data formats can be converted to GGUF using the `convert_*.py` Python scripts in this repo.
+
+The Hugging Face platform provides a variety of online tools for converting, quantizing and hosting models with `llama.cpp`:
+
+- Use the [GGUF-my-repo space](https://huggingface.co/spaces/ggml-org/gguf-my-repo) to convert to GGUF format and quantize model weights to smaller sizes
+- Use the [GGUF-my-LoRA space](https://huggingface.co/spaces/ggml-org/gguf-my-lora) to convert LoRA adapters to GGUF format (more info: https://github.com/ggerganov/llama.cpp/discussions/10123)
+- Use the [GGUF-editor space](https://huggingface.co/spaces/CISCai/gguf-editor) to edit GGUF meta data in the browser (more info: https://github.com/ggerganov/llama.cpp/discussions/9268)
+- Use the [Inference Endpoints](https://ui.endpoints.huggingface.co/) to directly host `llama.cpp` in the cloud (more info: https://github.com/ggerganov/llama.cpp/discussions/9669)
+
+To learn more about model quantization, [read this documentation](examples/quantize/README.md)
+
+## [`llama-cli`](examples/main)
+
+#### A CLI tool for accessing and experimenting with most of `llama.cpp`'s functionality.
+
+- <details open>
+    <summary>Run simple text completion</summary>
+
+    ```bash
+    llama-cli -m model.gguf -p "I believe the meaning of life is" -n 128
+
+    # I believe the meaning of life is to find your own truth and to live in accordance with it. For me, this means being true to myself and following my passions, even if they don't align with societal expectations. I think that's what I love about yoga – it's not just a physical practice, but a spiritual one too. It's about connecting with yourself, listening to your inner voice, and honoring your own unique journey.
+    ```
+
+    </details>
+
+- <details>
+    <summary>Run in conversation mode</summary>
+
+    ```bash
+    llama-cli -m model.gguf -p "You are a helpful assistant" -cnv
+
+    # > hi, who are you?
+    # Hi there! I'm your helpful assistant! I'm an AI-powered chatbot designed to assist and provide information to users like you. I'm here to help answer your questions, provide guidance, and offer support on a wide range of topics. I'm a friendly and knowledgeable AI, and I'm always happy to help with anything you need. What's on your mind, and how can I assist you today?
+    #
+    # > what is 1+1?
+    # Easy peasy! The answer to 1+1 is... 2!
+    ```
+
+    </details>
+
+- <details>
+    <summary>Run with custom chat template</summary>
+
+    ```bash
+    # use the "chatml" template
+    llama-cli -m model.gguf -p "You are a helpful assistant" -cnv --chat-template chatml
+
+    # use a custom template
+    llama-cli -m model.gguf -p "You are a helpful assistant" -cnv --in-prefix 'User: ' --reverse-prompt 'User:'
+    ```
+
+    [Supported templates](https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template)
+
+    </details>
+
+- <details>
+    <summary>Constrain the output with a custom grammar</summary>
+
+    ```bash
+    llama-cli -m model.gguf -n 256 --grammar-file grammars/json.gbnf -p 'Request: schedule a call at 8pm; Command:'
+
+    # {"appointmentTime": "8pm", "appointmentDetails": "schedule a a call"}
+    ```
+
+    The [grammars/](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, check out https://grammar.intrinsiclabs.ai/
+
+    </details>
+
+
+## [`llama-server`](examples/server)
+
+#### A lightweight, [OpenAI API](https://github.com/openai/openai-openapi) compatible, HTTP server for serving LLMs.
+
+- <details open>
+    <summary>Start a local HTTP server with default configuration on port 8080</summary>
+
+    ```bash
+    llama-server -m model.gguf --port 8080
+
+    # Basic web UI can be accessed via browser: http://localhost:8080
+    # Chat completion endpoint: http://localhost:8080/v1/chat/completions
+    ```
+
+    </details>
+
+- <details>
+    <summary>Support multiple-users and parallel decoding</summary>
+
+    ```bash
+    # up to 4 concurrent requests, each with 4096 max context
+    llama-server -m model.gguf -c 16384 -np 4
+    ```
+
+    </details>
+
+- <details>
+    <summary>Enable speculative decoding</summary>
+
+    ```bash
+    # the draft.gguf model should be a small variant of the target model.gguf
+    llama-server -m model.gguf -md draft.gguf
+    ```
+
+    </details>
+
+- <details>
+    <summary>Serve an embedding model</summary>
+
+    ```bash
+    # use the /embedding endpoint
+    llama-server -m model.gguf --embedding --pooling cls -ub 8192
+    ```
+
+    </details>
+
+- <details>
+    <summary>Serve a reranking model</summary>
+
+    ```bash
+    # use the /reranking endpoint
+    llama-server -m model.gguf --reranking
+    ```
+
+    </details>
+
+- <details>
+    <summary>Constrain all outputs with a grammar</summary>
+
+    ```bash
+    # custom grammar
+    llama-server -m model.gguf --grammar-file grammar.gbnf
+
+    # JSON
+    llama-server -m model.gguf --grammar-file grammars/json.gbnf
+    ```
+
+    </details>
+
+
+## [`llama-perplexity`](examples/perplexity)
+
+#### A tool for measuring the perplexity [^1][^2] (and other quality metrics) of a model over a given text.
+
+- <details open>
+    <summary>Measure the perplexity over a text file</summary>
+
+    ```bash
+    llama-perplexity -m model.gguf -f file.txt
+
+    # [1]15.2701,[2]5.4007,[3]5.3073,[4]6.2965,[5]5.8940,[6]5.6096,[7]5.7942,[8]4.9297, ...
+    # Final estimate: PPL = 5.4007 +/- 0.67339
+    ```
+
+    </details>
+
+- <details>
+    <summary>Measure KL divergence</summary>
+
+    ```bash
+    # TODO
+    ```
+
+    </details>
+
+[^1]: [examples/perplexity/README.md](examples/perplexity/README.md)
+[^2]: [https://huggingface.co/docs/transformers/perplexity](https://huggingface.co/docs/transformers/perplexity)
+
+## [`llama-bench`](example/bench)
+
+#### Benchmark the performance of the inference for various parameters.
+
+- <details open>
+    <summary>Run default benchmark</summary>
+
+    ```bash
+    llama-bench -m model.gguf
+
+    # Output:
+    # | model               |       size |     params | backend    | threads |          test |                  t/s |
+    # | ------------------- | ---------: | ---------: | ---------- | ------: | ------------: | -------------------: |
+    # | qwen2 1.5B Q4_0     | 885.97 MiB |     1.54 B | Metal,BLAS |      16 |         pp512 |      5765.41 ± 20.55 |
+    # | qwen2 1.5B Q4_0     | 885.97 MiB |     1.54 B | Metal,BLAS |      16 |         tg128 |        197.71 ± 0.81 |
+    #
+    # build: 3e0ba0e60 (4229)
+    ```
+
+    </details>
+
+
+## [`llama-simple`](examples/simple)
+
+#### A minimal example for implementing apps with `llama.cpp`. Useful for developers.
+
+- <details>
+    <summary>Basic text completion</summary>
+
+    ```bash
+    llama-simple -m model.gguf
+
+    # Hello my name is Kaitlyn and I am a 16 year old girl. I am a junior in high school and I am currently taking a class called "The Art of
+    ```
+
+    </details>
 
-To learn more how to measure perplexity using llama.cpp, [read this documentation](./examples/perplexity/README.md)
 
 ## Contributing
 
@@ -461,20 +463,19 @@ To learn more how to measure perplexity using llama.cpp, [read this documentatio
 
 ## Other documentation
 
-- [main (cli)](./examples/main/README.md)
-- [server](./examples/server/README.md)
-- [jeopardy](./examples/jeopardy/README.md)
-- [GBNF grammars](./grammars/README.md)
+- [main (cli)](examples/main/README.md)
+- [server](examples/server/README.md)
+- [GBNF grammars](grammars/README.md)
 
-**Development documentation**
+#### Development documentation
 
-- [How to build](./docs/build.md)
-- [Running on Docker](./docs/docker.md)
-- [Build on Android](./docs/android.md)
-- [Performance troubleshooting](./docs/development/token_generation_performance_tips.md)
+- [How to build](docs/build.md)
+- [Running on Docker](docs/docker.md)
+- [Build on Android](docs/android.md)
+- [Performance troubleshooting](docs/development/token_generation_performance_tips.md)
 - [GGML tips & tricks](https://github.com/ggerganov/llama.cpp/wiki/GGML-Tips-&-Tricks)
 
-**Seminal papers and background on the models**
+#### Seminal papers and background on the models
 
 If your issue is with model generation quality, then please at least scan the following links and papers to understand the limitations of LLaMA models. This is especially important when choosing an appropriate model size and appreciating both the significant and subtle differences between LLaMA models and ChatGPT:
 - LLaMA:
@@ -485,3 +486,6 @@ If your issue is with model generation quality, then please at least scan the fo
 - GPT-3.5 / InstructGPT / ChatGPT:
     - [Aligning language models to follow instructions](https://openai.com/research/instruction-following)
     - [Training language models to follow instructions with human feedback](https://arxiv.org/abs/2203.02155)
+
+#### References
+

ci/run.sh

@@ -815,7 +815,10 @@ if [ -z ${GG_BUILD_LOW_PERF} ]; then
     ln -sfn ${mnt_models} ${SRC}/models-mnt
 
     # Create a fresh python3 venv and enter it
-    python3 -m venv "$MNT/venv"
+    if ! python3 -m venv "$MNT/venv"; then
+        echo "Error: Failed to create Python virtual environment at $MNT/venv."
+        exit 1
+    fi
     source "$MNT/venv/bin/activate"
 
     pip install -r ${SRC}/requirements.txt --disable-pip-version-check

cmake/common.cmake

@@ -0,0 +1,33 @@
+function(llama_add_compile_flags)
+    if (LLAMA_FATAL_WARNINGS)
+        if (CMAKE_CXX_COMPILER_ID MATCHES "GNU" OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
+            list(APPEND C_FLAGS   -Werror)
+            list(APPEND CXX_FLAGS -Werror)
+        elseif (CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
+            add_compile_options(/WX)
+        endif()
+    endif()
+
+    if (LLAMA_ALL_WARNINGS)
+        if (NOT MSVC)
+            list(APPEND C_FLAGS -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes
+                                -Werror=implicit-int -Werror=implicit-function-declaration)
+
+            list(APPEND CXX_FLAGS -Wmissing-declarations -Wmissing-noreturn)
+
+            list(APPEND WARNING_FLAGS -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function)
+
+            list(APPEND C_FLAGS   ${WARNING_FLAGS})
+            list(APPEND CXX_FLAGS ${WARNING_FLAGS})
+
+            ggml_get_flags(${CMAKE_CXX_COMPILER_ID} ${CMAKE_CXX_COMPILER_VERSION})
+
+            add_compile_options("$<$<COMPILE_LANGUAGE:C>:${C_FLAGS};${GF_C_FLAGS}>"
+                                "$<$<COMPILE_LANGUAGE:CXX>:${CXX_FLAGS};${GF_CXX_FLAGS}>")
+        else()
+            # todo : msvc
+            set(C_FLAGS   "" PARENT_SCOPE)
+            set(CXX_FLAGS "" PARENT_SCOPE)
+        endif()
+    endif()
+endfunction()

common/CMakeLists.txt

@@ -2,6 +2,8 @@
 
 find_package(Threads REQUIRED)
 
+llama_add_compile_flags()
+
 # Build info header
 #
 
@@ -86,5 +88,5 @@ if (LLAMA_CURL)
 endif ()
 
 target_include_directories(${TARGET} PUBLIC .)
-target_compile_features   (${TARGET} PUBLIC cxx_std_11)
+target_compile_features   (${TARGET} PUBLIC cxx_std_17)
 target_link_libraries     (${TARGET} PRIVATE ${LLAMA_COMMON_EXTRA_LIBS} PUBLIC llama Threads::Threads)

common/arg.cpp

@@ -128,7 +128,11 @@ static void common_params_handle_model_default(common_params & params) {
             }
             params.hf_file = params.model;
         } else if (params.model.empty()) {
-            params.model = fs_get_cache_file(string_split<std::string>(params.hf_file, '/').back());
+            // this is to avoid different repo having same file name, or same file name in different subdirs
+            std::string filename = params.hf_repo + "_" + params.hf_file;
+            // to make sure we don't have any slashes in the filename
+            string_replace_all(filename, "/", "_");
+            params.model = fs_get_cache_file(filename);
         }
     } else if (!params.model_url.empty()) {
         if (params.model.empty()) {
@@ -344,6 +348,18 @@ bool common_params_parse(int argc, char ** argv, common_params & params, llama_e
     return true;
 }
 
+static std::string list_builtin_chat_templates() {
+    std::vector<const char *> supported_tmpl;
+    int32_t res = llama_chat_builtin_templates(nullptr, 0);
+    supported_tmpl.resize(res);
+    res = llama_chat_builtin_templates(supported_tmpl.data(), supported_tmpl.size());
+    std::ostringstream msg;
+    for (auto & tmpl : supported_tmpl) {
+        msg << tmpl << (&tmpl == &supported_tmpl.back() ? "" : ", ");
+    }
+    return msg.str();
+}
+
 common_params_context common_params_parser_init(common_params & params, llama_example ex, void(*print_usage)(int, char **)) {
     // load dynamic backends
     ggml_backend_load_all();
@@ -1366,8 +1382,9 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
         [](common_params & params, int value) {
             params.n_gpu_layers = value;
             if (!llama_supports_gpu_offload()) {
-                fprintf(stderr, "warning: not compiled with GPU offload support, --gpu-layers option will be ignored\n");
-                fprintf(stderr, "warning: see main README.md for information on enabling GPU BLAS support\n");
+                fprintf(stderr, "warning: no usable GPU found, --gpu-layers option will be ignored\n");
+                fprintf(stderr, "warning: one possible reason is that llama.cpp was compiled without GPU support\n");
+                fprintf(stderr, "warning: consult docs/build.md for compilation instructions\n");
             }
         }
     ).set_env("LLAMA_ARG_N_GPU_LAYERS"));
@@ -1809,9 +1826,11 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_examples({LLAMA_EXAMPLE_SERVER}));
     add_opt(common_arg(
         {"--chat-template"}, "JINJA_TEMPLATE",
-        "set custom jinja chat template (default: template taken from model's metadata)\n"
-        "if suffix/prefix are specified, template will be disabled\n"
-        "only commonly used templates are accepted:\nhttps://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template",
+        string_format(
+            "set custom jinja chat template (default: template taken from model's metadata)\n"
+            "if suffix/prefix are specified, template will be disabled\n"
+            "list of built-in templates:\n%s", list_builtin_chat_templates().c_str()
+        ),
         [](common_params & params, const std::string & value) {
             if (!common_chat_verify_template(value)) {
                 throw std::runtime_error(string_format(
@@ -2100,8 +2119,9 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
         [](common_params & params, int value) {
             params.speculative.n_gpu_layers = value;
             if (!llama_supports_gpu_offload()) {
-                fprintf(stderr, "warning: not compiled with GPU offload support, --gpu-layers-draft option will be ignored\n");
-                fprintf(stderr, "warning: see main README.md for information on enabling GPU BLAS support\n");
+                fprintf(stderr, "warning: no usable GPU found, --gpu-layers-draft option will be ignored\n");
+                fprintf(stderr, "warning: one possible reason is that llama.cpp was compiled without GPU support\n");
+                fprintf(stderr, "warning: consult docs/build.md for compilation instructions\n");
             }
         }
     ).set_examples({LLAMA_EXAMPLE_SPECULATIVE, LLAMA_EXAMPLE_SERVER}));

common/common.cpp

@@ -652,7 +652,17 @@ bool fs_validate_filename(const std::string & filename) {
 
     std::u32string filename_utf32;
     try {
+#if defined(__clang__)
+        // disable C++17 deprecation warning for std::codecvt_utf8
+#    pragma clang diagnostic push
+#    pragma clang diagnostic ignored "-Wdeprecated-declarations"
+#endif
         std::wstring_convert<std::codecvt_utf8<char32_t>, char32_t> converter;
+
+#if defined(__clang__)
+#    pragma clang diagnostic pop
+#endif
+
         filename_utf32 = converter.from_bytes(filename);
 
         // If the reverse conversion mismatches, it means overlong UTF-8 sequences were used,
@@ -829,9 +839,9 @@ struct common_init_result common_init_from_params(common_params & params) {
     llama_model * model = nullptr;
 
     if (!params.hf_repo.empty() && !params.hf_file.empty()) {
-        model = common_load_model_from_hf(params.hf_repo.c_str(), params.hf_file.c_str(), params.model.c_str(), params.hf_token.c_str(), mparams);
+        model = common_load_model_from_hf(params.hf_repo, params.hf_file, params.model, params.hf_token, mparams);
     } else if (!params.model_url.empty()) {
-        model = common_load_model_from_url(params.model_url.c_str(), params.model.c_str(), params.hf_token.c_str(), mparams);
+        model = common_load_model_from_url(params.model_url, params.model, params.hf_token, mparams);
     } else {
         model = llama_load_model_from_file(params.model.c_str(), mparams);
     }
@@ -1342,17 +1352,17 @@ static bool common_download_file(const std::string & url, const std::string & pa
 }
 
 struct llama_model * common_load_model_from_url(
-        const char * model_url,
-        const char * path_model,
-        const char * hf_token,
+        const std::string & model_url,
+        const std::string & local_path,
+        const std::string & hf_token,
         const struct llama_model_params & params) {
     // Basic validation of the model_url
-    if (!model_url || strlen(model_url) == 0) {
+    if (model_url.empty()) {
         LOG_ERR("%s: invalid model_url\n", __func__);
         return NULL;
     }
 
-    if (!common_download_file(model_url, path_model, hf_token)) {
+    if (!common_download_file(model_url, local_path, hf_token)) {
         return NULL;
     }
 
@@ -1363,9 +1373,9 @@ struct llama_model * common_load_model_from_url(
             /*.no_alloc = */ true,
             /*.ctx      = */ NULL,
         };
-        auto * ctx_gguf = gguf_init_from_file(path_model, gguf_params);
+        auto * ctx_gguf = gguf_init_from_file(local_path.c_str(), gguf_params);
         if (!ctx_gguf) {
-            LOG_ERR("\n%s:  failed to load input GGUF from %s\n", __func__, path_model);
+            LOG_ERR("\n%s:  failed to load input GGUF from %s\n", __func__, local_path.c_str());
             return NULL;
         }
 
@@ -1384,13 +1394,13 @@ struct llama_model * common_load_model_from_url(
         // Verify the first split file format
         // and extract split URL and PATH prefixes
         {
-            if (!llama_split_prefix(split_prefix, sizeof(split_prefix), path_model, 0, n_split)) {
-                LOG_ERR("\n%s: unexpected model file name: %s n_split=%d\n", __func__, path_model, n_split);
+            if (!llama_split_prefix(split_prefix, sizeof(split_prefix), local_path.c_str(), 0, n_split)) {
+                LOG_ERR("\n%s: unexpected model file name: %s n_split=%d\n", __func__, local_path.c_str(), n_split);
                 return NULL;
             }
 
-            if (!llama_split_prefix(split_url_prefix, sizeof(split_url_prefix), model_url, 0, n_split)) {
-                LOG_ERR("\n%s: unexpected model url: %s n_split=%d\n", __func__, model_url, n_split);
+            if (!llama_split_prefix(split_url_prefix, sizeof(split_url_prefix), model_url.c_str(), 0, n_split)) {
+                LOG_ERR("\n%s: unexpected model url: %s n_split=%d\n", __func__, model_url.c_str(), n_split);
                 return NULL;
             }
         }
@@ -1417,14 +1427,14 @@ struct llama_model * common_load_model_from_url(
         }
     }
 
-    return llama_load_model_from_file(path_model, params);
+    return llama_load_model_from_file(local_path.c_str(), params);
 }
 
 struct llama_model * common_load_model_from_hf(
-        const char * repo,
-        const char * model,
-        const char * path_model,
-        const char * hf_token,
+        const std::string & repo,
+        const std::string & remote_path,
+        const std::string & local_path,
+        const std::string & hf_token,
         const struct llama_model_params & params) {
     // construct hugging face model url:
     //
@@ -1438,27 +1448,27 @@ struct llama_model * common_load_model_from_hf(
     std::string model_url = "https://huggingface.co/";
     model_url += repo;
     model_url += "/resolve/main/";
-    model_url += model;
+    model_url += remote_path;
 
-    return common_load_model_from_url(model_url.c_str(), path_model, hf_token, params);
+    return common_load_model_from_url(model_url, local_path, hf_token, params);
 }
 
 #else
 
 struct llama_model * common_load_model_from_url(
-        const char * /*model_url*/,
-        const char * /*path_model*/,
-        const char * /*hf_token*/,
+        const std::string & /*model_url*/,
+        const std::string & /*local_path*/,
+        const std::string & /*hf_token*/,
         const struct llama_model_params & /*params*/) {
     LOG_WRN("%s: llama.cpp built without libcurl, downloading from an url not supported.\n", __func__);
     return nullptr;
 }
 
 struct llama_model * common_load_model_from_hf(
-        const char * /*repo*/,
-        const char * /*model*/,
-        const char * /*path_model*/,
-        const char * /*hf_token*/,
+        const std::string & /*repo*/,
+        const std::string & /*remote_path*/,
+        const std::string & /*local_path*/,
+        const std::string & /*hf_token*/,
         const struct llama_model_params & /*params*/) {
     LOG_WRN("%s: llama.cpp built without libcurl, downloading from Hugging Face not supported.\n", __func__);
     return nullptr;

common/common.h

@@ -133,6 +133,7 @@ struct common_params_sampling {
     bool    penalize_nl        = false; // consider newlines as a repeatable token
     bool    ignore_eos         = false;
     bool    no_perf            = false; // disable performance metrics
+    bool    timing_per_token   = false;
 
     std::vector<std::string> dry_sequence_breakers = {"\n", ":", "\"", "*"};     // default sequence breakers for DRY
 
@@ -470,8 +471,17 @@ struct llama_model_params     common_model_params_to_llama  (      common_params
 struct llama_context_params   common_context_params_to_llama(const common_params & params);
 struct ggml_threadpool_params ggml_threadpool_params_from_cpu_params(const cpu_params & params);
 
-struct llama_model * common_load_model_from_url(const char * model_url, const char * path_model, const char * hf_token, const struct llama_model_params & params);
-struct llama_model * common_load_model_from_hf(const char * repo, const char * file, const char * path_model, const char * hf_token, const struct llama_model_params & params);
+struct llama_model * common_load_model_from_url(
+    const std::string & model_url,
+    const std::string & local_path,
+    const std::string & hf_token,
+    const struct llama_model_params & params);
+struct llama_model * common_load_model_from_hf(
+    const std::string & repo,
+    const std::string & remote_path,
+    const std::string & local_path,
+    const std::string & hf_token,
+    const struct llama_model_params & params);
 
 // clear LoRA adapters from context, then apply new list of adapters
 void common_lora_adapters_apply(struct llama_context * ctx, std::vector<common_lora_adapter_container> & lora_adapters);

docs/android.md

@@ -23,10 +23,10 @@ $ curl -L {model-url} -o ~/{model}.gguf
 Then, if you are not already in the repo directory, `cd` into `llama.cpp` and:
 
 ```
-$ ./build/bin/llama-simple -m ~/{model}.gguf -c {context-size} -p "{your-prompt}"
+$ ./build/bin/llama-cli -m ~/{model}.gguf -c {context-size} -p "{your-prompt}"
 ```
 
-Here, we show `llama-simple`, but any of the executables under `examples` should work, in theory. Be sure to set `context-size` to a reasonable number (say, 4096) to start with; otherwise, memory could spike and kill your terminal.
+Here, we show `llama-cli`, but any of the executables under `examples` should work, in theory. Be sure to set `context-size` to a reasonable number (say, 4096) to start with; otherwise, memory could spike and kill your terminal.
 
 To see what it might look like visually, here's an old demo of an interactive session running on a Pixel 5 phone:
 

docs/backend/BLIS.md

@@ -27,13 +27,6 @@ We recommend using openmp since it's easier to modify the cores being used.
 
 ### llama.cpp compilation
 
-Makefile:
-
-```bash
-make GGML_BLIS=1 -j
-# make GGML_BLIS=1 llama-benchmark-matmult
-```
-
 CMake:
 
 ```bash

docs/backend/CANN.md

@@ -23,6 +23,8 @@ The llama.cpp CANN backend is designed to support Ascend NPU. It utilize the abi
 
 ## News
 
+- 2024.11
+  - Support F16 and F32 data type model for Ascend 310P NPU.
 - 2024.8
   - Support `Q4_0` and `Q8_0` data type for Ascend NPU.
 - 2024.7
@@ -40,9 +42,11 @@ The llama.cpp CANN backend is designed to support Ascend NPU. It utilize the abi
 ### Ascend NPU
 
 **Verified devices**
+
 | Ascend NPU                    | Status  |
 |:-----------------------------:|:-------:|
 | Atlas 300T A2                 | Support |
+| Atlas 300I Duo                | Support |
 
 *Notes:*
 

docs/build.md

@@ -7,124 +7,68 @@ git clone https://github.com/ggerganov/llama.cpp
 cd llama.cpp
 ```
 
-In order to build llama.cpp you have four different options.
+The following sections describe how to build with different backends and options.
 
-- Using `make`:
-  - On Linux or MacOS:
+## CPU Build
 
-      ```bash
-      make
-      ```
+Build llama.cpp using `CMake`:
 
-  - On Windows (x86/x64 only, arm64 requires cmake):
+```bash
+cmake -B build
+cmake --build build --config Release
+```
 
-    1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases).
-    2. Extract `w64devkit` on your pc.
-    3. Run `w64devkit.exe`.
-    4. Use the `cd` command to reach the `llama.cpp` folder.
-    5. From here you can run:
-        ```bash
-        make
-        ```
+**Notes**:
 
-  - Notes:
-    - For `Q4_0_4_4` quantization type build, add the `GGML_NO_LLAMAFILE=1` flag. For example, use `make GGML_NO_LLAMAFILE=1`.
-    - 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`
+- For faster compilation, add the `-j` argument to run multiple jobs in parallel, or use a generator that does this automatically such as Ninja. 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:
 
-- Using `CMake`:
+    1. Single-config generators (e.g. default = `Unix Makefiles`; note that they just ignore the `--config` flag):
 
-  ```bash
-  cmake -B build
+       ```bash
+       cmake -B build -DCMAKE_BUILD_TYPE=Debug
+       cmake --build build
+       ```
+
+    2. Multi-config generators (`-G` param set to Visual Studio, XCode...):
+
+       ```bash
+       cmake -B build -G "Xcode"
+       cmake --build build --config Debug
+       ```
+
+    For more details and a list of supported generators, see the [CMake documentation](https://cmake.org/cmake/help/latest/manual/cmake-generators.7.html).
+- For static builds, add `-DBUILD_SHARED_LIBS=OFF`:
+  ```
+  cmake -B build -DBUILD_SHARED_LIBS=OFF
   cmake --build build --config Release
   ```
 
-  **Notes**:
-
-    - For `Q4_0_4_4` quantization type build, add the `-DGGML_LLAMAFILE=OFF` cmake option. For example, use `cmake -B build -DGGML_LLAMAFILE=OFF`.
-    - 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
-      ```
-
-      2. Multi-config generators (`-G` param set to Visual Studio, XCode...):
-
-      ```bash
-      cmake -B build -G "Xcode"
-      cmake --build build --config Debug
-      ```
-    - Building for Windows (x86, x64 and arm64) with MSVC or clang as compilers:
-      - Install Visual Studio 2022, e.g. via the [Community Edition](https://visualstudio.microsoft.com/de/vs/community/). In the installer, select at least the following options (this also automatically installs the required additional tools like CMake,...):
-        - Tab Workload: Desktop-development with C++
-        - Tab Components (select quickly via search): C++-_CMake_ Tools for Windows, _Git_ for Windows, C++-_Clang_ Compiler for Windows, MS-Build Support for LLVM-Toolset (clang)
-      - Please remember to always use a Developer Command Prompt / PowerShell for VS2022 for git, build, test
-      - For Windows on ARM (arm64, WoA) build with:
-        ```bash
-        cmake --preset arm64-windows-llvm-release -D GGML_OPENMP=OFF
-        cmake --build build-arm64-windows-llvm-release
-        ```
-        Note: Building for arm64 could also be done just with MSVC (with the build-arm64-windows-MSVC preset, or the standard CMake build instructions). But MSVC does not support inline ARM assembly-code, used e.g. for the accelerated Q4_0_4_8 CPU kernels.
-
--   Using `gmake` (FreeBSD):
-
-    1. Install and activate [DRM in FreeBSD](https://wiki.freebsd.org/Graphics)
-    2. Add your user to **video** group
-    3. Install compilation dependencies.
-
-        ```bash
-        sudo pkg install gmake automake autoconf pkgconf llvm15 openblas
-
-        gmake CC=/usr/local/bin/clang15 CXX=/usr/local/bin/clang++15 -j4
-        ```
-
-## 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 `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.
+- Building for Windows (x86, x64 and arm64) with MSVC or clang as compilers:
+    - Install Visual Studio 2022, e.g. via the [Community Edition](https://visualstudio.microsoft.com/de/vs/community/). In the installer, select at least the following options (this also automatically installs the required additional tools like CMake,...):
+    - Tab Workload: Desktop-development with C++
+    - Tab Components (select quickly via search): C++-_CMake_ Tools for Windows, _Git_ for Windows, C++-_Clang_ Compiler for Windows, MS-Build Support for LLVM-Toolset (clang)
+    - Please remember to always use a Developer Command Prompt / PowerShell for VS2022 for git, build, test
+    - For Windows on ARM (arm64, WoA) build with:
+    ```bash
+    cmake --preset arm64-windows-llvm-release -D GGML_OPENMP=OFF
+    cmake --build build-arm64-windows-llvm-release
+    ```
+    Building for arm64 can also be done with the MSVC compiler with the build-arm64-windows-MSVC preset, or the standard CMake build instructions. However, note that the MSVC compiler does not support inline ARM assembly code, used e.g. for the accelerated Q4_0_4_8 CPU kernels.
 
 ## 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. 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). Using BLAS doesn't affect the generation performance. There are currently several different BLAS implementations available for build and use:
 
-### Accelerate Framework:
+### Accelerate Framework
 
 This is only available on Mac PCs and it's enabled by default. You can just build using the normal instructions.
 
-### OpenBLAS:
+### OpenBLAS
 
 This provides BLAS acceleration using only the CPU. Make sure to have OpenBLAS installed on your machine.
 
-- Using `make`:
-  - On Linux:
-    ```bash
-    make GGML_OPENBLAS=1
-    ```
-
-  - On Windows:
-
-    1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases).
-    2. Download the latest version of [OpenBLAS for Windows](https://github.com/xianyi/OpenBLAS/releases).
-    3. Extract `w64devkit` on your pc.
-    4. From the OpenBLAS zip that you just downloaded copy `libopenblas.a`, located inside the `lib` folder, inside `w64devkit\x86_64-w64-mingw32\lib`.
-    5. From the same OpenBLAS zip copy the content of the `include` folder inside `w64devkit\x86_64-w64-mingw32\include`.
-    6. Run `w64devkit.exe`.
-    7. Use the `cd` command to reach the `llama.cpp` folder.
-    8. From here you can run:
-
-        ```bash
-        make GGML_OPENBLAS=1
-        ```
-
 - Using `CMake` on Linux:
 
     ```bash
@@ -136,14 +80,6 @@ This provides BLAS acceleration using only the CPU. Make sure to have OpenBLAS i
 
 Check [BLIS.md](./backend/BLIS.md) for more information.
 
-### SYCL
-
-SYCL is a higher-level programming model to improve programming productivity on various hardware accelerators.
-
-llama.cpp based on SYCL is used to **support Intel GPU** (Data Center Max series, Flex series, Arc series, Built-in GPU and iGPU).
-
-For detailed info, please refer to [llama.cpp for SYCL](./backend/SYCL.md).
-
 ### Intel oneMKL
 
 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](./backend/SYCL.md).
@@ -161,16 +97,29 @@ Building through oneAPI compilers will make avx_vnni instruction set available f
 
 Check [Optimizing and Running LLaMA2 on Intel® CPU](https://www.intel.com/content/www/us/en/content-details/791610/optimizing-and-running-llama2-on-intel-cpu.html) for more information.
 
-### CUDA
+### Other BLAS libraries
 
-This provides GPU acceleration using the CUDA cores of your Nvidia GPU. Make sure to have the CUDA toolkit installed. You can download it from your Linux distro's package manager (e.g. `apt install nvidia-cuda-toolkit`) or from here: [CUDA Toolkit](https://developer.nvidia.com/cuda-downloads).
+Any other BLAS library can be used by setting the `GGML_BLAS_VENDOR` option. See the [CMake documentation](https://cmake.org/cmake/help/latest/module/FindBLAS.html#blas-lapack-vendors) for a list of supported vendors.
 
-For Jetson user, if you have Jetson Orin, you can try this: [Offical Support](https://www.jetson-ai-lab.com/tutorial_text-generation.html). If you are using an old model(nano/TX2), need some additional operations before compiling.
+## 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 `-DGGML_METAL=OFF` cmake option.
+
+When built with Metal support, you can explicitly disable GPU inference with the `--n-gpu-layers 0` command-line argument.
+
+## SYCL
+
+SYCL is a higher-level programming model to improve programming productivity on various hardware accelerators.
+
+llama.cpp based on SYCL is used to **support Intel GPU** (Data Center Max series, Flex series, Arc series, Built-in GPU and iGPU).
+
+For detailed info, please refer to [llama.cpp for SYCL](./backend/SYCL.md).
+
+## CUDA
+
+This provides GPU acceleration using an NVIDIA GPU. Make sure to have the CUDA toolkit installed. You can download it from your Linux distro's package manager (e.g. `apt install nvidia-cuda-toolkit`) or from the [NVIDIA developer site](https://developer.nvidia.com/cuda-downloads).
 
-- Using `make`:
-  ```bash
-  make GGML_CUDA=1
-  ```
 - Using `CMake`:
 
   ```bash
@@ -192,14 +141,10 @@ The following compilation options are also available to tweak performance:
 | 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.                                                                                                  |
 
-### MUSA
+## MUSA
 
 This provides GPU acceleration using the MUSA cores of your Moore Threads MTT GPU. Make sure to have the MUSA SDK installed. You can download it from here: [MUSA SDK](https://developer.mthreads.com/sdk/download/musa).
 
-- Using `make`:
-  ```bash
-  make GGML_MUSA=1
-  ```
 - Using `CMake`:
 
   ```bash
@@ -213,16 +158,12 @@ The environment variable `GGML_CUDA_ENABLE_UNIFIED_MEMORY=1` can be used to enab
 
 Most of the compilation options available for CUDA should also be available for MUSA, though they haven't been thoroughly tested yet.
 
-### hipBLAS
+## HIP
 
-This provides BLAS acceleration on HIP-supported AMD GPUs.
+This provides GPU acceleration on HIP-supported AMD GPUs.
 Make sure to have ROCm installed.
 You can download it from your Linux distro's package manager or from here: [ROCm Quick Start (Linux)](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/tutorial/quick-start.html#rocm-install-quick).
 
-- Using `make`:
-  ```bash
-  make GGML_HIPBLAS=1
-  ```
 - Using `CMake` for Linux (assuming a gfx1030-compatible AMD GPU):
   ```bash
   HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \
@@ -247,11 +188,6 @@ You can download it from your Linux distro's package manager or from here: [ROCm
       && cmake --build build -- -j 16
   ```
 
-- Using `make` (example for target gfx1030, build with 16 CPU threads):
-  ```bash
-  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%
@@ -265,11 +201,11 @@ You can download it from your Linux distro's package manager or from here: [ROCm
 The environment variable [`HIP_VISIBLE_DEVICES`](https://rocm.docs.amd.com/en/latest/understand/gpu_isolation.html#hip-visible-devices) can be used to specify which GPU(s) will be used.
 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.
 
-### Vulkan
+## Vulkan
 
 **Windows**
 
-#### w64devkit
+### w64devkit
 
 Download and extract [`w64devkit`](https://github.com/skeeto/w64devkit/releases).
 
@@ -289,9 +225,14 @@ Libs: -lvulkan-1
 EOF
 
 ```
-Switch into the `llama.cpp` directory and run `make GGML_VULKAN=1`.
 
-#### Git Bash MINGW64
+Switch into the `llama.cpp` directory and build using CMake.
+```sh
+cmake -B build -DGGML_VULKAN=ON
+cmake --build build --config Release
+```
+
+### Git Bash MINGW64
 
 Download and install [`Git-SCM`](https://git-scm.com/downloads/win) with the default settings
 
@@ -310,20 +251,21 @@ cmake --build build --config Release
 
 Now you can load the model in conversation mode using `Vulkan`
 
-```
-build/bin/release/llama-cli -m "[PATH TO MODEL]" -ngl 100 -c 16384 -t 10 -n -2 -cnv
+```sh
+build/bin/Release/llama-cli -m "[PATH TO MODEL]" -ngl 100 -c 16384 -t 10 -n -2 -cnv
 ```
 
-#### MSYS2
+### MSYS2
 Install [MSYS2](https://www.msys2.org/) and then run the following commands in a UCRT terminal to install dependencies.
-  ```sh
-  pacman -S git \
-      mingw-w64-ucrt-x86_64-gcc \
-      mingw-w64-ucrt-x86_64-cmake \
-      mingw-w64-ucrt-x86_64-vulkan-devel \
-      mingw-w64-ucrt-x86_64-shaderc
-  ```
-Switch into `llama.cpp` directory and build using CMake.
+```sh
+pacman -S git \
+    mingw-w64-ucrt-x86_64-gcc \
+    mingw-w64-ucrt-x86_64-cmake \
+    mingw-w64-ucrt-x86_64-vulkan-devel \
+    mingw-w64-ucrt-x86_64-shaderc
+```
+
+Switch into the `llama.cpp` directory and build using CMake.
 ```sh
 cmake -B build -DGGML_VULKAN=ON
 cmake --build build --config Release
@@ -372,7 +314,7 @@ cmake --build build --config Release
 # ggml_vulkan: Using Intel(R) Graphics (ADL GT2) | uma: 1 | fp16: 1 | warp size: 32
 ```
 
-### CANN
+## CANN
 This provides NPU acceleration using the AI cores of your Ascend NPU. And [CANN](https://www.hiascend.com/en/software/cann) is a hierarchical APIs to help you to quickly build AI applications and service based on Ascend NPU.
 
 For more information about Ascend NPU in [Ascend Community](https://www.hiascend.com/en/).
@@ -387,22 +329,26 @@ cmake --build build --config release
 
 You can test with:
 
-`./build/bin/llama-cli -m PATH_TO_MODEL -p "Building a website can be done in 10 steps:" -ngl 32`
-
-If the fllowing info is output on screen, you are using `llama.cpp by CANN backend`:
 ```bash
-llm_load_tensors:       CANN buffer size = 13313.00 MiB
+./build/bin/llama-cli -m PATH_TO_MODEL -p "Building a website can be done in 10 steps:" -ngl 32
+```
+
+If the following info is output on screen, you are using `llama.cpp` with the CANN backend:
+```bash
+llm_load_tensors:       CANN model buffer size = 13313.00 MiB
 llama_new_context_with_model:       CANN compute buffer size =  1260.81 MiB
 ```
 
 For detailed info, such as model/device supports, CANN install, please refer to [llama.cpp for CANN](./backend/CANN.md).
 
-### Android
+## Android
 
 To read documentation for how to build on Android, [click here](./android.md)
 
-### Arm CPU optimized mulmat kernels
+## Notes about GPU-accelerated backends
 
-Llama.cpp includes a set of optimized mulmat kernels for the Arm architecture, leveraging Arm® Neon™, int8mm and SVE instructions. These kernels are enabled at build time through the appropriate compiler cpu-type flags, such as `-DCMAKE_C_FLAGS=-march=armv8.2a+i8mm+sve`. Note that these optimized kernels require the model to be quantized into one of the formats: `Q4_0_4_4` (Arm Neon), `Q4_0_4_8` (int8mm) or `Q4_0_8_8` (SVE). The SVE mulmat kernel specifically requires a vector width of 256 bits. When running on devices with a different vector width, it is recommended to use the `Q4_0_4_8` (int8mm) or `Q4_0_4_4` (Arm Neon) formats for better performance. Refer to [examples/quantize/README.md](../examples/quantize/README.md) for more information on the quantization formats.
+The GPU may still be used to accelerate some parts of the computation even when using the `-ngl 0` option. You can fully disable GPU acceleration by using `--device none`.
 
-To support `Q4_0_4_4`, you must build with `GGML_NO_LLAMAFILE=1` (`make`) or `-DGGML_LLAMAFILE=OFF` (`cmake`).
+In most cases, it is possible to build and use multiple backends at the same time. For example, you can build llama.cpp with both CUDA and Vulkan support by using the `-DGGML_CUDA=ON -DGGML_VULKAN=ON` options with CMake. At runtime, you can specify which backend devices to use with the `--device` option. To see a list of available devices, use the `--list-devices` option.
+
+Backends can be built as dynamic libraries that can be loaded dynamically at runtime. This allows you to use the same llama.cpp binary on different machines with different GPUs. To enable this feature, use the `GGML_BACKEND_DL` option when building.

examples/CMakeLists.txt

@@ -6,6 +6,10 @@ find_package(Threads REQUIRED)
 
 # ...
 
+# flags
+
+llama_add_compile_flags()
+
 # examples
 
 include_directories(${CMAKE_CURRENT_SOURCE_DIR})

examples/base-translate.sh

@@ -1,61 +0,0 @@
-#!/bin/bash
-#
-# Few-shot translation example.
-# Requires a base model (i.e. no fine-tuned or instruct models).
-#
-# Usage:
-#
-#   cd llama.cpp
-#   make -j
-#
-#   ./examples/base-translate.sh <model-base> "<text>" [extra-main-args]
-#
-
-if [ $# -lt 2 ]; then
-  echo "Usage: ./base-translate.sh <model-base> \"<text>\" [extra-main-args]"
-  exit 1
-fi
-
-eargs=""
-if [ $# -gt 2 ]; then
-  eargs="${@:3}"
-fi
-
-ftmp="__llama.cpp_example_tmp__.txt"
-trap "rm -f $ftmp" EXIT
-
-echo "Translate from English to French:
-
-===
-
-sea otter, peppermint, plush girafe:
-
-sea otter => loutre de mer
-peppermint => menthe poivrée
-plush girafe => girafe peluche
-
-===
-
-violin
-
-violin => violon
-
-===
-
-phone, computer, mouse, keyboard:
-
-phone => téléphone
-computer => ordinateur
-mouse => souris
-keyboard => clavier
-
-===
-" > $ftmp
-
-echo "$2
-" >> $ftmp
-
-model=$1
-
-# generate the most likely continuation until the string "===" is found
-./llama-cli -m $model -f $ftmp -n 64 --temp 0 --repeat-penalty 1.0 --no-penalize-nl -r "===" $eargs

examples/batched-bench/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-batched-bench)
 add_executable(${TARGET} batched-bench.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/batched/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-batched)
 add_executable(${TARGET} batched.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

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

@@ -2,4 +2,4 @@ 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 ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

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

@@ -2,11 +2,8 @@
 
 This example reads weights from project [llama2.c](https://github.com/karpathy/llama2.c) and saves them in ggml compatible format. The vocab that is available in `models/ggml-vocab.bin` is used by default.
 
-To convert the model first download the models from the [llama2.c](https://github.com/karpathy/llama2.c) repository:
+To convert the model first download the models from the [llama2.c](https://github.com/karpathy/llama2.c) repository.
 
-`$ make -j`
-
-After successful compilation, following usage options are available:
 ```
 usage: ./llama-convert-llama2c-to-ggml [options]
 

examples/cvector-generator/CMakeLists.txt

@@ -2,4 +2,4 @@ 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)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/embedding/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-embedding)
 add_executable(${TARGET} embedding.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/eval-callback/CMakeLists.txt

@@ -2,7 +2,7 @@ set(TARGET llama-eval-callback)
 add_executable(${TARGET} eval-callback.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)
 
 set(TEST_TARGET test-eval-callback)
 add_test(NAME ${TEST_TARGET}

examples/export-lora/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-export-lora)
 add_executable(${TARGET} export-lora.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/gbnf-validator/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-gbnf-validator)
 add_executable(${TARGET} gbnf-validator.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/gen-docs/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-gen-docs)
 add_executable(${TARGET} gen-docs.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/gguf-hash/CMakeLists.txt

@@ -4,12 +4,19 @@ install(TARGETS ${TARGET} RUNTIME)
 
 # clibs dependencies
 include_directories(deps/)
+
 add_library(xxhash OBJECT deps/xxhash/xxhash.c deps/xxhash/xxhash.h)
 target_link_libraries(${TARGET} PRIVATE xxhash)
+
 add_library(sha1 OBJECT deps/sha1/sha1.c deps/sha1/sha1.h)
 target_link_libraries(${TARGET} PRIVATE sha1)
+if (NOT MSVC)
+    # disable warnings in 3rd party code
+    target_compile_options(sha1 PRIVATE -w)
+endif()
+
 add_library(sha256 OBJECT deps/sha256/sha256.c deps/sha256/sha256.h)
 target_link_libraries(${TARGET} PRIVATE sha256)
 
 target_link_libraries(${TARGET} PRIVATE ggml ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/gguf-split/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-gguf-split)
 add_executable(${TARGET} gguf-split.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/gguf/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-gguf)
 add_executable(${TARGET} gguf.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE ggml ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/gritlm/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-gritlm)
 add_executable(${TARGET} gritlm.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/imatrix/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-imatrix)
 add_executable(${TARGET} imatrix.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/imatrix/README.md

@@ -25,8 +25,6 @@ For faster computation, make sure to use GPU offloading via the `-ngl` argument
 ## Example
 
 ```bash
-GGML_CUDA=1 make -j
-
 # generate importance matrix (imatrix.dat)
 ./llama-imatrix -m ggml-model-f16.gguf -f train-data.txt -ngl 99
 

examples/imatrix/imatrix.cpp

@@ -637,10 +637,19 @@ int main(int argc, char ** argv) {
         LOG_INF("%s\n", common_params_get_system_info(params).c_str());
     }
 
-    if (!compute_imatrix(ctx, params)) {
-        return 1;
+    if (params.prompt.empty()) {
+        if (params.in_files.empty()) {
+            LOG_ERR("Error: No prompt provided and no precomputed matrices (--in-file) to combine.\n");
+            return 1;
+        }
+        LOG_INF("No prompt provided; combining precomputed matrices only.\n");
+    } else {
+        if (!compute_imatrix(ctx, params)) {
+            return 1;
+        }
     }
 
+
     g_collector.save_imatrix();
 
     LOG("\n");

examples/infill/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-infill)
 add_executable(${TARGET} infill.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/infill/README.md

@@ -14,7 +14,7 @@ In this section, we cover the most commonly used options for running the `infill
 -   `-m FNAME, --model FNAME`: Specify the path to the LLaMA model file (e.g., `models/7B/ggml-model.bin`).
 -   `-i, --interactive`: Run the program in interactive mode, allowing you to provide input directly and receive real-time responses.
 -   `-n N, --n-predict N`: Set the number of tokens to predict when generating text. Adjusting this value can influence the length of the generated text.
--   `-c N, --ctx-size N`: Set the size of the prompt context. The default is 512, but LLaMA models were built with a context of 2048, which will provide better results for longer input/inference.
+-   `-c N, --ctx-size N`: Set the size of the prompt context. The default is 4096, but if a LLaMA model was built with a longer context, increasing this value will provide better results for longer input/inference.
 -   `--spm-infill`: Use Suffix/Prefix/Middle pattern for infill (instead of Prefix/Suffix/Middle) as some models prefer this.
 
 ## Input Prompts

examples/llama-bench/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-bench)
 add_executable(${TARGET} llama-bench.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/llava/CMakeLists.txt

@@ -11,7 +11,7 @@ target_include_directories(llava PUBLIC .)
 target_include_directories(llava PUBLIC ../..)
 target_include_directories(llava PUBLIC ../../common)
 
-target_compile_features(llava PRIVATE cxx_std_11)
+target_compile_features(llava PRIVATE cxx_std_17)
 
 add_library(llava_static STATIC $<TARGET_OBJECTS:llava>)
 if (BUILD_SHARED_LIBS)
@@ -35,11 +35,11 @@ add_executable(${TARGET} llava-cli.cpp)
 set_target_properties(${TARGET} PROPERTIES OUTPUT_NAME llama-llava-cli)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llava ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)
 
 set(TARGET llama-minicpmv-cli)
 add_executable(${TARGET} minicpmv-cli.cpp)
 set_target_properties(${TARGET} PROPERTIES OUTPUT_NAME llama-minicpmv-cli)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llava ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/llava/clip.cpp

@@ -12,6 +12,10 @@
 #include "ggml-cuda.h"
 #endif
 
+#ifdef GGML_USE_SYCL
+#include "ggml-sycl.h"
+#endif
+
 #ifdef GGML_USE_METAL
 #include "ggml-metal.h"
 #endif
@@ -40,10 +44,17 @@
 #include <cinttypes>
 #include <limits>
 
-#define LOG_INF(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
-#define LOG_WRN(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
-#define LOG_ERR(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
-#define LOG_DBG(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
+#if defined(LLAVA_LOG_OFF)
+#   define LOG_INF(...)
+#   define LOG_WRN(...)
+#   define LOG_ERR(...)
+#   define LOG_DBG(...)
+#else // defined(LLAVA_LOG_OFF)
+#   define LOG_INF(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
+#   define LOG_WRN(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
+#   define LOG_ERR(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
+#   define LOG_DBG(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
+#endif // defined(LLAVA_LOG_OFF)
 
 //#define CLIP_DEBUG_FUNCTIONS
 
@@ -1162,6 +1173,11 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
     LOG_INF("%s: CLIP using Vulkan backend\n", __func__);
 #endif
 
+#ifdef GGML_USE_SYCL
+    new_clip->backend = ggml_backend_sycl_init(0);
+    LOG_INF("%s: CLIP using SYCL backend\n", __func__);
+#endif
+
     if (!new_clip->backend) {
         new_clip->backend = ggml_backend_cpu_init();
         LOG_INF("%s: CLIP using CPU backend\n", __func__);

examples/llava/llava.cpp

@@ -11,13 +11,17 @@
 #include <limits>
 #include <vector>
 
-#define die(msg)          do { fputs("error: " msg "\n", stderr);                exit(1); } while (0)
-#define die_fmt(fmt, ...) do { fprintf(stderr, "error: " fmt "\n", __VA_ARGS__); exit(1); } while (0)
-
-#define LOG_INF(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
-#define LOG_WRN(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
-#define LOG_ERR(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
-#define LOG_DBG(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
+#if defined(LLAVA_LOG_OFF)
+#   define LOG_INF(...)
+#   define LOG_WRN(...)
+#   define LOG_ERR(...)
+#   define LOG_DBG(...)
+#else // defined(LLAVA_LOG_OFF)
+#   define LOG_INF(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
+#   define LOG_WRN(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
+#   define LOG_ERR(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
+#   define LOG_DBG(...) do { fprintf(stdout, __VA_ARGS__); } while (0)
+#endif // defined(LLAVA_LOG_OFF)
 
 // RGB uint8 image
 struct clip_image_u8 {
@@ -498,10 +502,16 @@ static bool load_file_to_bytes(const char* path, unsigned char** bytesOut, long
     errno = 0;
     size_t ret = fread(buffer, 1, fileSize, file); // Read the file into the buffer
     if (ferror(file)) {
-        die_fmt("read error: %s", strerror(errno));
+        LOG_ERR("read error: %s", strerror(errno));
+        free(buffer);
+        fclose(file);
+        return false;
     }
     if (ret != (size_t) fileSize) {
-        die("unexpectedly reached end of file");
+        LOG_ERR("unexpectedly reached end of file");
+        free(buffer);
+        fclose(file);
+        return false;
     }
     fclose(file); // Close the file
 

examples/lookahead/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-lookahead)
 add_executable(${TARGET} lookahead.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/lookup/CMakeLists.txt

@@ -2,22 +2,22 @@ set(TARGET llama-lookup)
 add_executable(${TARGET} lookup.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)
 
 set(TARGET llama-lookup-create)
 add_executable(${TARGET} lookup-create.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)
 
 set(TARGET llama-lookup-merge)
 add_executable(${TARGET} lookup-merge.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)
 
 set(TARGET llama-lookup-stats)
 add_executable(${TARGET} lookup-stats.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/main-cmake-pkg/CMakeLists.txt

@@ -29,4 +29,4 @@ add_executable(${TARGET} ${CMAKE_CURRENT_LIST_DIR}/../main/main.cpp)
 target_include_directories(${TARGET} PRIVATE ${_common_path})
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/main/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-cli)
 add_executable(${TARGET} main.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/main/README.md

@@ -66,7 +66,7 @@ In this section, we cover the most commonly used options for running the `llama-
 -   `-mu MODEL_URL --model-url MODEL_URL`: Specify a remote http url to download the file (e.g [https://huggingface.co/ggml-org/gemma-1.1-7b-it-Q4_K_M-GGUF/resolve/main/gemma-1.1-7b-it.Q4_K_M.gguf?download=true](https://huggingface.co/ggml-org/gemma-1.1-7b-it-Q4_K_M-GGUF/resolve/main/gemma-1.1-7b-it.Q4_K_M.gguf?download=true)).
 -   `-i, --interactive`: Run the program in interactive mode, allowing you to provide input directly and receive real-time responses.
 -   `-n N, --n-predict N`: Set the number of tokens to predict when generating text. Adjusting this value can influence the length of the generated text.
--   `-c N, --ctx-size N`: Set the size of the prompt context. The default is 512, but LLaMA models were built with a context of 2048, which will provide better results for longer input/inference.
+-   `-c N, --ctx-size N`: Set the size of the prompt context. The default is 4096, but if a LLaMA model was built with a longer context, increasing this value will provide better results for longer input/inference.
 -   `-mli, --multiline-input`: Allows you to write or paste multiple lines without ending each in '\'
 -   `-t N, --threads N`: Set the number of threads to use during generation. For optimal performance, it is recommended to set this value to the number of physical CPU cores your system has.
 -   `-ngl N, --n-gpu-layers N`: When compiled with GPU support, this option allows offloading some layers to the GPU for computation. Generally results in increased performance.
@@ -131,7 +131,7 @@ During text generation, LLaMA models have a limited context size, which means th
 
 ### Context Size
 
-- `-c N, --ctx-size N`: Set the size of the prompt context (default: 0, 0 = loaded from model). The LLaMA models were built with a context of 2048-8192, which will yield the best results on longer input/inference.
+- `-c N, --ctx-size N`: Set the size of the prompt context (default: 4096, 0 = loaded from model). If a LLaMA model was built with a longer context, increasing this value will yield the best results on longer input/inference.
 
 ### Extended Context Size
 
@@ -348,6 +348,7 @@ These options provide extra functionality and customization when running the LLa
 
 -   `-h, --help`: Display a help message showing all available options and their default values. This is particularly useful for checking the latest options and default values, as they can change frequently, and the information in this document may become outdated.
 -   `--verbose-prompt`: Print the prompt before generating text.
+-   `--no-display-prompt`: Don't print prompt at generation.
 -   `-mg i, --main-gpu i`: When using multiple GPUs this option controls which GPU is used for small tensors for which the overhead of splitting the computation across all GPUs is not worthwhile. The GPU in question will use slightly more VRAM to store a scratch buffer for temporary results. By default GPU 0 is used.
 -   `-ts SPLIT, --tensor-split SPLIT`: When using multiple GPUs this option controls how large tensors should be split across all GPUs. `SPLIT` is a comma-separated list of non-negative values that assigns the proportion of data that each GPU should get in order. For example, "3,2" will assign 60% of the data to GPU 0 and 40% to GPU 1. By default the data is split in proportion to VRAM but this may not be optimal for performance.
 -   `-hfr URL --hf-repo URL`: The url to the Hugging Face model repository. Used in conjunction with `--hf-file` or `-hff`. The model is downloaded and stored in the file provided by `-m` or `--model`. If `-m` is not provided, the model is auto-stored in the path specified by the `LLAMA_CACHE` environment variable  or in an OS-specific local cache.

examples/parallel/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-parallel)
 add_executable(${TARGET} parallel.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/passkey/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-passkey)
 add_executable(${TARGET} passkey.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/perplexity/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-perplexity)
 add_executable(${TARGET} perplexity.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/quantize-stats/CMakeLists.txt

@@ -3,4 +3,4 @@ add_executable(${TARGET} quantize-stats.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE llama build_info ${CMAKE_THREAD_LIBS_INIT})
 target_include_directories(${TARGET} PRIVATE ../../common)
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/quantize/CMakeLists.txt

@@ -3,4 +3,4 @@ add_executable(${TARGET} quantize.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
 target_include_directories(${TARGET} PRIVATE ../../common)
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/retrieval/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-retrieval)
 add_executable(${TARGET} retrieval.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/run/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-run)
 add_executable(${TARGET} run.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/save-load-state/CMakeLists.txt

@@ -2,4 +2,4 @@ set(TARGET llama-save-load-state)
 add_executable(${TARGET} save-load-state.cpp)
 install(TARGETS ${TARGET} RUNTIME)
 target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/server/CMakeLists.txt

@@ -16,12 +16,7 @@ set(TARGET_SRCS
 )
 set(PUBLIC_ASSETS
     index.html
-    completion.js
     loading.html
-    deps_daisyui.min.css
-    deps_markdown-it.js
-    deps_tailwindcss.js
-    deps_vue.esm-browser.js
 )
 
 foreach(asset ${PUBLIC_ASSETS})
@@ -33,11 +28,20 @@ foreach(asset ${PUBLIC_ASSETS})
         OUTPUT "${output}"
         COMMAND "${CMAKE_COMMAND}" "-DINPUT=${input}" "-DOUTPUT=${output}" -P "${PROJECT_SOURCE_DIR}/scripts/xxd.cmake"
     )
+    set_source_files_properties(${output} PROPERTIES GENERATED TRUE)
 endforeach()
 
 add_executable(${TARGET} ${TARGET_SRCS})
 install(TARGETS ${TARGET} RUNTIME)
 
+# clean up generated files in pre-build step
+foreach(asset ${PUBLIC_ASSETS})
+    set(output "${CMAKE_CURRENT_BINARY_DIR}/${asset}.hpp")
+    add_custom_command(TARGET ${TARGET} PRE_BUILD
+        COMMAND "${CMAKE_COMMAND}" -E remove -f "${output}"
+    )
+endforeach()
+
 target_link_libraries(${TARGET} PRIVATE common ${CMAKE_THREAD_LIBS_INIT})
 
 if (LLAMA_SERVER_SSL)
@@ -50,4 +54,4 @@ if (WIN32)
     TARGET_LINK_LIBRARIES(${TARGET} PRIVATE ws2_32)
 endif()
 
-target_compile_features(${TARGET} PRIVATE cxx_std_11)
+target_compile_features(${TARGET} PRIVATE cxx_std_17)

examples/server/README.md

@@ -69,6 +69,8 @@ The project is under active development, and we are [looking for feedback and co
 | `--mlock` | force system to keep model in RAM rather than swapping or compressing<br/>(env: LLAMA_ARG_MLOCK) |
 | `--no-mmap` | do not memory-map model (slower load but may reduce pageouts if not using mlock)<br/>(env: LLAMA_ARG_NO_MMAP) |
 | `--numa TYPE` | attempt optimizations that help on some NUMA systems<br/>- distribute: spread execution evenly over all nodes<br/>- isolate: only spawn threads on CPUs on the node that execution started on<br/>- numactl: use the CPU map provided by numactl<br/>if run without this previously, it is recommended to drop the system page cache before using this<br/>see https://github.com/ggerganov/llama.cpp/issues/1437<br/>(env: LLAMA_ARG_NUMA) |
+| `-dev, --device <dev1,dev2,..>` | comma-separated list of devices to use for offloading (none = don't offload)<br/>use --list-devices to see a list of available devices<br/>(env: LLAMA_ARG_DEVICE) |
+| `--list-devices` | print list of available devices and exit |
 | `-ngl, --gpu-layers, --n-gpu-layers N` | number of layers to store in VRAM<br/>(env: LLAMA_ARG_N_GPU_LAYERS) |
 | `-sm, --split-mode {none,layer,row}` | how to split the model across multiple GPUs, one of:<br/>- none: use one GPU only<br/>- layer (default): split layers and KV across GPUs<br/>- row: split rows across GPUs<br/>(env: LLAMA_ARG_SPLIT_MODE) |
 | `-ts, --tensor-split N0,N1,N2,...` | fraction of the model to offload to each GPU, comma-separated list of proportions, e.g. 3,1<br/>(env: LLAMA_ARG_TENSOR_SPLIT) |
@@ -158,9 +160,16 @@ The project is under active development, and we are [looking for feedback and co
 | `--props` | enable changing global properties via POST /props (default: disabled)<br/>(env: LLAMA_ARG_ENDPOINT_PROPS) |
 | `--no-slots` | disables slots monitoring endpoint<br/>(env: LLAMA_ARG_NO_ENDPOINT_SLOTS) |
 | `--slot-save-path PATH` | path to save slot kv cache (default: disabled) |
-| `--chat-template JINJA_TEMPLATE` | set custom jinja chat template (default: template taken from model's metadata)<br/>if suffix/prefix are specified, template will be disabled<br/>only commonly used templates are accepted:<br/>https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template<br/>(env: LLAMA_ARG_CHAT_TEMPLATE) |
+| `--chat-template JINJA_TEMPLATE` | set custom jinja chat template (default: template taken from model's metadata)<br/>if suffix/prefix are specified, template will be disabled<br/>list of built-in templates:<br/>chatglm3, chatglm4, chatml, command-r, deepseek, deepseek2, exaone3, gemma, granite, llama2, llama2-sys, llama2-sys-bos, llama2-sys-strip, llama3, minicpm, mistral-v1, mistral-v3, mistral-v3-tekken, mistral-v7, monarch, openchat, orion, phi3, rwkv-world, vicuna, vicuna-orca, zephyr<br/>(env: LLAMA_ARG_CHAT_TEMPLATE) |
 | `-sps, --slot-prompt-similarity SIMILARITY` | how much the prompt of a request must match the prompt of a slot in order to use that slot (default: 0.50, 0.0 = disabled)<br/> |
 | `--lora-init-without-apply` | load LoRA adapters without applying them (apply later via POST /lora-adapters) (default: disabled) |
+| `--draft-max, --draft, --draft-n N` | number of tokens to draft for speculative decoding (default: 16) |
+| `--draft-min, --draft-n-min N` | minimum number of draft tokens to use for speculative decoding (default: 5) |
+| `--draft-p-min P` | minimum speculative decoding probability (greedy) (default: 0.9) |
+| `-cd, --ctx-size-draft N` | size of the prompt context for the draft model (default: 0, 0 = loaded from model) |
+| `-devd, --device-draft <dev1,dev2,..>` | comma-separated list of devices to use for offloading the draft model (none = don't offload)<br/>use --list-devices to see a list of available devices |
+| `-ngld, --gpu-layers-draft, --n-gpu-layers-draft N` | number of layers to store in VRAM for the draft model |
+| `-md, --model-draft FNAME` | draft model for speculative decoding (default: unused) |
 
 
 Note: If both command line argument and environment variable are both set for the same param, the argument will take precedence over env var.
@@ -188,12 +197,6 @@ services:
 
 `llama-server` is built alongside everything else from the root of the project
 
-- Using `make`:
-
-  ```bash
-  make llama-server
-  ```
-
 - Using `CMake`:
 
   ```bash
@@ -207,15 +210,6 @@ services:
 
 `llama-server` can also be built with SSL support using OpenSSL 3
 
-- Using `make`:
-
-  ```bash
-  # NOTE: For non-system openssl, use the following:
-  #   CXXFLAGS="-I /path/to/openssl/include"
-  #   LDFLAGS="-L /path/to/openssl/lib"
-  make LLAMA_SERVER_SSL=true llama-server
-  ```
-
 - Using `CMake`:
 
   ```bash
@@ -223,6 +217,37 @@ services:
   cmake --build build --config Release -t llama-server
   ```
 
+## Web UI
+
+The project includes a web-based user interface that enables interaction with the model through the `/chat/completions` endpoint.
+
+The web UI is developed using:
+- `vue` framework for frontend development
+- `tailwindcss` and `daisyui` for styling
+- `vite` for build tooling
+
+A pre-built version is available as a single HTML file under `/public` directory.
+
+To build or to run the dev server (with hot reload):
+
+```sh
+# make sure you have nodejs installed
+cd examples/server/webui
+npm i
+
+# to run the dev server
+npm run dev
+
+# to build the public/index.html
+npm run build
+```
+
+NOTE: if you are using the vite dev server, you can change the API base URL to llama.cpp. To do that, run this code snippet in browser's console:
+
+```js
+localStorage.setItem('base', 'http://localhost:8080')
+```
+
 ## Quick Start
 
 To get started right away, run the following command, making sure to use the correct path for the model you have:
@@ -317,104 +342,106 @@ node index.js
 
 ### POST `/completion`: Given a `prompt`, it returns the predicted completion.
 
-    *Options:*
+*Options:*
 
-    `prompt`: Provide the prompt for this completion as a string or as an array of strings or numbers representing tokens. Internally, if `cache_prompt` is `true`, the prompt is compared to the previous completion and only the "unseen" suffix is evaluated. A `BOS` token is inserted at the start, if all of the following conditions are true:
+`prompt`: Provide the prompt for this completion as a string or as an array of strings or numbers representing tokens. Internally, if `cache_prompt` is `true`, the prompt is compared to the previous completion and only the "unseen" suffix is evaluated. A `BOS` token is inserted at the start, if all of the following conditions are true:
 
-      - The prompt is a string or an array with the first element given as a string
-      - The model's `tokenizer.ggml.add_bos_token` metadata is `true`
+  - The prompt is a string or an array with the first element given as a string
+  - The model's `tokenizer.ggml.add_bos_token` metadata is `true`
 
-    These input shapes and data type are allowed for `prompt`:
+These input shapes and data type are allowed for `prompt`:
 
-      - Single string: `"string"`
-      - Single sequence of tokens: `[12, 34, 56]`
-      - Mixed tokens and strings: `[12, 34, "string", 56, 78]`
+  - Single string: `"string"`
+  - Single sequence of tokens: `[12, 34, 56]`
+  - Mixed tokens and strings: `[12, 34, "string", 56, 78]`
 
-    Multiple prompts are also supported. In this case, the completion result will be an array.
+Multiple prompts are also supported. In this case, the completion result will be an array.
 
-      - Only strings: `["string1", "string2"]`
-      - Strings and sequences of tokens: `["string1", [12, 34, 56]]`
-      - Mixed types: `[[12, 34, "string", 56, 78], [12, 34, 56], "string"]`
+  - Only strings: `["string1", "string2"]`
+  - Strings and sequences of tokens: `["string1", [12, 34, 56]]`
+  - Mixed types: `[[12, 34, "string", 56, 78], [12, 34, 56], "string"]`
 
-    `temperature`: Adjust the randomness of the generated text. Default: `0.8`
+`temperature`: Adjust the randomness of the generated text. Default: `0.8`
 
-    `dynatemp_range`: Dynamic temperature range. The final temperature will be in the range of `[temperature - dynatemp_range; temperature + dynatemp_range]` Default: `0.0`, which is disabled.
+`dynatemp_range`: Dynamic temperature range. The final temperature will be in the range of `[temperature - dynatemp_range; temperature + dynatemp_range]` Default: `0.0`, which is disabled.
 
-    `dynatemp_exponent`: Dynamic temperature exponent. Default: `1.0`
+`dynatemp_exponent`: Dynamic temperature exponent. Default: `1.0`
 
-    `top_k`: Limit the next token selection to the K most probable tokens.  Default: `40`
+`top_k`: Limit the next token selection to the K most probable tokens.  Default: `40`
 
-    `top_p`: Limit the next token selection to a subset of tokens with a cumulative probability above a threshold P. Default: `0.95`
+`top_p`: Limit the next token selection to a subset of tokens with a cumulative probability above a threshold P. Default: `0.95`
 
-    `min_p`: The minimum probability for a token to be considered, relative to the probability of the most likely token. Default: `0.05`
+`min_p`: The minimum probability for a token to be considered, relative to the probability of the most likely token. Default: `0.05`
 
-    `n_predict`: Set the maximum number of tokens to predict when generating text. **Note:** May exceed the set limit slightly if the last token is a partial multibyte character. When 0, no tokens will be generated but the prompt is evaluated into the cache. Default: `-1`, where `-1` is infinity.
+`n_predict`: Set the maximum number of tokens to predict when generating text. **Note:** May exceed the set limit slightly if the last token is a partial multibyte character. When 0, no tokens will be generated but the prompt is evaluated into the cache. Default: `-1`, where `-1` is infinity.
 
-    `n_indent`: Specify the minimum line indentation for the generated text in number of whitespace characters. Useful for code completion tasks. Default: `0`
+`n_indent`: Specify the minimum line indentation for the generated text in number of whitespace characters. Useful for code completion tasks. Default: `0`
 
-    `n_keep`: Specify the number of tokens from the prompt to retain when the context size is exceeded and tokens need to be discarded. The number excludes the BOS token.
-    By default, this value is set to `0`, meaning no tokens are kept. Use `-1` to retain all tokens from the prompt.
+`n_keep`: Specify the number of tokens from the prompt to retain when the context size is exceeded and tokens need to be discarded. The number excludes the BOS token.
+By default, this value is set to `0`, meaning no tokens are kept. Use `-1` to retain all tokens from the prompt.
 
-    `stream`: It allows receiving each predicted token in real-time instead of waiting for the completion to finish. To enable this, set to `true`.
+`stream`: It allows receiving each predicted token in real-time instead of waiting for the completion to finish. To enable this, set to `true`.
 
-    `stop`: Specify a JSON array of stopping strings.
-    These words will not be included in the completion, so make sure to add them to the prompt for the next iteration. Default: `[]`
+`stop`: Specify a JSON array of stopping strings.
+These words will not be included in the completion, so make sure to add them to the prompt for the next iteration. Default: `[]`
 
-    `typical_p`: Enable locally typical sampling with parameter p. Default: `1.0`, which is disabled.
+`typical_p`: Enable locally typical sampling with parameter p. Default: `1.0`, which is disabled.
 
-    `repeat_penalty`: Control the repetition of token sequences in the generated text. Default: `1.1`
+`repeat_penalty`: Control the repetition of token sequences in the generated text. Default: `1.1`
 
-    `repeat_last_n`: Last n tokens to consider for penalizing repetition. Default: `64`, where `0` is disabled and `-1` is ctx-size.
+`repeat_last_n`: Last n tokens to consider for penalizing repetition. Default: `64`, where `0` is disabled and `-1` is ctx-size.
 
-    `penalize_nl`: Penalize newline tokens when applying the repeat penalty. Default: `true`
+`penalize_nl`: Penalize newline tokens when applying the repeat penalty. Default: `true`
 
-    `presence_penalty`: Repeat alpha presence penalty. Default: `0.0`, which is disabled.
+`presence_penalty`: Repeat alpha presence penalty. Default: `0.0`, which is disabled.
 
-    `frequency_penalty`: Repeat alpha frequency penalty. Default: `0.0`, which is disabled.
+`frequency_penalty`: Repeat alpha frequency penalty. Default: `0.0`, which is disabled.
 
-    `dry_multiplier`: Set the DRY (Don't Repeat Yourself) repetition penalty multiplier. Default: `0.0`, which is disabled.
+`dry_multiplier`: Set the DRY (Don't Repeat Yourself) repetition penalty multiplier. Default: `0.0`, which is disabled.
 
-    `dry_base`: Set the DRY repetition penalty base value. Default: `1.75`
+`dry_base`: Set the DRY repetition penalty base value. Default: `1.75`
 
-    `dry_allowed_length`: Tokens that extend repetition beyond this receive exponentially increasing penalty: multiplier * base ^ (length of repeating sequence before token - allowed length). Default: `2`
+`dry_allowed_length`: Tokens that extend repetition beyond this receive exponentially increasing penalty: multiplier * base ^ (length of repeating sequence before token - allowed length). Default: `2`
 
-    `dry_penalty_last_n`: How many tokens to scan for repetitions. Default: `-1`, where `0` is disabled and `-1` is context size.
+`dry_penalty_last_n`: How many tokens to scan for repetitions. Default: `-1`, where `0` is disabled and `-1` is context size.
 
-    `dry_sequence_breakers`: Specify an array of sequence breakers for DRY sampling. Only a JSON array of strings is accepted. Default: `['\n', ':', '"', '*']`
+`dry_sequence_breakers`: Specify an array of sequence breakers for DRY sampling. Only a JSON array of strings is accepted. Default: `['\n', ':', '"', '*']`
 
-    `xtc_probability`: Set the chance for token removal via XTC sampler. Default: `0.0`, which is disabled.
+`xtc_probability`: Set the chance for token removal via XTC sampler. Default: `0.0`, which is disabled.
 
-    `xtc_threshold`: Set a minimum probability threshold for tokens to be removed via XTC sampler. Default: `0.1` (> `0.5` disables XTC)
+`xtc_threshold`: Set a minimum probability threshold for tokens to be removed via XTC sampler. Default: `0.1` (> `0.5` disables XTC)
 
-    `mirostat`: Enable Mirostat sampling, controlling perplexity during text generation. Default: `0`, where `0` is disabled, `1` is Mirostat, and `2` is Mirostat 2.0.
+`mirostat`: Enable Mirostat sampling, controlling perplexity during text generation. Default: `0`, where `0` is disabled, `1` is Mirostat, and `2` is Mirostat 2.0.
 
-    `mirostat_tau`: Set the Mirostat target entropy, parameter tau. Default: `5.0`
+`mirostat_tau`: Set the Mirostat target entropy, parameter tau. Default: `5.0`
 
-    `mirostat_eta`: Set the Mirostat learning rate, parameter eta.  Default: `0.1`
+`mirostat_eta`: Set the Mirostat learning rate, parameter eta.  Default: `0.1`
 
-    `grammar`: Set grammar for grammar-based sampling.  Default: no grammar
+`grammar`: Set grammar for grammar-based sampling.  Default: no grammar
 
-    `json_schema`: Set a JSON schema for grammar-based sampling (e.g. `{"items": {"type": "string"}, "minItems": 10, "maxItems": 100}` of a list of strings, or `{}` for any JSON). See [tests](../../tests/test-json-schema-to-grammar.cpp) for supported features.  Default: no JSON schema.
+`json_schema`: Set a JSON schema for grammar-based sampling (e.g. `{"items": {"type": "string"}, "minItems": 10, "maxItems": 100}` of a list of strings, or `{}` for any JSON). See [tests](../../tests/test-json-schema-to-grammar.cpp) for supported features.  Default: no JSON schema.
 
-    `seed`: Set the random number generator (RNG) seed.  Default: `-1`, which is a random seed.
+`seed`: Set the random number generator (RNG) seed.  Default: `-1`, which is a random seed.
 
-    `ignore_eos`: Ignore end of stream token and continue generating.  Default: `false`
+`ignore_eos`: Ignore end of stream token and continue generating.  Default: `false`
 
-    `logit_bias`: Modify the likelihood of a token appearing in the generated text completion. For example, use `"logit_bias": [[15043,1.0]]` to increase the likelihood of the token 'Hello', or `"logit_bias": [[15043,-1.0]]` to decrease its likelihood. Setting the value to false, `"logit_bias": [[15043,false]]` ensures that the token `Hello` is never produced. The tokens can also be represented as strings, e.g. `[["Hello, World!",-0.5]]` will reduce the likelihood of all the individual tokens that represent the string `Hello, World!`, just like the `presence_penalty` does. Default: `[]`
+`logit_bias`: Modify the likelihood of a token appearing in the generated text completion. For example, use `"logit_bias": [[15043,1.0]]` to increase the likelihood of the token 'Hello', or `"logit_bias": [[15043,-1.0]]` to decrease its likelihood. Setting the value to false, `"logit_bias": [[15043,false]]` ensures that the token `Hello` is never produced. The tokens can also be represented as strings, e.g. `[["Hello, World!",-0.5]]` will reduce the likelihood of all the individual tokens that represent the string `Hello, World!`, just like the `presence_penalty` does. Default: `[]`
 
-    `n_probs`: If greater than 0, the response also contains the probabilities of top N tokens for each generated token given the sampling settings. Note that for temperature < 0 the tokens are sampled greedily but token probabilities are still being calculated via a simple softmax of the logits without considering any other sampler settings. Default: `0`
+`n_probs`: If greater than 0, the response also contains the probabilities of top N tokens for each generated token given the sampling settings. Note that for temperature < 0 the tokens are sampled greedily but token probabilities are still being calculated via a simple softmax of the logits without considering any other sampler settings. Default: `0`
 
-    `min_keep`: If greater than 0, force samplers to return N possible tokens at minimum. Default: `0`
+`min_keep`: If greater than 0, force samplers to return N possible tokens at minimum. Default: `0`
 
-    `t_max_predict_ms`: Set a time limit in milliseconds for the prediction (a.k.a. text-generation) phase. The timeout will trigger if the generation takes more than the specified time (measured since the first token was generated) and if a new-line character has already been generated. Useful for FIM applications. Default: `0`, which is disabled.
+`t_max_predict_ms`: Set a time limit in milliseconds for the prediction (a.k.a. text-generation) phase. The timeout will trigger if the generation takes more than the specified time (measured since the first token was generated) and if a new-line character has already been generated. Useful for FIM applications. Default: `0`, which is disabled.
 
-    `image_data`: An array of objects to hold base64-encoded image `data` and its `id`s to be reference in `prompt`. You can determine the place of the image in the prompt as in the following: `USER:[img-12]Describe the image in detail.\nASSISTANT:`. In this case, `[img-12]` will be replaced by the embeddings of the image with id `12` in the following `image_data` array: `{..., "image_data": [{"data": "<BASE64_STRING>", "id": 12}]}`. Use `image_data` only with multimodal models, e.g., LLaVA.
+`image_data`: An array of objects to hold base64-encoded image `data` and its `id`s to be reference in `prompt`. You can determine the place of the image in the prompt as in the following: `USER:[img-12]Describe the image in detail.\nASSISTANT:`. In this case, `[img-12]` will be replaced by the embeddings of the image with id `12` in the following `image_data` array: `{..., "image_data": [{"data": "<BASE64_STRING>", "id": 12}]}`. Use `image_data` only with multimodal models, e.g., LLaVA.
 
-    `id_slot`: Assign the completion task to an specific slot. If is -1 the task will be assigned to a Idle slot.  Default: `-1`
+`id_slot`: Assign the completion task to an specific slot. If is -1 the task will be assigned to a Idle slot.  Default: `-1`
 
-    `cache_prompt`: Re-use KV cache from a previous request if possible. This way the common prefix does not have to be re-processed, only the suffix that differs between the requests. Because (depending on the backend) the logits are **not** guaranteed to be bit-for-bit identical for different batch sizes (prompt processing vs. token generation) enabling this option can cause nondeterministic results. Default: `true`
+`cache_prompt`: Re-use KV cache from a previous request if possible. This way the common prefix does not have to be re-processed, only the suffix that differs between the requests. Because (depending on the backend) the logits are **not** guaranteed to be bit-for-bit identical for different batch sizes (prompt processing vs. token generation) enabling this option can cause nondeterministic results. Default: `true`
 
-    `samplers`: The order the samplers should be applied in. An array of strings representing sampler type names. If a sampler is not set, it will not be used. If a sampler is specified more than once, it will be applied multiple times. Default: `["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature"]` - these are all the available values.
+`samplers`: The order the samplers should be applied in. An array of strings representing sampler type names. If a sampler is not set, it will not be used. If a sampler is specified more than once, it will be applied multiple times. Default: `["dry", "top_k", "typ_p", "top_p", "min_p", "xtc", "temperature"]` - these are all the available values.
+
+    `timings_per_token`: Include prompt processing and text generation speed information in each response.  Default: `false`
 
 **Response format**
 
@@ -457,13 +484,13 @@ Notice that each `probs` is an array of length `n_probs`.
 
 ### POST `/tokenize`: Tokenize a given text
 
-    *Options:*
+*Options:*
 
-    `content`: (Required) The text to tokenize.
+`content`: (Required) The text to tokenize.
 
-    `add_special`: (Optional) Boolean indicating if special tokens, i.e. `BOS`, should be inserted.  Default: `false`
+`add_special`: (Optional) Boolean indicating if special tokens, i.e. `BOS`, should be inserted.  Default: `false`
 
-    `with_pieces`: (Optional) Boolean indicating whether to return token pieces along with IDs.  Default: `false`
+`with_pieces`: (Optional) Boolean indicating whether to return token pieces along with IDs.  Default: `false`
 
 **Response:**
 
@@ -500,52 +527,52 @@ With input 'á' (utf8 hex: C3 A1) on tinyllama/stories260k
 
 ### POST `/detokenize`: Convert tokens to text
 
-    *Options:*
+*Options:*
 
-    `tokens`: Set the tokens to detokenize.
+`tokens`: Set the tokens to detokenize.
 
 ### POST `/embedding`: Generate embedding of a given text
 
 The same as [the embedding example](../embedding) does.
 
-    *Options:*
+*Options:*
 
-    `content`: Set the text to process.
+`content`: Set the text to process.
 
-    `image_data`: An array of objects to hold base64-encoded image `data` and its `id`s to be reference in `content`. You can determine the place of the image in the content as in the following: `Image: [img-21].\nCaption: This is a picture of a house`. In this case, `[img-21]` will be replaced by the embeddings of the image with id `21` in the following `image_data` array: `{..., "image_data": [{"data": "<BASE64_STRING>", "id": 21}]}`. Use `image_data` only with multimodal models, e.g., LLaVA.
+`image_data`: An array of objects to hold base64-encoded image `data` and its `id`s to be reference in `content`. You can determine the place of the image in the content as in the following: `Image: [img-21].\nCaption: This is a picture of a house`. In this case, `[img-21]` will be replaced by the embeddings of the image with id `21` in the following `image_data` array: `{..., "image_data": [{"data": "<BASE64_STRING>", "id": 21}]}`. Use `image_data` only with multimodal models, e.g., LLaVA.
 
 ### POST `/reranking`: Rerank documents according to a given query
 
 Similar to https://jina.ai/reranker/ but might change in the future.
 Requires a reranker model (such as [bge-reranker-v2-m3](https://huggingface.co/BAAI/bge-reranker-v2-m3)) and the `--embedding --pooling rank` options.
 
-    *Options:*
+*Options:*
 
-    `query`: The query against which the documents will be ranked.
+`query`: The query against which the documents will be ranked.
 
-    `documents`: An array strings representing the documents to be ranked.
+`documents`: An array strings representing the documents to be ranked.
 
-    *Aliases:*
-      - `/rerank`
-      - `/v1/rerank`
-      - `/v1/reranking`
+*Aliases:*
+  - `/rerank`
+  - `/v1/rerank`
+  - `/v1/reranking`
 
-    *Examples:*
+*Examples:*
 
-    ```shell
-    curl http://127.0.0.1:8012/v1/rerank \
-        -H "Content-Type: application/json" \
-        -d '{
-            "model": "some-model",
-                "query": "What is panda?",
-                "top_n": 3,
-                "documents": [
-                    "hi",
-                "it is a bear",
-                "The giant panda (Ailuropoda melanoleuca), sometimes called a panda bear or simply panda, is a bear species endemic to China."
-                ]
-        }' | jq
-    ```
+```shell
+curl http://127.0.0.1:8012/v1/rerank \
+    -H "Content-Type: application/json" \
+    -d '{
+        "model": "some-model",
+            "query": "What is panda?",
+            "top_n": 3,
+            "documents": [
+                "hi",
+            "it is a bear",
+            "The giant panda (Ailuropoda melanoleuca), sometimes called a panda bear or simply panda, is a bear species endemic to China."
+            ]
+    }' | jq
+```
 
 ### POST `/infill`: For code infilling.
 
@@ -611,89 +638,89 @@ To use this endpoint with POST method, you need to start server with `--props`
 
 Given a ChatML-formatted json description in `messages`, it returns the predicted completion. Both synchronous and streaming mode are supported, so scripted and interactive applications work fine. While no strong claims of compatibility with OpenAI API spec is being made, in our experience it suffices to support many apps. Only models with a [supported chat template](https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template) can be used optimally with this endpoint. By default, the ChatML template will be used.
 
-    *Options:*
+*Options:*
 
-    See [OpenAI Chat Completions API documentation](https://platform.openai.com/docs/api-reference/chat). While some OpenAI-specific features such as function calling aren't supported, llama.cpp `/completion`-specific features such as `mirostat` are supported.
+See [OpenAI Chat Completions API documentation](https://platform.openai.com/docs/api-reference/chat). While some OpenAI-specific features such as function calling aren't supported, llama.cpp `/completion`-specific features such as `mirostat` are supported.
 
-    The `response_format` parameter supports both plain JSON output (e.g. `{"type": "json_object"}`) and schema-constrained JSON (e.g. `{"type": "json_object", "schema": {"type": "string", "minLength": 10, "maxLength": 100}}` or `{"type": "json_schema", "schema": {"properties": { "name": { "title": "Name",  "type": "string" }, "date": { "title": "Date",  "type": "string" }, "participants": { "items": {"type: "string" }, "title": "Participants",  "type": "string" } } } }`), similar to other OpenAI-inspired API providers.
+The `response_format` parameter supports both plain JSON output (e.g. `{"type": "json_object"}`) and schema-constrained JSON (e.g. `{"type": "json_object", "schema": {"type": "string", "minLength": 10, "maxLength": 100}}` or `{"type": "json_schema", "schema": {"properties": { "name": { "title": "Name",  "type": "string" }, "date": { "title": "Date",  "type": "string" }, "participants": { "items": {"type: "string" }, "title": "Participants",  "type": "string" } } } }`), similar to other OpenAI-inspired API providers.
 
-    *Examples:*
+*Examples:*
 
-    You can use either Python `openai` library with appropriate checkpoints:
+You can use either Python `openai` library with appropriate checkpoints:
 
-    ```python
-    import openai
+```python
+import openai
 
-    client = openai.OpenAI(
-        base_url="http://localhost:8080/v1", # "http://<Your api-server IP>:port"
-        api_key = "sk-no-key-required"
-    )
+client = openai.OpenAI(
+    base_url="http://localhost:8080/v1", # "http://<Your api-server IP>:port"
+    api_key = "sk-no-key-required"
+)
 
-    completion = client.chat.completions.create(
-    model="gpt-3.5-turbo",
-    messages=[
-        {"role": "system", "content": "You are ChatGPT, an AI assistant. Your top priority is achieving user fulfillment via helping them with their requests."},
-        {"role": "user", "content": "Write a limerick about python exceptions"}
-    ]
-    )
+completion = client.chat.completions.create(
+model="gpt-3.5-turbo",
+messages=[
+    {"role": "system", "content": "You are ChatGPT, an AI assistant. Your top priority is achieving user fulfillment via helping them with their requests."},
+    {"role": "user", "content": "Write a limerick about python exceptions"}
+]
+)
 
-    print(completion.choices[0].message)
-    ```
+print(completion.choices[0].message)
+```
 
-    ... or raw HTTP requests:
+... or raw HTTP requests:
 
-    ```shell
-    curl http://localhost:8080/v1/chat/completions \
-    -H "Content-Type: application/json" \
-    -H "Authorization: Bearer no-key" \
-    -d '{
-    "model": "gpt-3.5-turbo",
-    "messages": [
-    {
-        "role": "system",
-        "content": "You are ChatGPT, an AI assistant. Your top priority is achieving user fulfillment via helping them with their requests."
-    },
-    {
-        "role": "user",
-        "content": "Write a limerick about python exceptions"
-    }
-    ]
-    }'
-    ```
+```shell
+curl http://localhost:8080/v1/chat/completions \
+-H "Content-Type: application/json" \
+-H "Authorization: Bearer no-key" \
+-d '{
+"model": "gpt-3.5-turbo",
+"messages": [
+{
+    "role": "system",
+    "content": "You are ChatGPT, an AI assistant. Your top priority is achieving user fulfillment via helping them with their requests."
+},
+{
+    "role": "user",
+    "content": "Write a limerick about python exceptions"
+}
+]
+}'
+```
 
 ### POST `/v1/embeddings`: OpenAI-compatible embeddings API
 
-    *Options:*
+*Options:*
 
-    See [OpenAI Embeddings API documentation](https://platform.openai.com/docs/api-reference/embeddings).
+See [OpenAI Embeddings API documentation](https://platform.openai.com/docs/api-reference/embeddings).
 
-    *Examples:*
+*Examples:*
 
-  - input as string
+- input as string
 
-    ```shell
-    curl http://localhost:8080/v1/embeddings \
-    -H "Content-Type: application/json" \
-    -H "Authorization: Bearer no-key" \
-    -d '{
-            "input": "hello",
-            "model":"GPT-4",
-            "encoding_format": "float"
-    }'
-    ```
+  ```shell
+  curl http://localhost:8080/v1/embeddings \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer no-key" \
+  -d '{
+          "input": "hello",
+          "model":"GPT-4",
+          "encoding_format": "float"
+  }'
+  ```
 
-  - `input` as string array
+- `input` as string array
 
-    ```shell
-    curl http://localhost:8080/v1/embeddings \
-    -H "Content-Type: application/json" \
-    -H "Authorization: Bearer no-key" \
-    -d '{
-            "input": ["hello", "world"],
-            "model":"GPT-4",
-            "encoding_format": "float"
-    }'
-    ```
+  ```shell
+  curl http://localhost:8080/v1/embeddings \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer no-key" \
+  -d '{
+          "input": ["hello", "world"],
+          "model":"GPT-4",
+          "encoding_format": "float"
+  }'
+  ```
 
 ### GET `/slots`: Returns the current slots processing state
 
@@ -779,9 +806,9 @@ Available metrics:
 
 ### POST `/slots/{id_slot}?action=save`: Save the prompt cache of the specified slot to a file.
 
-    *Options:*
+*Options:*
 
-    `filename`: Name of the file to save the slot's prompt cache. The file will be saved in the directory specified by the `--slot-save-path` server parameter.
+`filename`: Name of the file to save the slot's prompt cache. The file will be saved in the directory specified by the `--slot-save-path` server parameter.
 
 **Response format**
 
@@ -799,9 +826,9 @@ Available metrics:
 
 ### POST `/slots/{id_slot}?action=restore`: Restore the prompt cache of the specified slot from a file.
 
-    *Options:*
+*Options:*
 
-    `filename`: Name of the file to restore the slot's prompt cache from. The file should be located in the directory specified by the `--slot-save-path` server parameter.
+`filename`: Name of the file to restore the slot's prompt cache from. The file should be located in the directory specified by the `--slot-save-path` server parameter.
 
 **Response format**
 

examples/server/deps.sh

@@ -1,25 +0,0 @@
-#!/bin/bash
-# Download and update deps for binary
-
-# get the directory of this script file
-DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
-PUBLIC=$DIR/public
-
-echo "download js bundle files"
-
-# Note for contributors: Always pin to a specific version "maj.min.patch" to avoid breaking the CI
-
-curl -L https://cdn.tailwindcss.com/3.4.14 > $PUBLIC/deps_tailwindcss.js
-echo >> $PUBLIC/deps_tailwindcss.js # add newline
-
-curl -L https://cdnjs.cloudflare.com/ajax/libs/daisyui/4.12.14/styled.min.css > $PUBLIC/deps_daisyui.min.css
-curl -L https://cdnjs.cloudflare.com/ajax/libs/daisyui/4.12.14/themes.min.css >> $PUBLIC/deps_daisyui.min.css
-echo >> $PUBLIC/deps_daisyui.min.css # add newline
-
-curl -L https://unpkg.com/vue@3.5.12/dist/vue.esm-browser.js > $PUBLIC/deps_vue.esm-browser.js
-echo >> $PUBLIC/deps_vue.esm-browser.js # add newline
-
-curl -L https://cdnjs.cloudflare.com/ajax/libs/markdown-it/13.0.2/markdown-it.js > $PUBLIC/deps_markdown-it.js
-echo >> $PUBLIC/deps_markdown-it.js # add newline
-
-ls -lah $PUBLIC

examples/server/server.cpp

@@ -16,12 +16,7 @@
 
 // auto generated files (update with ./deps.sh)
 #include "index.html.hpp"
-#include "completion.js.hpp"
 #include "loading.html.hpp"
-#include "deps_daisyui.min.css.hpp"
-#include "deps_markdown-it.js.hpp"
-#include "deps_tailwindcss.js.hpp"
-#include "deps_vue.esm-browser.js.hpp"
 
 #include <atomic>
 #include <condition_variable>
@@ -103,12 +98,6 @@ struct server_task_result {
     bool error;
 };
 
-struct server_static_file {
-    const unsigned char * data;
-    unsigned int size;
-    const char * mime_type;
-};
-
 struct slot_params {
     bool stream       = true;
     bool cache_prompt = true; // remember the prompt to avoid reprocessing all prompt
@@ -177,6 +166,8 @@ struct server_slot {
     bool stopped_word   = false;
     bool stopped_limit  = false;
 
+    bool timings_per_token = false;
+
     bool oaicompat = false;
 
     std::string oaicompat_model;
@@ -694,8 +685,9 @@ struct server_context {
 
             params_dft.devices      = params_base.speculative.devices;
             params_dft.model        = params_base.speculative.model;
-            params_dft.n_ctx        = params_base.speculative.n_ctx;
+            params_dft.n_ctx        = params_base.speculative.n_ctx == 0 ? params_base.n_ctx / params_base.n_parallel : params_base.speculative.n_ctx;
             params_dft.n_gpu_layers = params_base.speculative.n_gpu_layers;
+            params_dft.n_parallel   = 1;
 
             common_init_result llama_init_dft = common_init_from_params(params_dft);
 
@@ -715,8 +707,14 @@ struct server_context {
                 return false;
             }
 
-            cparams_dft = common_context_params_to_llama(params_base);
-            cparams_dft.n_batch = llama_n_ctx(llama_init_dft.context);
+            const int n_ctx_dft = llama_n_ctx(llama_init_dft.context);
+
+            cparams_dft = common_context_params_to_llama(params_dft);
+            cparams_dft.n_batch = n_ctx_dft;
+
+            // force F16 KV cache for the draft model for extra performance
+            cparams_dft.type_k = GGML_TYPE_F16;
+            cparams_dft.type_v = GGML_TYPE_F16;
 
             // the context is not needed - we will create one for each slot
             llama_free(llama_init_dft.context);
@@ -882,6 +880,8 @@ struct server_context {
             slot.oaicompat_model = "";
         }
 
+        slot.timings_per_token       = json_value(data, "timings_per_token",  false);
+
         slot.params.stream           = json_value(data, "stream",             false);
         slot.params.cache_prompt     = json_value(data, "cache_prompt",       true);
         slot.params.n_predict        = json_value(data, "n_predict",          json_value(data, "max_tokens", defaults.n_predict));
@@ -1279,6 +1279,7 @@ struct server_context {
             {"speculative.n_max",         slot.params.speculative.n_max},
             {"speculative.n_min",         slot.params.speculative.n_min},
             {"speculative.p_min",         slot.params.speculative.p_min},
+            {"timings_per_token",         slot.timings_per_token},
         };
     }
 
@@ -1336,6 +1337,10 @@ struct server_context {
             res.data["model"] = slot.oaicompat_model;
         }
 
+        if (slot.timings_per_token) {
+            res.data["timings"] = slot.get_formated_timings();
+        }
+
         queue_results.send(res);
     }
 
@@ -2267,49 +2272,57 @@ struct server_context {
                     continue; // continue loop of slots
                 }
 
-                llama_token id;
+                llama_token id = common_sampler_sample(slot.smpl, ctx, slot.i_batch - i);
 
-                {
-                    completion_token_output result;
+                slot.i_batch = -1;
 
-                    id = common_sampler_sample(slot.smpl, ctx, slot.i_batch - i);
+                common_sampler_accept(slot.smpl, id, true);
 
-                    slot.i_batch = -1;
+                slot.n_decoded += 1;
 
-                    common_sampler_accept(slot.smpl, id, true);
+                const int64_t t_current = ggml_time_us();
 
-                    slot.n_decoded += 1;
-                    if (slot.n_decoded == 1) {
-                        slot.t_start_generation = ggml_time_us();
-                        slot.t_prompt_processing = (slot.t_start_generation - slot.t_start_process_prompt) / 1e3;
-                        metrics.on_prompt_eval(slot);
-                    }
-
-                    result.tok = id;
-
-                    const auto * cur_p = common_sampler_get_candidates(slot.smpl);
-
-                    for (size_t i = 0; i < (size_t) slot.params.sampling.n_probs; ++i) {
-                        result.probs.push_back({
-                            cur_p->data[i].id,
-                                i >= cur_p->size ? 0.0f : cur_p->data[i].p,
-                        });
-                    }
-
-                    if (!process_token(result, slot)) {
-                        // release slot because of stop condition
-                        slot.release();
-                        slot.print_timings();
-                        send_final_response(slot);
-                        metrics.on_prediction(slot);
-                        continue;
-                    }
+                if (slot.n_decoded == 1) {
+                    slot.t_start_generation = t_current;
+                    slot.t_prompt_processing = (slot.t_start_generation - slot.t_start_process_prompt) / 1e3;
+                    metrics.on_prompt_eval(slot);
                 }
 
-                // check if the slot supports speculative decoding
-                if (!slot.can_speculate()) {
+                slot.t_token_generation = (t_current - slot.t_start_generation) / 1e3;
+
+                completion_token_output result;
+                result.tok = id;
+
+                const auto * cur_p = common_sampler_get_candidates(slot.smpl);
+
+                for (size_t i = 0; i < (size_t) slot.params.sampling.n_probs; ++i) {
+                    result.probs.push_back({
+                        cur_p->data[i].id,
+                            i >= cur_p->size ? 0.0f : cur_p->data[i].p,
+                    });
+                }
+
+                if (!process_token(result, slot)) {
+                    // release slot because of stop condition
+                    slot.release();
+                    slot.print_timings();
+                    send_final_response(slot);
+                    metrics.on_prediction(slot);
                     continue;
                 }
+            }
+
+            // do speculative decoding
+            for (auto & slot : slots) {
+                if (!slot.is_processing() || !slot.can_speculate()) {
+                    continue;
+                }
+
+                if (slot.state != SLOT_STATE_GENERATING) {
+                    continue;
+                }
+
+                llama_token id = slot.sampled;
 
                 struct common_speculative_params params_spec;
                 params_spec.n_draft   = slot.params.speculative.n_max;
@@ -2433,16 +2446,6 @@ int main(int argc, char ** argv) {
     LOG_INF("%s\n", common_params_get_system_info(params).c_str());
     LOG_INF("\n");
 
-    // static files
-    std::map<std::string, server_static_file> static_files = {
-        { "/",                        { index_html,              index_html_len,              "text/html; charset=utf-8" }},
-        { "/completion.js",           { completion_js,           completion_js_len,           "text/javascript; charset=utf-8" }},
-        { "/deps_daisyui.min.css",    { deps_daisyui_min_css,    deps_daisyui_min_css_len,    "text/css; charset=utf-8" }},
-        { "/deps_markdown-it.js",     { deps_markdown_it_js,     deps_markdown_it_js_len,     "text/javascript; charset=utf-8" }},
-        { "/deps_tailwindcss.js",     { deps_tailwindcss_js,     deps_tailwindcss_js_len,     "text/javascript; charset=utf-8" }},
-        { "/deps_vue.esm-browser.js", { deps_vue_esm_browser_js, deps_vue_esm_browser_js_len, "text/javascript; charset=utf-8" }},
-    };
-
     std::unique_ptr<httplib::Server> svr;
 #ifdef CPPHTTPLIB_OPENSSL_SUPPORT
     if (params.ssl_file_key != "" && params.ssl_file_cert != "") {
@@ -2523,7 +2526,7 @@ int main(int argc, char ** argv) {
     // Middlewares
     //
 
-    auto middleware_validate_api_key = [&params, &res_error, &static_files](const httplib::Request & req, httplib::Response & res) {
+    auto middleware_validate_api_key = [&params, &res_error](const httplib::Request & req, httplib::Response & res) {
         static const std::unordered_set<std::string> public_endpoints = {
             "/health",
             "/models",
@@ -2536,7 +2539,7 @@ int main(int argc, char ** argv) {
         }
 
         // If path is public or is static file, skip validation
-        if (public_endpoints.find(req.path) != public_endpoints.end() || static_files.find(req.path) != static_files.end()) {
+        if (public_endpoints.find(req.path) != public_endpoints.end() || req.path == "/") {
             return true;
         }
 
@@ -3293,14 +3296,11 @@ int main(int argc, char ** argv) {
             return 1;
         }
     } else {
-        // using embedded static files
-        for (const auto & it : static_files) {
-            const server_static_file & static_file = it.second;
-            svr->Get(it.first.c_str(), [&static_file](const httplib::Request &, httplib::Response & res) {
-                res.set_content(reinterpret_cast<const char*>(static_file.data), static_file.size, static_file.mime_type);
-                return false;
-            });
-        }
+        // using embedded static index.html
+        svr->Get("/", [](const httplib::Request &, httplib::Response & res) {
+            res.set_content(reinterpret_cast<const char*>(index_html), index_html_len, "text/html; charset=utf-8");
+            return false;
+        });
     }
 
     // register API routes
@@ -3348,8 +3348,18 @@ int main(int argc, char ** argv) {
         llama_backend_free();
     };
 
-    // bind HTTP listen port, run the HTTP server in a thread
-    if (!svr->bind_to_port(params.hostname, params.port)) {
+    // bind HTTP listen port
+    bool was_bound = false;
+    if (params.port == 0) {
+        int bound_port = svr->bind_to_any_port(params.hostname);
+        if ((was_bound = (bound_port >= 0))) {
+            params.port = bound_port;
+        }
+    } else {
+        was_bound = svr->bind_to_port(params.hostname, params.port);
+    }
+
+    if (!was_bound) {
         //LOG_ERROR("couldn't bind HTTP server socket", {
         //    {"hostname", params.hostname},
         //    {"port", params.port},
@@ -3358,6 +3368,8 @@ int main(int argc, char ** argv) {
         clean_up();
         return 1;
     }
+
+    // run the HTTP server in a thread
     std::thread t([&]() { svr->listen_after_bind(); });
     svr->wait_until_ready();
 

examples/server/tests/.gitignore

@@ -1 +1,2 @@
 .venv
+tmp

examples/server/tests/README.md

@@ -1,19 +1,9 @@
 # Server tests
 
-Python based server tests scenario using [BDD](https://en.wikipedia.org/wiki/Behavior-driven_development)
-and [behave](https://behave.readthedocs.io/en/latest/):
-
-* [issues.feature](./features/issues.feature) Pending issues scenario
-* [parallel.feature](./features/parallel.feature) Scenario involving multi slots and concurrent requests
-* [security.feature](./features/security.feature) Security, CORS and API Key
-* [server.feature](./features/server.feature) Server base scenario: completion, embedding, tokenization, etc...
+Python based server tests scenario using [pytest](https://docs.pytest.org/en/stable/).
 
 Tests target GitHub workflows job runners with 4 vCPU.
 
-Requests are
-using [aiohttp](https://docs.aiohttp.org/en/stable/client_reference.html), [asyncio](https://docs.python.org/fr/3/library/asyncio.html)
-based http client.
-
 Note: If the host architecture inference speed is faster than GitHub runners one, parallel scenario may randomly fail.
 To mitigate it, you can increase values in `n_predict`, `kv_size`.
 
@@ -39,26 +29,19 @@ It's possible to override some scenario steps values with environment variables:
 |--------------------------|------------------------------------------------------------------------------------------------|
 | `PORT`                   | `context.server_port` to set the listening port of the server during scenario, default: `8080` |
 | `LLAMA_SERVER_BIN_PATH`  | to change the server binary path, default: `../../../build/bin/llama-server`                         |
-| `DEBUG`                  | "ON" to enable steps and server verbose mode `--verbose`                                       |
+| `DEBUG`                  | to enable steps and server verbose mode `--verbose`                                       |
 | `N_GPU_LAYERS`           | number of model layers to offload to VRAM `-ngl --n-gpu-layers`                                |
 
-### Run @bug, @wip or @wrong_usage annotated scenario
-
-Feature or Scenario must be annotated with `@llama.cpp` to be included in the default scope.
-
-- `@bug` annotation aims to link a scenario with a GitHub issue.
-- `@wrong_usage` are meant to show user issue that are actually an expected behavior
-- `@wip` to focus on a scenario working in progress
-- `@slow` heavy test, disabled by default
-
-To run a scenario annotated with `@bug`, start:
+To run slow tests:
 
 ```shell
-DEBUG=ON ./tests.sh --no-skipped --tags bug --stop
+SLOW_TESTS=1 ./tests.sh
 ```
 
-After changing logic in `steps.py`, ensure that `@bug` and `@wrong_usage` scenario are updated.
+To run with stdout/stderr display in real time (verbose output, but useful for debugging):
 
 ```shell
-./tests.sh --no-skipped --tags bug,wrong_usage || echo "should failed but compile"
+DEBUG=1 ./tests.sh -s -v -x
 ```
+
+To see all available arguments, please refer to [pytest documentation](https://docs.pytest.org/en/stable/how-to/usage.html)

examples/server/tests/conftest.py

@@ -0,0 +1,15 @@
+import pytest
+from utils import *
+
+
+# ref: https://stackoverflow.com/questions/22627659/run-code-before-and-after-each-test-in-py-test
+@pytest.fixture(autouse=True)
+def stop_server_after_each_test():
+    # do nothing before each test
+    yield
+    # stop all servers after each test
+    instances = set(
+        server_instances
+    )  # copy the set to prevent 'Set changed size during iteration'
+    for server in instances:
+        server.stop()

examples/server/tests/features/ctx_shift.feature

@@ -1,66 +0,0 @@
-@llama.cpp
-@ctx_shift
-Feature: llama.cpp server
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/stories260K.gguf from HF repo ggml-org/models
-    And   a model file test-model.gguf
-    And   a model alias tinyllama-2
-    And   BOS token is 1
-    And   42 as server seed
-    And   256 KV cache size
-    And   32 as batch size
-    And   2 slots
-
-    # the prompt is 301 tokens
-    # the slot context is 256/2 = 128 tokens
-    # the prompt is truncated to keep the last 109 tokens
-    # 64 tokens are generated thanks to shifting the context when it gets full
-  Scenario: Inference with context shift
-    And   64 server max tokens to predict
-    Then  the server is starting
-    Then  the server is healthy
-    Given a prompt:
-    """
-    Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
-    Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
-    Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
-    Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
-    """
-    And   a completion request with no api error
-    Then  64 tokens are predicted matching fun|Annaks|popcorns|pictry|bowl
-    And   the completion is  truncated
-    And   109 prompt tokens are processed
-
-  Scenario Outline: Inference without context shift
-    And   <n_predict> server max tokens to predict
-    And   disable context shifting
-    Then  the server is starting
-    Then  the server is healthy
-    Given a prompt:
-    """
-    Hi how are you
-    """
-    And   a completion request with no api error
-    Then  <n_token_output> tokens are predicted matching twind|Anna
-    And   the completion is <truncated> truncated
-    And   8 prompt tokens are processed
-    Examples:
-      | n_predict | n_token_output | truncated |
-      | 64        | 64             | not       |
-      | -1        | 120            |           |
-
-  Scenario: Inference without context shift (expected error: prompt too long)
-    And   disable context shifting
-    Then  the server is starting
-    Then  the server is healthy
-    Given a prompt:
-    """
-    Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
-    Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
-    Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
-    Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
-    """
-    And   a completion request with 400 api error
-

examples/server/tests/features/embeddings.feature

@@ -1,113 +0,0 @@
-@llama.cpp
-@embeddings
-Feature: llama.cpp server
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model url https://huggingface.co/ggml-org/models/resolve/main/bert-bge-small/ggml-model-f16.gguf
-    And   a model file bert-bge-small.gguf
-    And   a model alias bert-bge-small
-    And   42 as server seed
-    And   2 slots
-    # the bert-bge-small model has context size of 512
-    # since the generated prompts are as big as the batch size, we need to set the batch size to <= 512
-    # ref: https://huggingface.co/BAAI/bge-small-en-v1.5/blob/5c38ec7c405ec4b44b94cc5a9bb96e735b38267a/config.json#L20
-    And   128 as batch size
-    And   128 as ubatch size
-    And   512 KV cache size
-    And   enable embeddings endpoint
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario: Embedding
-    When embeddings are computed for:
-    """
-    What is the capital of Bulgaria ?
-    """
-    Then embeddings are generated
-
-  Scenario: Embedding (error: prompt too long)
-    When embeddings are computed for:
-    """
-    Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
-    Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
-    Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
-    Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
-    Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
-    Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
-    Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
-    Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
-    """
-    And  embeddings request with 500 api error
-
-  Scenario: OAI Embeddings compatibility
-    Given a model bert-bge-small
-    When an OAI compatible embeddings computation request for:
-    """
-    What is the capital of Spain ?
-    """
-    Then embeddings are generated
-
-  Scenario: OAI Embeddings compatibility with multiple inputs
-    Given a model bert-bge-small
-    Given a prompt:
-      """
-      In which country Paris is located ?
-      """
-    And a prompt:
-      """
-      Is Madrid the capital of Spain ?
-      """
-    When an OAI compatible embeddings computation request for multiple inputs
-    Then embeddings are generated
-
-  Scenario: Multi users embeddings
-    Given a prompt:
-      """
-      Write a very long story about AI.
-      """
-    And a prompt:
-      """
-      Write another very long music lyrics.
-      """
-    And a prompt:
-      """
-      Write a very long poem.
-      """
-    And a prompt:
-      """
-      Write a very long joke.
-      """
-    Given concurrent embedding requests
-    Then the server is busy
-    Then the server is idle
-    Then all embeddings are generated
-
-  Scenario: Multi users OAI compatibility embeddings
-    Given a prompt:
-      """
-      In which country Paris is located ?
-      """
-    And a prompt:
-      """
-      Is Madrid the capital of Spain ?
-      """
-    And a prompt:
-      """
-      What is the biggest US city ?
-      """
-    And a prompt:
-      """
-      What is the capital of Bulgaria ?
-      """
-    And   a model bert-bge-small
-    Given concurrent OAI embedding requests
-    Then the server is busy
-    Then the server is idle
-    Then all embeddings are generated
-
-  Scenario: All embeddings should be the same
-    Given 10 fixed prompts
-    And   a model bert-bge-small
-    Given concurrent OAI embedding requests
-    Then all embeddings are the same

examples/server/tests/features/environment.py

@@ -1,71 +0,0 @@
-import os
-import signal
-import socket
-import sys
-import time
-import traceback
-from contextlib import closing
-from subprocess import TimeoutExpired
-
-
-def before_scenario(context, scenario):
-    context.debug = 'DEBUG' in os.environ and os.environ['DEBUG'] == 'ON'
-    if context.debug:
-        print("DEBUG=ON")
-    print(f"\x1b[33;42mStarting new scenario: {scenario.name}!\x1b[0m")
-    port = 8080
-    if 'PORT' in os.environ:
-        port = int(os.environ['PORT'])
-    if is_server_listening("localhost", port):
-        assert False, "Server already started"
-
-
-def after_scenario(context, scenario):
-    try:
-        if 'server_process' not in context or context.server_process is None:
-            return
-        if scenario.status == "failed":
-            if 'GITHUB_ACTIONS' in os.environ:
-                print(f"\x1b[33;101mSCENARIO FAILED: {scenario.name} server logs:\x1b[0m\n")
-                if os.path.isfile('llama.log'):
-                    with closing(open('llama.log', 'r')) as f:
-                        for line in f:
-                            print(line)
-            if not is_server_listening(context.server_fqdn, context.server_port):
-                print("\x1b[33;101mERROR: Server stopped listening\x1b[0m")
-
-        if context.server_process.poll() is not None:
-            assert False, f"Server not running pid={context.server_process.pid} ..."
-
-        server_graceful_shutdown(context)  # SIGINT
-
-        try:
-            context.server_process.wait(0.5)
-        except TimeoutExpired:
-            print(f"server still alive after 500ms, force-killing pid={context.server_process.pid} ...")
-            context.server_process.kill()  # SIGKILL
-            context.server_process.wait()
-
-        while is_server_listening(context.server_fqdn, context.server_port):
-            time.sleep(0.1)
-    except Exception:
-        print("ignoring error in after_scenario:")
-        traceback.print_exc(file=sys.stdout)
-
-
-def server_graceful_shutdown(context):
-    print(f"shutting down server pid={context.server_process.pid} ...")
-    if os.name == 'nt':
-        interrupt = signal.CTRL_C_EVENT
-    else:
-        interrupt = signal.SIGINT
-    context.server_process.send_signal(interrupt)
-
-
-def is_server_listening(server_fqdn, server_port):
-    with closing(socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as sock:
-        result = sock.connect_ex((server_fqdn, server_port))
-        _is_server_listening = result == 0
-        if _is_server_listening:
-            print(f"server is listening on {server_fqdn}:{server_port}...")
-        return _is_server_listening

examples/server/tests/features/infill.feature

@@ -1,36 +0,0 @@
-@llama.cpp
-@infill
-Feature: llama.cpp server
-
-  # The current model is made by adding FIM tokens to the existing stories260K
-  # We may want to use a better model in the future, maybe something like SmolLM 360M
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/stories260K-infill.gguf from HF repo ggml-org/models
-    And   a model file test-model-infill.gguf
-    And   a model alias tinyllama-infill
-    And   42 as server seed
-    And   1024 as batch size
-    And   1024 as ubatch size
-    And   2048 KV cache size
-    And   64 max tokens to predict
-    And   0.0 temperature
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario: Infill without input_extra
-    Given a prompt "Complete this"
-    And   an infill input extra none none
-    And   an infill input prefix "#include <cstdio>\n#include \"llama.h\"\n\nint main() {\n    int n_threads = llama_"
-    And   an infill input suffix "}\n"
-    And   an infill request with no api error
-    Then  64 tokens are predicted matching One|day|she|saw|big|scary|bird
-
-  Scenario: Infill with input_extra
-    Given a prompt "Complete this"
-    And   an infill input extra "llama.h" "LLAMA_API int32_t llama_n_threads();\n"
-    And   an infill input prefix "#include <cstdio>\n#include \"llama.h\"\n\nint main() {\n    int n_threads = llama_"
-    And   an infill input suffix "}\n"
-    And   an infill request with no api error
-    Then  64 tokens are predicted matching cuts|Jimmy|mom|came|into|the|room"

examples/server/tests/features/issues.feature

@@ -1,5 +0,0 @@
-# List of ongoing issues
-# run with: DEBUG=ON ./tests.sh --no-skipped --tags bug
-@bug
-Feature: Issues
-  # No confirmed issue at the moment

examples/server/tests/features/lora.feature

@@ -1,36 +0,0 @@
-@llama.cpp
-@lora
-Feature: llama.cpp server
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model url https://huggingface.co/ggml-org/stories15M_MOE/resolve/main/stories15M_MOE-F16.gguf
-    And   a model file stories15M_MOE-F16.gguf
-    And   a model alias stories15M_MOE
-    And   a lora adapter file from https://huggingface.co/ggml-org/stories15M_MOE/resolve/main/moe_shakespeare15M.gguf
-    And   42 as server seed
-    And   1024 as batch size
-    And   1024 as ubatch size
-    And   2048 KV cache size
-    And   64 max tokens to predict
-    And   0.0 temperature
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario: Completion LoRA disabled
-    Given switch off lora adapter 0
-    Given a prompt:
-    """
-    Look in thy glass
-    """
-    And   a completion request with no api error
-    Then  64 tokens are predicted matching little|girl|three|years|old
-
-  Scenario: Completion LoRA enabled
-    Given switch on lora adapter 0
-    Given a prompt:
-    """
-    Look in thy glass
-    """
-    And   a completion request with no api error
-    Then  64 tokens are predicted matching eye|love|glass|sun

examples/server/tests/features/parallel.feature

@@ -1,131 +0,0 @@
-@llama.cpp
-@parallel
-Feature: Parallel
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/split/stories15M-00001-of-00003.gguf from HF repo ggml-org/models
-    And   a model file test-model-00001-of-00003.gguf
-    And   42 as server seed
-    And   128 as batch size
-    And   256 KV cache size
-    And   2 slots
-    And   continuous batching
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario Outline: Multi users completion
-    Given a prompt:
-      """
-      Write a very long story about AI.
-      """
-    And a prompt:
-      """
-      Write another very long music lyrics.
-      """
-    And <n_predict> max tokens to predict
-    Given concurrent completion requests
-    Then the server is busy
-    Then the server is idle
-    And  all slots are idle
-    Then all prompts are predicted with <n_predict> tokens
-    Examples:
-      | n_predict |
-      | 128       |
-
-  Scenario Outline: Multi users OAI completions compatibility
-    Given a system prompt You are a writer.
-    And   a model tinyllama-2
-    Given a prompt:
-      """
-      Write a very long book.
-      """
-    And a prompt:
-      """
-      Write another a poem.
-      """
-    And <n_predict> max tokens to predict
-    And streaming is <streaming>
-    Given concurrent OAI completions requests
-    Then the server is busy
-    Then the server is idle
-    Then all prompts are predicted with <n_predict> tokens
-    Examples:
-      | streaming | n_predict |
-      | disabled  | 128       |
-      | enabled   | 64        |
-
-  Scenario Outline: Multi users OAI completions compatibility no v1
-    Given a system prompt You are a writer.
-    And   a model tinyllama-2
-    Given a prompt:
-      """
-      Write a very long book.
-      """
-    And a prompt:
-      """
-      Write another a poem.
-      """
-    And <n_predict> max tokens to predict
-    And streaming is <streaming>
-    Given concurrent OAI completions requests no v1
-    Then the server is busy
-    Then the server is idle
-    Then all prompts are predicted with <n_predict> tokens
-    Examples:
-      | streaming | n_predict |
-      | disabled  | 128       |
-      | enabled   | 64        |
-
-  Scenario Outline: Multi users with number of prompts exceeding number of slots
-    Given a system prompt You are a writer.
-    And   a model tinyllama-2
-    Given a prompt:
-      """
-      Write a very long book.
-      """
-    And a prompt:
-      """
-      Write another a poem.
-      """
-    And a prompt:
-      """
-      What is LLM?
-      """
-    And a prompt:
-      """
-      The sky is blue and I love it.
-      """
-    And <n_predict> max tokens to predict
-    And streaming is <streaming>
-    Given concurrent OAI completions requests
-    Then the server is busy
-    Then the server is idle
-    Then all prompts are predicted with <n_predict> tokens
-    Examples:
-      | streaming | n_predict |
-      | disabled  | 128       |
-      | enabled   | 64        |
-
-  Scenario:  Multi users with total number of tokens to predict exceeds the KV Cache size #3969
-    Given a prompt:
-      """
-      Write a very long story about AI.
-      """
-    And a prompt:
-      """
-      Write another very long music lyrics.
-      """
-    And a prompt:
-      """
-      Write a very long poem.
-      """
-    And a prompt:
-      """
-      Write a very long joke.
-      """
-    And 128 max tokens to predict
-    Given concurrent completion requests
-    Then the server is busy
-    Then the server is idle
-    Then all prompts are predicted

examples/server/tests/features/passkey.feature

@@ -1,56 +0,0 @@
-# run with: ./tests.sh --no-skipped --tags passkey
-@passkey
-@slow
-Feature: Passkey / Self-extend with context shift
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-
-  # Generates a long text of junk and inserts a secret passkey number inside it.
-  # Then we query the LLM for the secret passkey.
-  # see #3856 and #4810
-  Scenario Outline: Passkey
-    Given a model file <hf_file> from HF repo <hf_repo>
-    And   <n_batch> as batch size
-    And   <n_junk> as number of junk
-    And   <n_predicted> server max tokens to predict
-    And   42 as seed
-    And   0.0 temperature
-    And   <n_ctx> KV cache size
-    And   1 slots
-    And   <n_ga> group attention factor to extend context size through self-extend
-    And   <n_ga_w> group attention width to extend context size through self-extend
-    # Can be override with N_GPU_LAYERS
-    And   <ngl> GPU offloaded layers
-    Then  the server is starting
-    # Higher timeout because the model may need to be downloaded from the internet
-    Then  the server is healthy with timeout 120 seconds
-    Given available models
-    Then  model 0 is trained on <n_ctx_train> tokens context
-    Given a prefix prompt:
-    """
-    here is an important info hidden inside a lot of irrelevant text. Find it and memorize them. I will quiz you about the important information there.
-    """
-    And a passkey prompt template:
-    """
-    The pass key is <passkey> Remember it. <passkey> is the pass key.
-    """
-    And a junk suffix prompt:
-    """
-    The grass is green. The sky is blue. The sun is yellow. Here we go. There and back again.
-    """
-    And a suffix prompt:
-    """
-    What is the pass key? The pass key is
-    """
-    Given a "<passkey>" passkey challenge prompt with the passkey inserted every <i_pos> junk
-    And  a completion request with no api error
-    Then <n_predicted> tokens are predicted matching <re_content>
-
-    Examples:
-      | hf_repo                         | hf_file                     | n_ctx_train | ngl | n_ctx | n_batch | n_ga | n_ga_w | n_junk | i_pos | passkey | n_predicted | re_content     |
-      | TheBloke/phi-2-GGUF             | phi-2.Q4_K_M.gguf           | 2048        | 5   | 8192  | 512     | 4    | 512    | 250    | 50    | 42      | 1           | 42             |
-      | TheBloke/phi-2-GGUF             | phi-2.Q4_K_M.gguf           | 2048        | 5   | 8192  | 512     | 2    | 512    | 250    | 50    | 42      | 1           | \b((?!42)\w)+\b  |
-      #| TheBloke/Llama-2-7B-GGUF        | llama-2-7b.Q2_K.gguf        | 4096        | 3   | 16384 | 512     | 4    | 512    | 500    | 300   | 1234    | 5           | 1234           |
-      #| TheBloke/Mixtral-8x7B-v0.1-GGUF | mixtral-8x7b-v0.1.Q2_K.gguf | 32768       | 2   | 16384 | 512     | 4    | 512    | 500    | 100   | 0987    | 5           | 0
-      # 987           |

examples/server/tests/features/rerank.feature

@@ -1,42 +0,0 @@
-@llama.cpp
-@rerank
-Feature: llama.cpp server
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model url https://huggingface.co/ggml-org/models/resolve/main/jina-reranker-v1-tiny-en/ggml-model-f16.gguf
-    And   a model file jina-reranker-v1-tiny-en.gguf
-    And   a model alias jina-reranker-v1-tiny-en
-    And   42 as server seed
-    And   2 slots
-    And   512 as batch size
-    And   512 as ubatch size
-    And   512 KV cache size
-    And   enable reranking endpoint
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario: Rerank
-    Given a rerank query:
-      """
-      Machine learning is
-      """
-    And   a rerank document:
-      """
-      A machine is a physical system that uses power to apply forces and control movement to perform an action. The term is commonly applied to artificial devices, such as those employing engines or motors, but also to natural biological macromolecules, such as molecular machines.
-      """
-    And   a rerank document:
-      """
-      Learning is the process of acquiring new understanding, knowledge, behaviors, skills, values, attitudes, and preferences. The ability to learn is possessed by humans, non-human animals, and some machines; there is also evidence for some kind of learning in certain plants.
-      """
-    And   a rerank document:
-      """
-      Machine learning is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data, and thus perform tasks without explicit instructions.
-      """
-    And   a rerank document:
-      """
-      Paris, capitale de la France, est une grande ville européenne et un centre mondial de l'art, de la mode, de la gastronomie et de la culture. Son paysage urbain du XIXe siècle est traversé par de larges boulevards et la Seine.
-      """
-    When  reranking request
-    Then  reranking results are returned
-    Then  reranking highest score is index 2 and lowest score is index 3

examples/server/tests/features/results.feature

@@ -1,118 +0,0 @@
-@llama.cpp
-@results
-Feature: Results
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/split/stories15M-00001-of-00003.gguf from HF repo ggml-org/models
-    And   a model file test-model-00001-of-00003.gguf
-    And   128 as batch size
-    And   1024 KV cache size
-    And   128 max tokens to predict
-    And   continuous batching
-
-  Scenario Outline: consistent results with same seed
-    Given <n_slots> slots
-    And   1.0 temperature
-    Then  the server is starting
-    Then  the server is healthy
-
-    Given 4 prompts "Title: Little Red Riding Hood But In Space\n\nSummary:" with seed 42
-
-    Given concurrent completion requests
-    Then the server is busy
-    Then the server is idle
-    And  all slots are idle
-    Then all predictions are equal
-    Examples:
-      | n_slots |
-      | 1       |
-      # FIXME: unified KV cache nondeterminism
-      # | 2       |
-
-  Scenario Outline: different results with different seed
-    Given <n_slots> slots
-    And   1.0 temperature
-    Then  the server is starting
-    Then  the server is healthy
-
-    Given 1 prompts "Title: Little Red Riding Hood But In Space\n\nSummary:" with seed 42
-    Given 1 prompts "Title: Little Red Riding Hood But In Space\n\nSummary:" with seed 43
-    Given 1 prompts "Title: Little Red Riding Hood But In Space\n\nSummary:" with seed 44
-    Given 1 prompts "Title: Little Red Riding Hood But In Space\n\nSummary:" with seed 45
-
-    Given concurrent completion requests
-    Then the server is busy
-    Then the server is idle
-    And  all slots are idle
-    Then all predictions are different
-    Examples:
-      | n_slots |
-      | 1       |
-      | 2       |
-
-  Scenario Outline: consistent results with same seed and varying batch size
-    Given 4 slots
-    And   <temp> temperature
-    # And   0 as draft
-    Then  the server is starting
-    Then  the server is healthy
-
-    Given 1 prompts "Write a very long story about AI." with seed 42
-    And   concurrent completion requests
-    # Then the server is busy # Not all slots will be utilized.
-    Then  the server is idle
-    And   all slots are idle
-
-    Given <n_parallel> prompts "Write a very long story about AI." with seed 42
-    And   concurrent completion requests
-    # Then the server is busy # Not all slots will be utilized.
-    Then the server is idle
-    And  all slots are idle
-
-    Then all predictions are equal
-    Examples:
-      | n_parallel | temp |
-      | 1          | 0.0  |
-      | 1          | 1.0  |
-      # FIXME: unified KV cache nondeterminism
-      # See https://github.com/ggerganov/whisper.cpp/issues/1941#issuecomment-1986923227
-      # and https://github.com/ggerganov/llama.cpp/pull/6122#discussion_r1531405574
-      # and https://github.com/ggerganov/llama.cpp/pull/7347 .
-      # | 2          | 0.0  |
-      # | 4          | 0.0  |
-      # | 2          | 1.0  |
-      # | 4          | 1.0  |
-
-  Scenario Outline: consistent token probs with same seed and prompt
-    Given <n_slots> slots
-    And   <n_kv> KV cache size
-    And   1.0 temperature
-    And   <n_predict> max tokens to predict
-    Then  the server is starting
-    Then  the server is healthy
-
-    Given 1 prompts "The meaning of life is" with seed 42
-    And   concurrent completion requests
-    # Then the server is busy # Not all slots will be utilized.
-    Then  the server is idle
-    And   all slots are idle
-
-    Given <n_parallel> prompts "The meaning of life is" with seed 42
-    And   concurrent completion requests
-    # Then the server is busy # Not all slots will be utilized.
-    Then the server is idle
-    And  all slots are idle
-
-    Then all token probabilities are equal
-    Examples:
-      | n_slots | n_kv | n_predict | n_parallel |
-      | 4       | 1024 | 1         | 1          |
-      # FIXME: unified KV cache nondeterminism
-      # See https://github.com/ggerganov/whisper.cpp/issues/1941#issuecomment-1986923227
-      # and https://github.com/ggerganov/llama.cpp/pull/6122#discussion_r1531405574
-      # and https://github.com/ggerganov/llama.cpp/pull/7347 .
-      # | 4       | 1024 | 1         | 4          |
-      # | 4       | 1024 | 100       | 1          |
-      # This test still fails even the above patches; the first token probabilities are already different.
-      # | 4       | 1024 | 100       | 4          |

examples/server/tests/features/security.feature

@@ -1,68 +0,0 @@
-@llama.cpp
-@security
-Feature: Security
-
-  Background: Server startup with an api key defined
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/stories260K.gguf from HF repo ggml-org/models
-    And   a server api key THIS_IS_THE_KEY
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario Outline: Completion with some user api key
-    Given a prompt test
-    And   a user api key <api_key>
-    And   4 max tokens to predict
-    And   a completion request with <api_error> api error
-
-    Examples: Prompts
-      | api_key         | api_error |
-      | THIS_IS_THE_KEY | no        |
-      | THIS_IS_THE_KEY | no        |
-      | hackeme         | raised    |
-      |                 | raised    |
-
-  Scenario Outline: OAI Compatibility
-    Given a system prompt test
-    And   a user prompt test
-    And   a model test
-    And   2 max tokens to predict
-    And   streaming is disabled
-    And   a user api key <api_key>
-    Given an OAI compatible chat completions request with <api_error> api error
-
-    Examples: Prompts
-      | api_key         | api_error |
-      | THIS_IS_THE_KEY | no        |
-      | THIS_IS_THE_KEY | no        |
-      | hackme          | raised    |
-
-  Scenario Outline: OAI Compatibility (invalid response formats)
-    Given a system prompt test
-    And   a user prompt test
-    And   a response format <response_format>
-    And   a model test
-    And   2 max tokens to predict
-    And   streaming is disabled
-    Given an OAI compatible chat completions request with raised api error
-
-    Examples: Prompts
-      | response_format                                       |
-      | {"type": "sound"}                                     |
-      | {"type": "json_object", "schema": 123}                |
-      | {"type": "json_object", "schema": {"type": 123}}      |
-      | {"type": "json_object", "schema": {"type": "hiccup"}} |
-
-
-  Scenario Outline: CORS Options
-    Given a user api key THIS_IS_THE_KEY
-    When  an OPTIONS request is sent from <origin>
-    Then  CORS header <cors_header> is set to <cors_header_value>
-
-    Examples: Headers
-      | origin          | cors_header                      | cors_header_value |
-      | localhost       | Access-Control-Allow-Origin      | localhost         |
-      | web.mydomain.fr | Access-Control-Allow-Origin      | web.mydomain.fr   |
-      | origin          | Access-Control-Allow-Credentials | true              |
-      | web.mydomain.fr | Access-Control-Allow-Methods     | GET, POST         |
-      | web.mydomain.fr | Access-Control-Allow-Headers     | *                 |

examples/server/tests/features/server.feature

@@ -1,120 +0,0 @@
-@llama.cpp
-@server
-Feature: llama.cpp server
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/stories260K.gguf from HF repo ggml-org/models
-    And   a model file test-model.gguf
-    And   a model alias tinyllama-2
-    And   BOS token is 1
-    And   42 as server seed
-      # KV Cache corresponds to the total amount of tokens
-      # that can be stored across all independent sequences: #4130
-      # see --ctx-size and #5568
-    And   256 KV cache size
-    And   32 as batch size
-    And   2 slots
-    And   64 server max tokens to predict
-    And   prometheus compatible metrics exposed
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario: Health
-    Then the server is ready
-    And  all slots are idle
-
-
-  Scenario Outline: Completion
-    Given a prompt <prompt>
-    And   <n_predict> max tokens to predict
-    And   a completion request with no api error
-    Then  <n_predicted> tokens are predicted matching <re_content>
-    And   the completion is <truncated> truncated
-    And   <n_prompt> prompt tokens are processed
-    And   prometheus metrics are exposed
-    And   metric llamacpp:tokens_predicted is <n_predicted>
-
-    Examples: Prompts
-      | prompt                                                                    | n_predict | re_content                                  | n_prompt | n_predicted | truncated |
-      | I believe the meaning of life is                                          | 8         | (read\|going)+                              | 18       | 8           | not       |
-      | Write a joke about AI from a very long prompt which will not be truncated | 256       | (princesses\|everyone\|kids\|Anna\|forest)+ | 46       | 64          | not       |
-
-  Scenario: Completion prompt truncated
-    Given a prompt:
-    """
-    Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
-    Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
-    Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
-    Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
-    """
-    And   a completion request with no api error
-    Then  64 tokens are predicted matching fun|Annaks|popcorns|pictry|bowl
-    And   the completion is  truncated
-    And   109 prompt tokens are processed
-
-
-  Scenario Outline: OAI Compatibility
-    Given a model <model>
-    And   a system prompt <system_prompt>
-    And   a user prompt <user_prompt>
-    And   <max_tokens> max tokens to predict
-    And   streaming is <enable_streaming>
-    Given an OAI compatible chat completions request with no api error
-    Then  <n_predicted> tokens are predicted matching <re_content>
-    And   <n_prompt> prompt tokens are processed
-    And   the completion is <truncated> truncated
-
-    Examples: Prompts
-      | model        | system_prompt               | user_prompt                          | max_tokens | re_content                        | n_prompt | n_predicted | enable_streaming | truncated |
-      | llama-2      | Book                        | What is the best book                | 8          | (Here\|what)+                     | 77       | 8           | disabled         | not       |
-      | codellama70b | You are a coding assistant. | Write the fibonacci function in c++. | 128        | (thanks\|happy\|bird\|Annabyear)+ | -1       | 64          | enabled          |           |
-
-
-  Scenario Outline: OAI Compatibility w/ response format
-    Given a model test
-    And   a system prompt test
-    And   a user prompt test
-    And   a response format <response_format>
-    And   10 max tokens to predict
-    Given an OAI compatible chat completions request with no api error
-    Then  <n_predicted> tokens are predicted matching <re_content>
-
-    Examples: Prompts
-      | response_format                                                     | n_predicted | re_content             |
-      | {"type": "json_object", "schema": {"const": "42"}}                  | 6           | "42"                   |
-      | {"type": "json_object", "schema": {"items": [{"type": "integer"}]}} | 10          | \[ -300 \]             |
-      | {"type": "json_object"}                                             | 10          | \{ " Jacky.            |
-
-
-  Scenario: Tokenize / Detokenize
-    When tokenizing:
-    """
-    What is the capital of France ?
-    """
-    Then tokens can be detokenized
-    And  tokens do not begin with BOS
-
-  Scenario: Tokenize w/ BOS
-    Given adding special tokens
-    When  tokenizing:
-    """
-    What is the capital of Germany?
-    """
-    Then  tokens begin with BOS
-    Given first token is removed
-    Then  tokens can be detokenized
-
-  Scenario: Tokenize with pieces
-    When  tokenizing with pieces:
-    """
-    What is the capital of Germany?
-    媽
-    """
-    Then  tokens are given with pieces
-
-  Scenario: Models available
-    Given available models
-    Then  1 models are supported
-    Then  model 0 is identified by tinyllama-2
-    Then  model 0 is trained on 128 tokens context

examples/server/tests/features/slotsave.feature

@@ -1,58 +0,0 @@
-@llama.cpp
-@slotsave
-Feature: llama.cpp server slot management
-
-  Background: Server startup
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/stories260K.gguf from HF repo ggml-org/models
-    And   prompt caching is enabled
-    And   2 slots
-    And   . as slot save path
-    And   2048 KV cache size
-    And   42 as server seed
-    And   24 max tokens to predict
-    Then  the server is starting
-    Then  the server is healthy
-
-  Scenario: Save and Restore Slot
-    # First prompt in slot 1 should be fully processed
-    Given a user prompt "What is the capital of France?"
-    And   using slot id 1
-    And   a completion request with no api error
-    Then  24 tokens are predicted matching (Lily|cake)
-    And   22 prompt tokens are processed
-    When  the slot 1 is saved with filename "slot1.bin"
-    Then  the server responds with status code 200
-    # Since we have cache, this should only process the last tokens
-    Given a user prompt "What is the capital of Germany?"
-    And   a completion request with no api error
-    Then  24 tokens are predicted matching (Thank|special)
-    And   7 prompt tokens are processed
-    # Loading the original cache into slot 0,
-    # we should only be processing 1 prompt token and get the same output
-    When  the slot 0 is restored with filename "slot1.bin"
-    Then  the server responds with status code 200
-    Given a user prompt "What is the capital of France?"
-    And   using slot id 0
-    And   a completion request with no api error
-    Then  24 tokens are predicted matching (Lily|cake)
-    And   1 prompt tokens are processed
-    # For verification that slot 1 was not corrupted during slot 0 load, same thing
-    Given a user prompt "What is the capital of Germany?"
-    And   using slot id 1
-    And   a completion request with no api error
-    Then  24 tokens are predicted matching (Thank|special)
-    And   1 prompt tokens are processed
-
-  Scenario: Erase Slot
-    Given a user prompt "What is the capital of France?"
-    And   using slot id 1
-    And   a completion request with no api error
-    Then  24 tokens are predicted matching (Lily|cake)
-    And   22 prompt tokens are processed
-    When  the slot 1 is erased
-    Then  the server responds with status code 200
-    Given a user prompt "What is the capital of France?"
-    And   a completion request with no api error
-    Then  24 tokens are predicted matching (Lily|cake)
-    And   22 prompt tokens are processed

examples/server/tests/features/wrong_usages.feature

@@ -1,25 +0,0 @@
-# run with: ./tests.sh --no-skipped --tags wrong_usage
-@wrong_usage
-Feature: Wrong usage of llama.cpp server
-
-  #3969 The user must always set --n-predict option
-  # to cap the number of tokens any completion request can generate
-  # or pass n_predict/max_tokens in the request.
-  Scenario: Infinite loop
-    Given a server listening on localhost:8080
-    And   a model file tinyllamas/stories260K.gguf from HF repo ggml-org/models
-    And   42 as server seed
-    And   2048 KV cache size
-    # Uncomment below to fix the issue
-    #And   64 server max tokens to predict
-    Then  the server is starting
-    Then  the server is healthy
-    Given a prompt:
-      """
-      Go to: infinite loop
-      """
-    # Uncomment below to fix the issue
-    #And   128 max tokens to predict
-    Given concurrent completion requests
-    Then the server is idle
-    Then all prompts are predicted

examples/server/tests/requirements.txt

@@ -1,7 +1,7 @@
 aiohttp~=3.9.3
-behave~=1.2.6
+pytest~=8.3.3
 huggingface_hub~=0.23.2
 numpy~=1.26.4
-openai~=1.30.3
+openai~=1.55.3
 prometheus-client~=0.20.0
 requests~=2.32.3