ggml-hexagon: swiglu_oai operation (#18114)

* snapshot: debug ggml-hexagon swiglu-oai

* fix: fix hvx_min_scalar_f32

* feat: working swiglu-oai

* chore: fix formating isue

.devops/cann.Dockerfile

@@ -107,7 +107,7 @@ ENTRYPOINT ["/app/tools.sh"]
 # ENTRYPOINT ["/app/llama-server"]
 
 ### Target: light
-# Lightweight image containing only llama-cli
+# Lightweight image containing only llama-cli and llama-completion
 # ==============================================================================
 FROM base AS light
 

.devops/llama-cli-cann.Dockerfile

@@ -23,11 +23,12 @@ ENV LD_LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}/runtime/lib64/stub:$LD_LIBRARY_PATH
 RUN echo "Building with static libs" && \
     source /usr/local/Ascend/ascend-toolkit/set_env.sh --force && \
     cmake -B build -DGGML_NATIVE=OFF -DGGML_CANN=ON -DBUILD_SHARED_LIBS=OFF -DLLAMA_BUILD_TESTS=OFF  && \
-    cmake --build build --config Release --target llama-cli
+    cmake --build build --config Release --target llama-cli && \
+    cmake --build build --config Release --target llama-completion
 
 # TODO: use image with NNRT
 FROM ascendai/cann:$ASCEND_VERSION AS runtime
-COPY --from=build /app/build/bin/llama-cli /llama-cli
+COPY --from=build /app/build/bin/llama-cli /app/build/bin/llama-completion /
 
 ENV LC_ALL=C.utf8
 

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

@@ -37,6 +37,7 @@ make -j GGML_CUDA=1
 %install
 mkdir -p %{buildroot}%{_bindir}/
 cp -p llama-cli %{buildroot}%{_bindir}/llama-cuda-cli
+cp -p llama-completion %{buildroot}%{_bindir}/llama-cuda-completion
 cp -p llama-server %{buildroot}%{_bindir}/llama-cuda-server
 cp -p llama-simple %{buildroot}%{_bindir}/llama-cuda-simple
 
@@ -68,6 +69,7 @@ rm -rf %{_builddir}/*
 
 %files
 %{_bindir}/llama-cuda-cli
+%{_bindir}/llama-cuda-completion
 %{_bindir}/llama-cuda-server
 %{_bindir}/llama-cuda-simple
 /usr/lib/systemd/system/llamacuda.service

.devops/llama-cpp.srpm.spec

@@ -39,6 +39,7 @@ make -j
 %install
 mkdir -p %{buildroot}%{_bindir}/
 cp -p llama-cli %{buildroot}%{_bindir}/llama-cli
+cp -p llama-completion %{buildroot}%{_bindir}/llama-completion
 cp -p llama-server %{buildroot}%{_bindir}/llama-server
 cp -p llama-simple %{buildroot}%{_bindir}/llama-simple
 
@@ -70,6 +71,7 @@ rm -rf %{_builddir}/*
 
 %files
 %{_bindir}/llama-cli
+%{_bindir}/llama-completion
 %{_bindir}/llama-server
 %{_bindir}/llama-simple
 /usr/lib/systemd/system/llama.service

.github/ISSUE_TEMPLATE/019-bug-misc.yml

@@ -86,6 +86,7 @@ body:
       description: >
           If applicable, please copy and paste any relevant log output, including any generated text.
           This will be automatically formatted into code, so no need for backticks.
+          If you are encountering problems specifically with the `llama_params_fit` module, always upload `--verbose` logs as well.
       render: shell
     validations:
       required: false

.github/workflows/server-webui.yml

@@ -0,0 +1,225 @@
+# Server WebUI build and tests
+name: Server WebUI
+
+on:
+  workflow_dispatch: # allows manual triggering
+    inputs:
+      sha:
+        description: 'Commit SHA1 to build'
+        required: false
+        type: string
+      slow_tests:
+        description: 'Run slow tests'
+        required: true
+        type: boolean
+  push:
+    branches:
+      - master
+    paths: ['.github/workflows/server-webui.yml', 'tools/server/webui/**.*', 'tools/server/tests/**.*', 'tools/server/public/**']
+  pull_request:
+    types: [opened, synchronize, reopened]
+    paths: ['.github/workflows/server-webui.yml', 'tools/server/webui/**.*', 'tools/server/tests/**.*', 'tools/server/public/**']
+
+env:
+  LLAMA_LOG_COLORS: 1
+  LLAMA_LOG_PREFIX: 1
+  LLAMA_LOG_TIMESTAMPS: 1
+  LLAMA_LOG_VERBOSITY: 10
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  webui-check:
+    name: WebUI Checks
+    runs-on: ubuntu-latest
+    continue-on-error: true
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
+
+      - name: Setup Node.js
+        id: node
+        uses: actions/setup-node@v4
+        with:
+          node-version: "22"
+          cache: "npm"
+          cache-dependency-path: "tools/server/webui/package-lock.json"
+
+      - name: Install dependencies
+        id: setup
+        if: ${{ steps.node.conclusion == 'success' }}
+        run: npm ci
+        working-directory: tools/server/webui
+
+      - name: Run type checking
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npm run check
+        working-directory: tools/server/webui
+
+      - name: Run linting
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npm run lint
+        working-directory: tools/server/webui
+
+      - name: Build application
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npm run build
+        working-directory: tools/server/webui
+
+      - name: Install Playwright browsers
+        id: playwright
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npx playwright install --with-deps
+        working-directory: tools/server/webui
+
+      - name: Build Storybook
+        if: ${{ always() && steps.playwright.conclusion == 'success' }}
+        run: npm run build-storybook
+        working-directory: tools/server/webui
+
+      - name: Run Client tests
+        if: ${{ always() && steps.playwright.conclusion == 'success' }}
+        run: npm run test:client
+        working-directory: tools/server/webui
+
+      - name: Run Unit tests
+        if: ${{ always() && steps.playwright.conclusion == 'success' }}
+        run: npm run test:unit
+        working-directory: tools/server/webui
+
+      - name: Run UI tests
+        if: ${{ always() && steps.playwright.conclusion == 'success' }}
+        run: npm run test:ui -- --testTimeout=60000
+        working-directory: tools/server/webui
+
+      - name: Run E2E tests
+        if: ${{ always() && steps.playwright.conclusion == 'success' }}
+        run: npm run test:e2e
+        working-directory: tools/server/webui
+
+  server-build:
+    runs-on: ubuntu-latest
+
+    strategy:
+      matrix:
+        sanitizer: [ADDRESS, UNDEFINED] # THREAD is broken
+        build_type: [RelWithDebInfo]
+        include:
+          - build_type: Release
+            sanitizer: ""
+      fail-fast: false # While -DLLAMA_SANITIZE_THREAD=ON is broken
+
+    steps:
+      - name: Dependencies
+        id: depends
+        run: |
+          sudo apt-get update
+          sudo apt-get -y install \
+            build-essential \
+            xxd \
+            git \
+            cmake \
+            curl \
+            wget \
+            language-pack-en \
+            libssl-dev
+
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
+
+      - name: Python setup
+        id: setup_python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Tests dependencies
+        id: test_dependencies
+        run: |
+          pip install -r tools/server/tests/requirements.txt
+
+      - name: Setup Node.js for WebUI
+        uses: actions/setup-node@v4
+        with:
+          node-version: "22"
+          cache: "npm"
+          cache-dependency-path: "tools/server/webui/package-lock.json"
+
+      - name: Install WebUI dependencies
+        run: npm ci
+        working-directory: tools/server/webui
+
+      - name: Build WebUI
+        run: npm run build
+        working-directory: tools/server/webui
+
+      - name: Build (no OpenMP)
+        id: cmake_build_no_openmp
+        if: ${{ matrix.sanitizer == 'THREAD' }}
+        run: |
+          cmake -B build \
+              -DGGML_NATIVE=OFF \
+              -DLLAMA_CURL=OFF \
+              -DLLAMA_OPENSSL=ON \
+              -DLLAMA_BUILD_SERVER=ON \
+              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
+              -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON \
+              -DGGML_OPENMP=OFF ;
+          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
+
+      - name: Build (sanitizers)
+        id: cmake_build_sanitizers
+        if: ${{ matrix.sanitizer != '' && matrix.sanitizer != 'THREAD' }}
+        run: |
+          cmake -B build \
+              -DGGML_NATIVE=OFF \
+              -DLLAMA_CURL=OFF \
+              -DLLAMA_OPENSSL=ON \
+              -DLLAMA_BUILD_SERVER=ON \
+              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
+              -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ;
+          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
+
+      - name: Build (sanitizers)
+        id: cmake_build
+        if: ${{ matrix.sanitizer == '' }}
+        run: |
+          cmake -B build \
+              -DGGML_NATIVE=OFF \
+              -DLLAMA_CURL=OFF \
+              -DLLAMA_OPENSSL=ON \
+              -DLLAMA_BUILD_SERVER=ON \
+              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} ;
+          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
+
+      - name: Tests
+        id: server_integration_tests
+        if: ${{ matrix.sanitizer == '' }}
+        env:
+          GITHUB_ACTIONS: "true"
+        run: |
+          cd tools/server/tests
+          ./tests.sh
+
+      - name: Tests (sanitizers)
+        id: server_integration_tests_sanitizers
+        if: ${{ matrix.sanitizer != '' }}
+        run: |
+          cd tools/server/tests
+          LLAMA_SANITIZE=1 ./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 tools/server/tests
+          SLOW_TESTS=1 ./tests.sh

.github/workflows/server.yml

@@ -76,270 +76,6 @@ jobs:
         run: |
           pip install -r tools/server/tests/requirements.txt
 
-  webui-setup:
-    name: WebUI Setup
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-          cache: "npm"
-          cache-dependency-path: "tools/server/webui/package-lock.json"
-
-      - name: Cache node_modules
-        uses: actions/cache@v4
-        id: cache-node-modules
-        with:
-          path: tools/server/webui/node_modules
-          key: ${{ runner.os }}-node-modules-${{ hashFiles('tools/server/webui/package-lock.json') }}
-          restore-keys: |
-            ${{ runner.os }}-node-modules-
-
-      - name: Install dependencies
-        if: steps.cache-node-modules.outputs.cache-hit != 'true'
-        run: npm ci
-        working-directory: tools/server/webui
-
-  webui-check:
-    needs: webui-setup
-    name: WebUI Check
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-
-      - name: Restore node_modules cache
-        uses: actions/cache@v4
-        with:
-          path: tools/server/webui/node_modules
-          key: ${{ runner.os }}-node-modules-${{ hashFiles('tools/server/webui/package-lock.json') }}
-          restore-keys: |
-            ${{ runner.os }}-node-modules-
-
-      - name: Run type checking
-        run: npm run check
-        working-directory: tools/server/webui
-
-      - name: Run linting
-        run: npm run lint
-        working-directory: tools/server/webui
-
-  webui-build:
-    needs: webui-check
-    name: WebUI Build
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-
-      - name: Restore node_modules cache
-        uses: actions/cache@v4
-        with:
-          path: tools/server/webui/node_modules
-          key: ${{ runner.os }}-node-modules-${{ hashFiles('tools/server/webui/package-lock.json') }}
-          restore-keys: |
-            ${{ runner.os }}-node-modules-
-
-      - name: Build application
-        run: npm run build
-        working-directory: tools/server/webui
-
-  webui-tests:
-    needs: webui-build
-    name: Run WebUI tests
-    permissions:
-      contents: read
-
-    runs-on: ubuntu-latest
-
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-
-      - name: Restore node_modules cache
-        uses: actions/cache@v4
-        with:
-          path: tools/server/webui/node_modules
-          key: ${{ runner.os }}-node-modules-${{ hashFiles('tools/server/webui/package-lock.json') }}
-          restore-keys: |
-            ${{ runner.os }}-node-modules-
-
-      - name: Install Playwright browsers
-        run: npx playwright install --with-deps
-        working-directory: tools/server/webui
-
-      - name: Build Storybook
-        run: npm run build-storybook
-        working-directory: tools/server/webui
-
-      - name: Run Client tests
-        run: npm run test:client
-        working-directory: tools/server/webui
-
-      - name: Run Server tests
-        run: npm run test:server
-        working-directory: tools/server/webui
-
-      - name: Run UI tests
-        run: npm run test:ui -- --testTimeout=60000
-        working-directory: tools/server/webui
-
-      - name: Run E2E tests
-        run: npm run test:e2e
-        working-directory: tools/server/webui
-
-  server-build:
-    needs: [webui-tests]
-    runs-on: ubuntu-latest
-
-    strategy:
-      matrix:
-        sanitizer: [ADDRESS, UNDEFINED] # THREAD is broken
-        build_type: [RelWithDebInfo]
-        include:
-          - build_type: Release
-            sanitizer: ""
-      fail-fast: false # While -DLLAMA_SANITIZE_THREAD=ON is broken
-
-    steps:
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get -y install \
-            build-essential \
-            xxd \
-            git \
-            cmake \
-            curl \
-            wget \
-            language-pack-en \
-            libssl-dev
-
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
-
-      - name: Python setup
-        id: setup_python
-        uses: actions/setup-python@v5
-        with:
-          python-version: '3.11'
-
-      - name: Tests dependencies
-        id: test_dependencies
-        run: |
-          pip install -r tools/server/tests/requirements.txt
-
-      - name: Setup Node.js for WebUI
-        uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-          cache: "npm"
-          cache-dependency-path: "tools/server/webui/package-lock.json"
-
-      - name: Install WebUI dependencies
-        run: npm ci
-        working-directory: tools/server/webui
-
-      - name: Build WebUI
-        run: npm run build
-        working-directory: tools/server/webui
-
-      - name: Build (no OpenMP)
-        id: cmake_build_no_openmp
-        if: ${{ matrix.sanitizer == 'THREAD' }}
-        run: |
-          cmake -B build \
-              -DGGML_NATIVE=OFF \
-              -DLLAMA_CURL=OFF \
-              -DLLAMA_OPENSSL=ON \
-              -DLLAMA_BUILD_SERVER=ON \
-              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
-              -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON \
-              -DGGML_OPENMP=OFF ;
-          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
-
-      - name: Build (sanitizers)
-        id: cmake_build_sanitizers
-        if: ${{ matrix.sanitizer != '' && matrix.sanitizer != 'THREAD' }}
-        run: |
-          cmake -B build \
-              -DGGML_NATIVE=OFF \
-              -DLLAMA_CURL=OFF \
-              -DLLAMA_OPENSSL=ON \
-              -DLLAMA_BUILD_SERVER=ON \
-              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
-              -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON ;
-          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
-
-      - name: Build (sanitizers)
-        id: cmake_build
-        if: ${{ matrix.sanitizer == '' }}
-        run: |
-          cmake -B build \
-              -DGGML_NATIVE=OFF \
-              -DLLAMA_CURL=OFF \
-              -DLLAMA_OPENSSL=ON \
-              -DLLAMA_BUILD_SERVER=ON \
-              -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} ;
-          cmake --build build --config ${{ matrix.build_type }} -j $(nproc) --target llama-server
-
-      - name: Tests
-        id: server_integration_tests
-        if: ${{ matrix.sanitizer == '' }}
-        env:
-          GITHUB_ACTIONS: "true"
-        run: |
-          cd tools/server/tests
-          ./tests.sh
-
-      - name: Tests (sanitizers)
-        id: server_integration_tests_sanitizers
-        if: ${{ matrix.sanitizer != '' }}
-        run: |
-          cd tools/server/tests
-          LLAMA_SANITIZE=1 ./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 tools/server/tests
-          SLOW_TESTS=1 ./tests.sh
-
-
   server-windows:
     runs-on: windows-2022
 

CODEOWNERS

@@ -32,7 +32,7 @@
 /examples/export-docs/                  @ggerganov
 /examples/gen-docs/                     @ggerganov
 /examples/gguf/                         @ggerganov
-/examples/llama.android/                @ggerganov
+/examples/llama.android/                @ggerganov @hanyin-arm @naco-siren
 /examples/llama.swiftui/                @ggerganov
 /examples/llama.vim                     @ggerganov
 /examples/lookahead/                    @ggerganov

README.md

@@ -190,6 +190,7 @@ Instructions for adding support for new models: [HOWTO-add-model.md](docs/develo
 - Swift [ShenghaiWang/SwiftLlama](https://github.com/ShenghaiWang/SwiftLlama)
 - Delphi [Embarcadero/llama-cpp-delphi](https://github.com/Embarcadero/llama-cpp-delphi)
 - Go (no CGo needed): [hybridgroup/yzma](https://github.com/hybridgroup/yzma)
+- Android: [llama.android](/examples/llama.android)
 
 </details>
 

SECURITY.md

@@ -68,3 +68,6 @@ Please disclose it as a private [security advisory](https://github.com/ggml-org/
 Please note that using AI to identify vulnerabilities and generate reports is permitted. However, you must (1) explicitly disclose how AI was used and (2) conduct a thorough manual review before submitting the report.
 
 A team of volunteers on a reasonable-effort basis maintains this project. As such, please give us at least 90 days to work on a fix before public exposure.
+
+> [!IMPORTANT]
+> For collaborators: if you are interested in helping out with reviewing privting security disclosures, please see: https://github.com/ggml-org/llama.cpp/discussions/18080

common/arg.cpp

@@ -420,6 +420,8 @@ static bool common_params_parse_ex(int argc, char ** argv, common_params_context
         }
     };
 
+    std::set<std::string> seen_args;
+
     for (int i = 1; i < argc; i++) {
         const std::string arg_prefix = "--";
 
@@ -430,6 +432,9 @@ static bool common_params_parse_ex(int argc, char ** argv, common_params_context
         if (arg_to_options.find(arg) == arg_to_options.end()) {
             throw std::invalid_argument(string_format("error: invalid argument: %s", arg.c_str()));
         }
+        if (!seen_args.insert(arg).second) {
+            LOG_WRN("DEPRECATED: argument '%s' specified multiple times, use comma-separated values instead (only last value will be used)\n", arg.c_str());
+        }
         auto & tmp = arg_to_options[arg];
         auto opt = *tmp.first;
         bool is_positive = tmp.second;
@@ -750,6 +755,8 @@ bool common_params_to_map(int argc, char ** argv, llama_example ex, std::map<com
         }
     };
 
+    std::set<std::string> seen_args;
+
     for (int i = 1; i < argc; i++) {
         const std::string arg_prefix = "--";
 
@@ -760,6 +767,9 @@ bool common_params_to_map(int argc, char ** argv, llama_example ex, std::map<com
         if (arg_to_options.find(arg) == arg_to_options.end()) {
             throw std::invalid_argument(string_format("error: invalid argument: %s", arg.c_str()));
         }
+        if (!seen_args.insert(arg).second) {
+            LOG_WRN("DEPRECATED: argument '%s' specified multiple times, use comma-separated values instead (only last value will be used)\n", arg.c_str());
+        }
         auto opt = *arg_to_options[arg];
         std::string val;
         if (opt.value_hint != nullptr) {
@@ -835,6 +845,19 @@ bool common_arg_utils::is_autoy(const std::string & value) {
 }
 
 common_params_context common_params_parser_init(common_params & params, llama_example ex, void(*print_usage)(int, char **)) {
+    // per-example default params
+    // we define here to make sure it's included in llama-gen-docs
+    if (ex == LLAMA_EXAMPLE_COMPLETION) {
+        params.use_jinja = false;   // disable jinja by default
+
+    } else if (ex == LLAMA_EXAMPLE_MTMD) {
+        params.use_jinja = false;   // disable jinja by default
+        params.sampling.temp = 0.2; // lower temp by default for better quality
+
+    } else if (ex == LLAMA_EXAMPLE_SERVER) {
+        params.n_parallel = -1;     // auto by default
+    }
+
     params.use_color = tty_can_use_colors();
 
     // load dynamic backends
@@ -1107,7 +1130,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_env("LLAMA_ARG_SWA_FULL"));
     add_opt(common_arg(
         {"--ctx-checkpoints", "--swa-checkpoints"}, "N",
-        string_format("max number of context checkpoints to create per slot (default: %d)\n"
+        string_format("max number of context checkpoints to create per slot (default: %d)"
             "[(more info)](https://github.com/ggml-org/llama.cpp/pull/15293)", params.n_ctx_checkpoints),
         [](common_params & params, int value) {
             params.n_ctx_checkpoints = value;
@@ -1115,7 +1138,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_env("LLAMA_ARG_CTX_CHECKPOINTS").set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_CLI}));
     add_opt(common_arg(
         {"--cache-ram", "-cram"}, "N",
-        string_format("set the maximum cache size in MiB (default: %d, -1 - no limit, 0 - disable)\n"
+        string_format("set the maximum cache size in MiB (default: %d, -1 - no limit, 0 - disable)"
             "[(more info)](https://github.com/ggml-org/llama.cpp/pull/16391)", params.cache_ram_mib),
         [](common_params & params, int value) {
             params.cache_ram_mib = value;
@@ -1123,12 +1146,11 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_env("LLAMA_ARG_CACHE_RAM").set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_CLI}));
     add_opt(common_arg(
         {"--kv-unified", "-kvu"},
-        string_format("use single unified KV buffer for the KV cache of all sequences (default: %s)\n"
-            "[(more info)](https://github.com/ggml-org/llama.cpp/pull/14363)", params.kv_unified ? "true" : "false"),
+        "use single unified KV buffer shared across all sequences (default: enabled if number of slots is auto)",
         [](common_params & params) {
             params.kv_unified = true;
         }
-    ).set_env("LLAMA_ARG_KV_UNIFIED"));
+    ).set_env("LLAMA_ARG_KV_UNIFIED").set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_PERPLEXITY}));
     add_opt(common_arg(
         {"--context-shift"},
         {"--no-context-shift"},
@@ -1214,13 +1236,15 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_examples({LLAMA_EXAMPLE_COMPLETION, LLAMA_EXAMPLE_CLI, LLAMA_EXAMPLE_DIFFUSION}));
     add_opt(common_arg(
         {"--in-file"}, "FNAME",
-        "an input file (repeat to specify multiple files)",
+        "an input file (use comma-separated values to specify multiple files)",
         [](common_params & params, const std::string & value) {
-            std::ifstream file(value);
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                std::ifstream file(item);
                 if (!file) {
-                throw std::runtime_error(string_format("error: failed to open file '%s'\n", value.c_str()));
+                    throw std::runtime_error(string_format("error: failed to open file '%s'\n", item.c_str()));
+                }
+                params.in_files.push_back(item);
             }
-            params.in_files.push_back(value);
         }
     ).set_examples({LLAMA_EXAMPLE_IMATRIX}));
     add_opt(common_arg(
@@ -1888,6 +1912,19 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             LOG_WRN("DEPRECATED: --defrag-thold is deprecated and no longer necessary to specify\n");
         }
     ).set_env("LLAMA_ARG_DEFRAG_THOLD"));
+    if (ex == LLAMA_EXAMPLE_SERVER) {
+        // this is to make sure this option appears in the server-specific section of the help message
+        add_opt(common_arg(
+            {"-np", "--parallel"}, "N",
+            string_format("number of server slots (default: %d, -1 = auto)", params.n_parallel),
+            [](common_params & params, int value) {
+                if (value == 0) {
+                    throw std::invalid_argument("error: invalid value for n_parallel\n");
+                }
+                params.n_parallel = value;
+            }
+        ).set_env("LLAMA_ARG_N_PARALLEL").set_examples({LLAMA_EXAMPLE_SERVER}));
+    } else {
         add_opt(common_arg(
             {"-np", "--parallel"}, "N",
             string_format("number of parallel sequences to decode (default: %d)", params.n_parallel),
@@ -1895,6 +1932,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
                 params.n_parallel = value;
             }
         ).set_env("LLAMA_ARG_N_PARALLEL"));
+    }
     add_opt(common_arg(
         {"-ns", "--sequences"}, "N",
         string_format("number of sequences to decode (default: %d)", params.n_sequences),
@@ -1943,9 +1981,11 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_examples(mmproj_examples).set_env("LLAMA_ARG_MMPROJ_OFFLOAD"));
     add_opt(common_arg(
         {"--image", "--audio"}, "FILE",
-        "path to an image or audio file. use with multimodal models, can be repeated if you have multiple files\n",
+        "path to an image or audio file. use with multimodal models, use comma-separated values for multiple files\n",
         [](common_params & params, const std::string & value) {
-            params.image.emplace_back(value);
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                params.image.emplace_back(item);
+            }
         }
     ).set_examples({LLAMA_EXAMPLE_MTMD, LLAMA_EXAMPLE_CLI}));
     add_opt(common_arg(
@@ -2192,12 +2232,39 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
         }
     ));
     add_opt(common_arg(
-        {"--override-kv"}, "KEY=TYPE:VALUE",
-        "advanced option to override model metadata by key. may be specified multiple times.\n"
-        "types: int, float, bool, str. example: --override-kv tokenizer.ggml.add_bos_token=bool:false",
+        {"--override-kv"}, "KEY=TYPE:VALUE,...",
+        "advanced option to override model metadata by key. to specify multiple overrides, either use comma-separated or repeat this argument.\n"
+        "types: int, float, bool, str. example: --override-kv tokenizer.ggml.add_bos_token=bool:false,tokenizer.ggml.add_eos_token=bool:false",
         [](common_params & params, const std::string & value) {
-            if (!string_parse_kv_override(value.c_str(), params.kv_overrides)) {
-                throw std::runtime_error(string_format("error: Invalid type for KV override: %s\n", value.c_str()));
+            std::vector<std::string> kv_overrides;
+
+            std::string current;
+            bool escaping = false;
+
+            for (const char c : value) {
+                if (escaping) {
+                    current.push_back(c);
+                    escaping = false;
+                } else if (c == '\\') {
+                    escaping = true;
+                } else if (c == ',') {
+                    kv_overrides.push_back(current);
+                    current.clear();
+                } else {
+                    current.push_back(c);
+                }
+            }
+
+            if (escaping) {
+                current.push_back('\\');
+            }
+
+            kv_overrides.push_back(current);
+
+            for (const auto & kv_override : kv_overrides) {
+                if (!string_parse_kv_override(kv_override.c_str(), params.kv_overrides)) {
+                    throw std::runtime_error(string_format("error: Invalid type for KV override: %s\n", kv_override.c_str()));
+                }
             }
         }
     ));
@@ -2211,33 +2278,50 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ));
     add_opt(common_arg(
         {"--lora"}, "FNAME",
-        "path to LoRA adapter (can be repeated to use multiple adapters)",
+        "path to LoRA adapter (use comma-separated values to load multiple adapters)",
         [](common_params & params, const std::string & value) {
-            params.lora_adapters.push_back({ std::string(value), 1.0, "", "", nullptr });
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                params.lora_adapters.push_back({ item, 1.0, "", "", nullptr });
+            }
         }
         // we define this arg on both COMMON and EXPORT_LORA, so when showing help message of export-lora, it will be categorized as "example-specific" arg
     ).set_examples({LLAMA_EXAMPLE_COMMON, LLAMA_EXAMPLE_EXPORT_LORA}));
     add_opt(common_arg(
-        {"--lora-scaled"}, "FNAME", "SCALE",
-        "path to LoRA adapter with user defined scaling (can be repeated to use multiple adapters)",
-        [](common_params & params, const std::string & fname, const std::string & scale) {
-            params.lora_adapters.push_back({ fname, std::stof(scale), "", "", nullptr });
+        {"--lora-scaled"}, "FNAME:SCALE,...",
+        "path to LoRA adapter with user defined scaling (format: FNAME:SCALE,...)\n"
+        "note: use comma-separated values",
+        [](common_params & params, const std::string & value) {
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                auto parts = string_split<std::string>(item, ':');
+                if (parts.size() != 2) {
+                    throw std::invalid_argument("lora-scaled format: FNAME:SCALE");
+                }
+                params.lora_adapters.push_back({ parts[0], std::stof(parts[1]), "", "", nullptr });
+            }
         }
         // we define this arg on both COMMON and EXPORT_LORA, so when showing help message of export-lora, it will be categorized as "example-specific" arg
     ).set_examples({LLAMA_EXAMPLE_COMMON, LLAMA_EXAMPLE_EXPORT_LORA}));
     add_opt(common_arg(
         {"--control-vector"}, "FNAME",
-        "add a control vector\nnote: this argument can be repeated to add multiple control vectors",
+        "add a control vector\nnote: use comma-separated values to add multiple control vectors",
         [](common_params & params, const std::string & value) {
-            params.control_vectors.push_back({ 1.0f, value, });
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                params.control_vectors.push_back({ 1.0f, item, });
+            }
         }
     ));
     add_opt(common_arg(
-        {"--control-vector-scaled"}, "FNAME", "SCALE",
+        {"--control-vector-scaled"}, "FNAME:SCALE,...",
         "add a control vector with user defined scaling SCALE\n"
-        "note: this argument can be repeated to add multiple scaled control vectors",
-        [](common_params & params, const std::string & fname, const std::string & scale) {
-            params.control_vectors.push_back({ std::stof(scale), fname });
+        "note: use comma-separated values (format: FNAME:SCALE,...)",
+        [](common_params & params, const std::string & value) {
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                auto parts = string_split<std::string>(item, ':');
+                if (parts.size() != 2) {
+                    throw std::invalid_argument("control-vector-scaled format: FNAME:SCALE");
+                }
+                params.control_vectors.push_back({ std::stof(parts[1]), parts[0] });
+            }
         }
     ));
     add_opt(common_arg(
@@ -2327,13 +2411,15 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
     ).set_env("HF_TOKEN"));
     add_opt(common_arg(
         {"--context-file"}, "FNAME",
-        "file to load context from (repeat to specify multiple files)",
+        "file to load context from (use comma-separated values to specify multiple files)",
         [](common_params & params, const std::string & value) {
-            std::ifstream file(value, std::ios::binary);
+            for (const auto & item : string_split<std::string>(value, ',')) {
+                std::ifstream file(item, std::ios::binary);
                 if (!file) {
-                throw std::runtime_error(string_format("error: failed to open file '%s'\n", value.c_str()));
+                    throw std::runtime_error(string_format("error: failed to open file '%s'\n", item.c_str()));
+                }
+                params.context_files.push_back(item);
             }
-            params.context_files.push_back(value);
         }
     ).set_examples({LLAMA_EXAMPLE_RETRIEVAL}));
     add_opt(common_arg(
@@ -2524,6 +2610,20 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             params.api_prefix = value;
         }
     ).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_API_PREFIX"));
+    add_opt(common_arg(
+        {"--webui-config"}, "JSON",
+        "JSON that provides default WebUI settings (overrides WebUI defaults)",
+        [](common_params & params, const std::string & value) {
+            params.webui_config_json = value;
+        }
+    ).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_WEBUI_CONFIG"));
+    add_opt(common_arg(
+        {"--webui-config-file"}, "PATH",
+        "JSON file that provides default WebUI settings (overrides WebUI defaults)",
+        [](common_params & params, const std::string & value) {
+            params.webui_config_json = read_file(value);
+        }
+    ).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_WEBUI_CONFIG_FILE"));
     add_opt(common_arg(
         {"--webui"},
         {"--no-webui"},

common/chat-peg-parser.cpp

@@ -4,9 +4,14 @@
 
 using json = nlohmann::json;
 
-static std::string_view trim_trailing_space(std::string_view sv) {
+static std::string_view trim_trailing_space(std::string_view sv, int max = -1) {
+    int count = 0;
     while (!sv.empty() && std::isspace(static_cast<unsigned char>(sv.back()))) {
+        if (max != -1 && count <= max) {
+            break;
+        }
         sv.remove_suffix(1);
+        count++;
     }
     return sv;
 }
@@ -93,7 +98,7 @@ void common_chat_peg_constructed_mapper::map(const common_peg_ast_node & node) {
 
     if (is_arg_string && current_tool) {
         // Serialize to JSON, but exclude the end quote
-        std::string dumped = json(node.text).dump();
+        std::string dumped = json(trim_trailing_space(node.text)).dump();
         current_tool->arguments += dumped.substr(0, dumped.size() - 1);
         needs_closing_quote = true;
     }
@@ -101,6 +106,7 @@ void common_chat_peg_constructed_mapper::map(const common_peg_ast_node & node) {
     if (is_arg_close && current_tool) {
         if (needs_closing_quote) {
             current_tool->arguments += "\"";
+            needs_closing_quote = false;
         }
     }
 
@@ -109,6 +115,10 @@ void common_chat_peg_constructed_mapper::map(const common_peg_ast_node & node) {
     }
 
     if (is_tool_close && current_tool) {
+        if (needs_closing_quote) {
+            current_tool->arguments += "\"";
+            needs_closing_quote = false;
+        }
         current_tool->arguments += "}";
     }
 }

common/chat.cpp

@@ -711,6 +711,25 @@ static void foreach_function(const json & tools, const std::function<void(const
     }
 }
 
+static void foreach_parameter(const json & function, const std::function<void(const std::string &, const json &, bool)> & fn) {
+    if (!function.contains("parameters") || !function.at("parameters").is_object()) {
+        return;
+    }
+    const auto & params = function.at("parameters");
+    if (!params.contains("properties") || !params.at("properties").is_object()) {
+        return;
+    }
+    const auto & props = params.at("properties");
+    std::set<std::string> required;
+    if (params.contains("required") && params.at("required").is_array()) {
+        params.at("required").get_to(required);
+    }
+    for (const auto & [name, prop] : props.items()) {
+        bool is_required = (required.find(name) != required.end());
+        fn(name, prop, is_required);
+    }
+}
+
 static std::string apply(
     const common_chat_template & tmpl,
     const struct templates_params & inputs,
@@ -1409,6 +1428,123 @@ static common_chat_params common_chat_params_init_nemotron_v2(const common_chat_
     return data;
 }
 
+static common_chat_params common_chat_params_init_nemotron_v3(const common_chat_template & tmpl, const struct templates_params & inputs) {
+    common_chat_params data;
+
+    data.prompt = apply(tmpl, inputs);
+    data.format = COMMON_CHAT_FORMAT_PEG_CONSTRUCTED;
+
+    // Handle thinking tags appropriately based on inputs.enable_thinking
+    if (string_ends_with(data.prompt, "<think>\n")) {
+        if (!inputs.enable_thinking) {
+            data.prompt += "</think>";
+        } else {
+            data.thinking_forced_open = true;
+        }
+    }
+
+    data.preserved_tokens = {
+        "<think>",
+        "</think>",
+        "<tool_call>",
+        "</tool_call>",
+    };
+
+    auto has_tools = inputs.tools.is_array() && !inputs.tools.empty();
+    auto extract_reasoning = inputs.reasoning_format != COMMON_REASONING_FORMAT_NONE;
+    auto include_grammar = true;
+
+    auto parser = build_chat_peg_constructed_parser([&](auto & p) {
+        auto reasoning = p.eps();
+        if (inputs.enable_thinking && extract_reasoning) {
+            auto reasoning_content = p.reasoning(p.until("</think>")) + ("</think>" | p.end());
+            if (data.thinking_forced_open) {
+                reasoning = reasoning_content;
+            }
+        }
+
+        // Response format parser
+        if (inputs.json_schema.is_object() && !inputs.json_schema.empty()) {
+            return reasoning << p.content(p.schema(p.json(), "response-format", inputs.json_schema));
+        }
+
+        // Tool call parser
+        if (has_tools && inputs.tool_choice != COMMON_CHAT_TOOL_CHOICE_NONE) {
+            auto tool_choice = p.choice();
+            foreach_function(inputs.tools, [&](const json & tool) {
+                const auto & function = tool.at("function");
+                std::string name = function.at("name");
+                auto parameters = function.at("parameters");
+
+                auto schema_info = common_schema_info();
+                schema_info.resolve_refs(parameters);
+
+                auto tool_open = "<function=" + p.tool_name(p.literal(name)) + ">\n";
+                auto tool_close = p.literal("</function>\n");
+                auto args = p.sequence();
+                auto arg_string = p.rule("xml-arg-string", p.until_one_of({
+                    "\n</parameter>",
+                    "\n<parameter=",
+                    "\n</function>"
+                }));
+
+                foreach_parameter(function, [&](const auto & param_name, const json & param_schema, bool is_required) {
+                    auto rule_name = "tool-" + name + "-arg-" + param_name;
+
+                    auto arg_open = "<parameter=" + p.tool_arg_name(p.literal(param_name)) + ">\n";
+                    auto arg_close = p.literal("</parameter>\n");
+                    auto arg_value = p.eps();
+
+                    if (schema_info.resolves_to_string(param_schema)) {
+                        arg_value = p.tool_arg_string_value(arg_string) + "\n";
+                    } else {
+                        arg_value = p.tool_arg_json_value(p.schema(p.json(), rule_name + "-schema", param_schema));
+                    }
+
+                    // Model may or my not close with </parameter>
+                    auto arg_rule = p.rule(rule_name, p.tool_arg_open(arg_open) + arg_value + p.optional(p.tool_arg_close(arg_close)));
+                    args += p.repeat(arg_rule, /* min = */ is_required ? 1 : 0, /* max = */ 1);
+                });
+
+                tool_choice |= p.rule("tool-" + name, p.tool_open(tool_open) + args + p.tool_close(tool_close));
+            });
+
+            auto min_calls = inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_REQUIRED ? 1 : 0;
+            auto max_calls = inputs.parallel_tool_calls ? -1 : 1;
+            auto tool_call = p.rule("tool-call", "<tool_call>\n" + tool_choice + "</tool_call>" + p.space());
+            auto tool_calls = p.trigger_rule("tool-call-root", p.repeat(tool_call, /* min = */ min_calls, /* max = */ max_calls));
+
+            return reasoning << p.content(p.until("<tool_call>")) << tool_calls;
+        }
+
+        // Content only parser
+        include_grammar = false;
+        return reasoning << p.content(p.rest());
+    });
+
+    data.parser = parser.save();
+
+    if (include_grammar) {
+        data.grammar_lazy = has_tools && inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_AUTO;
+
+        data.grammar = build_grammar([&](const common_grammar_builder & builder) {
+            foreach_function(inputs.tools, [&](const json & tool) {
+                const auto & function = tool.at("function");
+                auto schema = function.at("parameters");
+                builder.resolve_refs(schema);
+            });
+            parser.build_grammar(builder, data.grammar_lazy);
+        });
+
+        data.grammar_triggers = {
+            {COMMON_GRAMMAR_TRIGGER_TYPE_WORD, "<tool_call>"}
+        };
+    }
+
+    return data;
+}
+
+
 static common_chat_params common_chat_params_init_apertus(const common_chat_template & tmpl, const struct templates_params & inputs) {
     common_chat_params data;
 
@@ -2534,6 +2670,10 @@ static common_chat_params common_chat_templates_apply_jinja(
         src.find("<function=") != std::string::npos &&
         src.find("<parameters>") != std::string::npos &&
         src.find("<parameter=") != std::string::npos) {
+        // Nemotron 3 Nano 30B A3B
+        if (src.find("<think>") != std::string::npos) {
+            return common_chat_params_init_nemotron_v3(tmpl, params);
+        }
         return common_chat_params_init_qwen3_coder_xml(tmpl, params);
     }
 

common/common.cpp

@@ -1092,7 +1092,7 @@ common_init_result::common_init_result(common_params & params) :
     auto cparams = common_context_params_to_llama(params);
 
     if (params.fit_params) {
-        LOG_INF("%s: fitting params to device memory, to report bugs during this step use -fit off (or --verbose if you can't)\n", __func__);
+        LOG_INF("%s: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on\n", __func__);
         llama_params_fit(params.model.path.c_str(), &mparams, &cparams,
             params.tensor_split, params.tensor_buft_overrides.data(), params.fit_params_target, params.fit_params_min_ctx,
             params.verbosity >= 4 ? GGML_LOG_LEVEL_DEBUG : GGML_LOG_LEVEL_ERROR);

common/common.h

@@ -484,8 +484,11 @@ struct common_params {
 
     std::map<std::string, std::string> default_template_kwargs;
 
-    // "advanced" endpoints are disabled by default for better security
+    // webui configs
     bool webui = true;
+    std::string webui_config_json;
+
+    // "advanced" endpoints are disabled by default for better security
     bool endpoint_slots   = true;
     bool endpoint_props   = false; // only control POST requests, not GET
     bool endpoint_metrics = false;

common/json-schema-to-grammar.cpp

@@ -305,8 +305,9 @@ static std::string format_literal(const std::string & literal) {
 
 std::string gbnf_format_literal(const std::string & literal) { return format_literal(literal); }
 
-class SchemaConverter {
+class common_schema_converter {
 private:
+    friend class common_schema_info;
     friend std::string build_grammar(const std::function<void(const common_grammar_builder &)> & cb, const common_grammar_options & options);
     std::function<json(const std::string &)> _fetch_json;
     bool _dotall;
@@ -729,7 +730,7 @@ private:
     }
 
 public:
-    SchemaConverter(
+    common_schema_converter(
         const std::function<json(const std::string &)> & fetch_json,
         bool dotall)
           : _fetch_json(fetch_json), _dotall(dotall)
@@ -990,6 +991,134 @@ public:
     }
 };
 
+// common_schema_info implementation (pimpl)
+
+common_schema_info::common_schema_info()
+    : impl_(std::make_unique<common_schema_converter>(
+        [](const std::string &) { return json(); },
+        false)) {}
+
+common_schema_info::~common_schema_info() = default;
+
+common_schema_info::common_schema_info(common_schema_info &&) noexcept = default;
+common_schema_info & common_schema_info::operator=(common_schema_info &&) noexcept = default;
+
+void common_schema_info::resolve_refs(nlohmann::ordered_json & schema) {
+    impl_->resolve_refs(schema, "");
+}
+
+// Determines if a JSON schema can resolve to a string type through any path.
+// Some models emit raw string values rather than JSON-encoded strings for string parameters.
+// If any branch of the schema (via oneOf, anyOf, $ref, etc.) permits a string, this returns
+// true, allowing callers to handle the value as a raw string for simplicity.
+bool common_schema_info::resolves_to_string(const nlohmann::ordered_json & schema) {
+    std::unordered_set<std::string> visited_refs;
+
+    std::function<bool(const json &)> check = [&](const json & s) -> bool {
+        if (!s.is_object()) {
+            return false;
+        }
+
+        // Handle $ref
+        if (s.contains("$ref")) {
+            const std::string & ref = s["$ref"];
+            if (visited_refs.find(ref) != visited_refs.end()) {
+                // Circular reference, assume not a string to be safe
+                return false;
+            }
+            visited_refs.insert(ref);
+            auto it = impl_->_refs.find(ref);
+            if (it != impl_->_refs.end()) {
+                return check(it->second);
+            }
+            return false;
+        }
+
+        // Check type field
+        if (s.contains("type")) {
+            const json & schema_type = s["type"];
+            if (schema_type.is_string()) {
+                if (schema_type == "string") {
+                    return true;
+                }
+            } else if (schema_type.is_array()) {
+                // Type can be an array like ["string", "null"]
+                for (const auto & t : schema_type) {
+                    if (t == "string") {
+                        return true;
+                    }
+                }
+            }
+        }
+
+        // Check oneOf/anyOf - if any alternative can be a string
+        if (s.contains("oneOf")) {
+            for (const auto & alt : s["oneOf"]) {
+                if (check(alt)) {
+                    return true;
+                }
+            }
+        }
+        if (s.contains("anyOf")) {
+            for (const auto & alt : s["anyOf"]) {
+                if (check(alt)) {
+                    return true;
+                }
+            }
+        }
+
+        // Check allOf - all components must be compatible with string type
+        if (s.contains("allOf")) {
+            bool all_string = true;
+            for (const auto & component : s["allOf"]) {
+                if (!check(component)) {
+                    all_string = false;
+                    break;
+                }
+            }
+            if (all_string) {
+                return true;
+            }
+        }
+
+        // Check const - if the constant value is a string
+        if (s.contains("const")) {
+            if (s["const"].is_string()) {
+                return true;
+            }
+        }
+
+        // Check enum - if any enum value is a string
+        if (s.contains("enum")) {
+            for (const auto & val : s["enum"]) {
+                if (val.is_string()) {
+                    return true;
+                }
+            }
+        }
+
+        // String-specific keywords imply string type
+        if (s.contains("pattern") || s.contains("minLength") || s.contains("maxLength")) {
+            return true;
+        }
+
+        // Check format - many formats imply string
+        if (s.contains("format")) {
+            const std::string & fmt = s["format"];
+            if (fmt == "date" || fmt == "time" || fmt == "date-time" ||
+                fmt == "uri" || fmt == "email" || fmt == "hostname" ||
+                fmt == "ipv4" || fmt == "ipv6" || fmt == "uuid" ||
+                fmt.find("uuid") == 0) {
+                return true;
+            }
+        }
+
+        return false;
+    };
+
+    return check(schema);
+}
+
 std::string json_schema_to_grammar(const json & schema, bool force_gbnf) {
 #ifdef LLAMA_USE_LLGUIDANCE
     if (!force_gbnf) {
@@ -1006,7 +1135,7 @@ std::string json_schema_to_grammar(const json & schema, bool force_gbnf) {
 }
 
 std::string build_grammar(const std::function<void(const common_grammar_builder &)> & cb, const common_grammar_options & options) {
-    SchemaConverter converter([&](const std::string &) { return json(); }, options.dotall);
+    common_schema_converter converter([&](const std::string &) { return json(); }, options.dotall);
     common_grammar_builder builder {
         /* .add_rule = */ [&](const std::string & name, const std::string & rule) {
             return converter._add_rule(name, rule);

common/json-schema-to-grammar.h

@@ -3,11 +3,31 @@
 #include <nlohmann/json_fwd.hpp>
 
 #include <functional>
+#include <memory>
 #include <string>
 
 std::string json_schema_to_grammar(const nlohmann::ordered_json & schema,
                                    bool force_gbnf = false);
 
+class common_schema_converter;
+
+// Probes a JSON schema to extract information about its structure and type constraints.
+class common_schema_info {
+    std::unique_ptr<common_schema_converter> impl_;
+
+  public:
+    common_schema_info();
+    ~common_schema_info();
+
+    common_schema_info(const common_schema_info &) = delete;
+    common_schema_info & operator=(const common_schema_info &) = delete;
+    common_schema_info(common_schema_info &&) noexcept;
+    common_schema_info & operator=(common_schema_info &&) noexcept;
+
+    void resolve_refs(nlohmann::ordered_json & schema);
+    bool resolves_to_string(const nlohmann::ordered_json & schema);
+};
+
 struct common_grammar_builder {
     std::function<std::string(const std::string &, const std::string &)> add_rule;
     std::function<std::string(const std::string &, const nlohmann::ordered_json &)> add_schema;

common/peg-parser.cpp

@@ -425,7 +425,7 @@ struct parser_executor {
 
         if (result.need_more_input()) {
             // Propagate - need to know what child would match before negating
-            return result;
+            return common_peg_parse_result(COMMON_PEG_PARSE_RESULT_NEED_MORE_INPUT, start_pos);
         }
 
         // Child failed, so negation succeeds

common/sampling.cpp

@@ -104,10 +104,9 @@ struct ring_buffer {
 struct common_sampler {
     common_params_sampling params;
 
+    struct llama_sampler * grmr;
     struct llama_sampler * chain;
 
-    bool grammar;
-
     ring_buffer<llama_token> prev;
 
     std::vector<llama_token_data> cur;
@@ -167,15 +166,14 @@ struct common_sampler * common_sampler_init(const struct llama_model * model, co
 
     lparams.no_perf = params.no_perf;
 
+    llama_sampler * grmr = nullptr;
     llama_sampler * chain = llama_sampler_chain_init(lparams);
 
-    bool grammar = false;
     std::vector<llama_sampler *> samplers;
 
     if (params.grammar.compare(0, 11, "%llguidance") == 0) {
 #ifdef LLAMA_USE_LLGUIDANCE
-        samplers.push_back(llama_sampler_init_llg(vocab, "lark", params.grammar.c_str()));
-        grammar = true;
+        grmr = llama_sampler_init_llg(vocab, "lark", params.grammar.c_str());
 #else
         GGML_ABORT("llguidance (cmake -DLLAMA_LLGUIDANCE=ON) is not enabled");
 #endif // LLAMA_USE_LLGUIDANCE
@@ -224,15 +222,12 @@ struct common_sampler * common_sampler_init(const struct llama_model * model, co
 
         if (!params.grammar.empty()) {
              if (params.grammar_lazy) {
-                 samplers.push_back(
-                         llama_sampler_init_grammar_lazy_patterns(vocab, params.grammar.c_str(), "root",
+                 grmr = llama_sampler_init_grammar_lazy_patterns(vocab, params.grammar.c_str(), "root",
                          trigger_patterns_c.data(), trigger_patterns_c.size(),
-                             trigger_tokens.data(),     trigger_tokens.size()));
+                         trigger_tokens.data(), trigger_tokens.size());
              } else {
-                 samplers.push_back(llama_sampler_init_grammar(vocab, params.grammar.c_str(), "root"));
+                 grmr = llama_sampler_init_grammar(vocab, params.grammar.c_str(), "root");
              }
-
-             grammar = true;
         }
     }
 
@@ -303,8 +298,8 @@ struct common_sampler * common_sampler_init(const struct llama_model * model, co
 
     auto * result = new common_sampler {
         /* .params  = */ params,
+        /* .grmr    = */ grmr,
         /* .chain   = */ chain,
-        /* .grammar = */ grammar,
         /* .prev    = */ ring_buffer<llama_token>(std::max(32, params.n_prev)),
         /* .cur     = */ {},
         /* .cur_p   = */ {},
@@ -315,6 +310,7 @@ struct common_sampler * common_sampler_init(const struct llama_model * model, co
 
 void common_sampler_free(struct common_sampler * gsmpl) {
     if (gsmpl) {
+        llama_sampler_free(gsmpl->grmr);
         llama_sampler_free(gsmpl->chain);
 
         delete gsmpl;
@@ -324,24 +320,11 @@ void common_sampler_free(struct common_sampler * gsmpl) {
 void common_sampler_accept(struct common_sampler * gsmpl, llama_token token, bool accept_grammar) {
     const auto tm = gsmpl->tm();
 
-    if (gsmpl->grammar) {
-        const int n_smpl = llama_sampler_chain_n(gsmpl->chain);
+    if (gsmpl->grmr && accept_grammar) {
+        llama_sampler_accept(gsmpl->grmr, token);
+    }
 
-        for (int i = 0; i < n_smpl; i++) {
-            auto * smpl = llama_sampler_chain_get(gsmpl->chain, i);
-
-            // the grammar sampler is always the first one
-            if (i == 0) {
-                if (accept_grammar) {
-                    llama_sampler_accept(smpl, token);
-                }
-            } else {
-                llama_sampler_accept(smpl, token);
-            }
-        }
-    } else {
     llama_sampler_accept(gsmpl->chain, token);
-    }
 
     gsmpl->prev.push_back(token);
 }
@@ -353,8 +336,8 @@ void common_sampler_reset(struct common_sampler * gsmpl) {
 struct common_sampler * common_sampler_clone(common_sampler * gsmpl) {
     return new common_sampler {
         /* .params  = */ gsmpl->params,
+        /* .grmr    = */ llama_sampler_clone(gsmpl->grmr),
         /* .chain   = */ llama_sampler_clone(gsmpl->chain),
-        /* .grammar = */ gsmpl->grammar,
         /* .prev    = */ gsmpl->prev,
         /* .cur     = */ gsmpl->cur,
         /* .cur_p   = */ gsmpl->cur_p,
@@ -410,7 +393,7 @@ struct llama_sampler * common_sampler_get(const struct common_sampler * gsmpl) {
     return gsmpl->chain;
 }
 
-llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_context * ctx, int idx) {
+llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_context * ctx, int idx, bool grammar_first) {
     llama_synchronize(ctx);
 
     // start measuring sampling time after the llama_context synchronization in order to not measure any ongoing async operations
@@ -418,11 +401,42 @@ llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_co
 
     llama_token id = LLAMA_TOKEN_NULL;
 
+    auto & grmr  = gsmpl->grmr;
     auto & chain = gsmpl->chain;
     auto & cur_p = gsmpl->cur_p; // initialized by set_logits
 
     gsmpl->set_logits(ctx, idx);
 
+    if (grammar_first) {
+        llama_sampler_apply(grmr, &cur_p);
+    }
+
+    llama_sampler_apply(chain, &cur_p);
+
+    id = cur_p.data[cur_p.selected].id;
+
+    if (grammar_first) {
+        return id;
+    }
+
+    // check if it the sampled token fits the grammar (grammar-based rejection sampling)
+    {
+        llama_token_data       single_token_data       = { id, 1.0f, 0.0f };
+        llama_token_data_array single_token_data_array = { &single_token_data, 1, -1, false };
+
+        llama_sampler_apply(grmr, &single_token_data_array);
+
+        const bool is_valid = single_token_data_array.data[0].logit != -INFINITY;
+        if (is_valid) {
+            return id;
+        }
+    }
+
+    // resampling:
+    // if the token is not valid, sample again, but first apply the grammar sampler and then the sampling chain
+    gsmpl->set_logits(ctx, idx);
+
+    llama_sampler_apply(grmr,  &cur_p);
     llama_sampler_apply(chain, &cur_p);
 
     GGML_ASSERT(cur_p.selected != -1 && "no selected token during sampling - check your sampling configuration");
@@ -432,7 +446,7 @@ llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_co
     return id;
 }
 
-std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const std::vector<int> & idxs, const llama_tokens & draft) {
+std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const std::vector<int> & idxs, const llama_tokens & draft, bool grammar_first) {
     GGML_ASSERT(idxs.size() == draft.size() + 1 && "idxs.size() must be draft.size() + 1");
 
     std::vector<llama_token> result;
@@ -440,7 +454,7 @@ std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sample
 
     size_t i = 0;
     for (; i < draft.size(); i++) {
-        const llama_token id = common_sampler_sample(gsmpl, ctx, idxs[i]);
+        const llama_token id = common_sampler_sample(gsmpl, ctx, idxs[i], grammar_first);
 
         common_sampler_accept(gsmpl, id, true);
 
@@ -452,7 +466,7 @@ std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sample
     }
 
     if (i == draft.size()) {
-        const llama_token id = common_sampler_sample(gsmpl, ctx, idxs[i]);
+        const llama_token id = common_sampler_sample(gsmpl, ctx, idxs[i], grammar_first);
 
         common_sampler_accept(gsmpl, id, true);
 
@@ -462,13 +476,13 @@ std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sample
     return result;
 }
 
-std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const llama_tokens & draft) {
+std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const llama_tokens & draft, bool grammar_first) {
     std::vector<int> idxs(draft.size() + 1);
     for (size_t i = 0; i < idxs.size(); ++i) {
         idxs[i] = i;
     }
 
-    return common_sampler_sample_and_accept_n(gsmpl, ctx, idxs, draft);
+    return common_sampler_sample_and_accept_n(gsmpl, ctx, idxs, draft, grammar_first);
 }
 
 uint32_t common_sampler_get_seed(const struct common_sampler * gsmpl) {

common/sampling.h

@@ -57,7 +57,10 @@ struct llama_sampler * common_sampler_get(const struct common_sampler * gsmpl);
 // - check if the token fits the grammar (if any)
 // - if not: resample by first applying the grammar constraints and then sampling again (slower path)
 //
-llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_context * ctx, int idx);
+// if grammar_first is true, the grammar is applied before the samplers (slower)
+// useful in cases where all the resulting candidates (not just the sampled one) must fit the grammar
+//
+llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_context * ctx, int idx, bool grammar_first = false);
 
 // generalized version of common_sampler_sample
 //
@@ -75,10 +78,10 @@ llama_token common_sampler_sample(struct common_sampler * gsmpl, struct llama_co
 //
 // returns at least 1 token, up to idxs.size()
 //
-std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const std::vector<int> & idxs, const llama_tokens & draft);
+std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const std::vector<int> & idxs, const llama_tokens & draft, bool grammar_first = false);
 
 // assume idxs == [ 0, 1, 2, ..., draft.size() ]
-std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const llama_tokens & draft);
+std::vector<llama_token> common_sampler_sample_and_accept_n(struct common_sampler * gsmpl, struct llama_context * ctx, const llama_tokens & draft, bool grammar_first = false);
 
 uint32_t common_sampler_get_seed(const struct common_sampler * gsmpl);
 

common/speculative.cpp

@@ -315,7 +315,7 @@ llama_tokens common_speculative_gen_draft(
     for (int i = 0; i < params.n_draft; ++i) {
         common_batch_clear(batch);
 
-        common_sampler_sample(smpl, ctx_dft, 0);
+        common_sampler_sample(smpl, ctx_dft, 0, true);
 
         const auto * cur_p = common_sampler_get_candidates(smpl, true);
 

convert_hf_to_gguf.py

@@ -136,19 +136,11 @@ class ModelBase:
         self.remote_hf_model_id = remote_hf_model_id
         self.sentence_transformers_dense_modules = sentence_transformers_dense_modules
         self.hparams = ModelBase.load_hparams(self.dir_model, self.is_mistral_format) if hparams is None else hparams
-        self.rope_parameters = self.hparams.get("rope_parameters", self.hparams.get("rope_scaling")) or {}
         self.model_tensors = self.index_tensors(remote_hf_model_id=remote_hf_model_id)
         self.metadata_override = metadata_override
         self.model_name = model_name
         self.dir_model_card = dir_model  # overridden in convert_lora_to_gguf.py
 
-        # Ensure "rope_theta" and "rope_type" is mirrored in rope_parameters
-        if "full_attention" not in self.rope_parameters and "sliding_attention" not in self.rope_parameters:
-            if "rope_theta" not in self.rope_parameters and (rope_theta := self.find_hparam(["rope_theta", "global_rope_theta", "rotary_emb_base"], optional=True)) is not None:
-                self.rope_parameters["rope_theta"] = rope_theta
-            if "rope_type" not in self.rope_parameters and (rope_type := self.rope_parameters.get("type")) is not None:
-                self.rope_parameters["rope_type"] = rope_type
-
         # Apply heuristics to figure out typical tensor encoding based on first layer tensor encoding type
         if self.ftype == gguf.LlamaFileType.GUESSED:
             # NOTE: can't use field "torch_dtype" in config.json, because some finetunes lie.
@@ -765,6 +757,15 @@ class TextModel(ModelBase):
         self.block_count = self.find_hparam(["n_layers", "num_hidden_layers", "n_layer", "num_layers"])
         self.tensor_map = gguf.get_tensor_name_map(self.model_arch, self.block_count)
 
+        self.rope_parameters = self.hparams.get("rope_parameters", self.hparams.get("rope_scaling")) or {}
+
+        # Ensure "rope_theta" and "rope_type" is mirrored in rope_parameters
+        if "full_attention" not in self.rope_parameters and "sliding_attention" not in self.rope_parameters:
+            if "rope_theta" not in self.rope_parameters and (rope_theta := self.find_hparam(["rope_theta", "global_rope_theta", "rotary_emb_base"], optional=True)) is not None:
+                self.rope_parameters["rope_theta"] = rope_theta
+            if "rope_type" not in self.rope_parameters and (rope_type := self.rope_parameters.get("type")) is not None:
+                self.rope_parameters["rope_type"] = rope_type
+
     @classmethod
     def __init_subclass__(cls):
         # can't use an abstract property, because overriding it without type errors
@@ -861,6 +862,14 @@ class TextModel(ModelBase):
                 logger.warning(f"Unknown RoPE type: {rope_type}")
             logger.info(f"gguf: rope scaling type = {rope_gguf_type.name}")
 
+        if "mrope_section" in self.rope_parameters:
+            mrope_section = self.rope_parameters["mrope_section"]
+            # Pad to 4 dimensions [time, height, width, extra]
+            while len(mrope_section) < 4:
+                mrope_section.append(0)
+            self.gguf_writer.add_rope_dimension_sections(mrope_section[:4])
+            logger.info(f"gguf: mrope sections: {mrope_section[:4]}")
+
         if (rope_theta := rope_params.get("rope_theta")) is not None:
             self.gguf_writer.add_rope_freq_base(rope_theta)
             logger.info(f"gguf: rope theta = {rope_theta}")
@@ -1829,7 +1838,7 @@ class MmprojModel(ModelBase):
 
     def tensor_force_quant(self, name, new_name, bid, n_dims):
         del bid, name, n_dims  # unused
-        if ".patch_embd.weight" in new_name:
+        if ".patch_embd.weight" in new_name or ".patch_merger.weight" in new_name:
             return gguf.GGMLQuantizationType.F16 if self.ftype == gguf.LlamaFileType.MOSTLY_F16 else gguf.GGMLQuantizationType.F32
         return False
 
@@ -3738,9 +3747,6 @@ class Qwen2VLModel(TextModel):
 
     def set_gguf_parameters(self):
         super().set_gguf_parameters()
-        mrope_section = self.hparams["rope_scaling"]["mrope_section"]
-        mrope_section += [0] * max(0, 4 - len(mrope_section))
-        self.gguf_writer.add_rope_dimension_sections(mrope_section)
 
     def set_vocab(self):
         try:
@@ -4376,6 +4382,30 @@ class Qwen3VLVisionModel(MmprojModel):
         return super().modify_tensors(data_torch, name, bid)
 
 
+@ModelBase.register("Glm4vForConditionalGeneration", "Glm4vMoeForConditionalGeneration")
+class Glm4VVisionModel(Qwen3VLVisionModel):
+    def set_gguf_parameters(self):
+        MmprojModel.set_gguf_parameters(self) # skip Qwen3VLVisionModel parameters
+        assert self.hparams_vision is not None
+        self.gguf_writer.add_clip_projector_type(gguf.VisionProjectorType.GLM4V)
+
+        hidden_act = str(self.hparams_vision.get("hidden_act", "")).lower()
+        if hidden_act == "gelu":
+            self.gguf_writer.add_vision_use_gelu(True)
+        elif hidden_act == "silu":
+            self.gguf_writer.add_vision_use_silu(True)
+
+        rms_norm_eps = self.hparams_vision.get("rms_norm_eps", 1e-5)
+        self.gguf_writer.add_vision_attention_layernorm_eps(rms_norm_eps)
+
+    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
+        if name.startswith("model.visual."):
+            name = name.replace("model.visual.", "visual.")
+        if name.startswith("visual.merger."):
+            return [(self.map_tensor_name(name), data_torch)]
+        return super().modify_tensors(data_torch, name, bid)
+
+
 @ModelBase.register("Qwen3VLForConditionalGeneration")
 class Qwen3VLTextModel(Qwen3Model):
     model_arch = gguf.MODEL_ARCH.QWEN3VL
@@ -4384,20 +4414,6 @@ class Qwen3VLTextModel(Qwen3Model):
         super().set_gguf_parameters()
 
         # Handle MRoPE (Multi-axis Rotary Position Embedding) for Qwen3-VL
-        text_config = self.hparams.get("text_config", {})
-        # rope_scaling is deprecated in V5, use rope_parameters instead
-        rope_scaling = text_config.get("rope_scaling") or text_config.get("rope_parameters") or {}
-
-        if rope_scaling.get("mrope_section"):
-            # mrope_section contains [time, height, width] dimensions
-            mrope_section = rope_scaling["mrope_section"]
-            # Pad to 4 dimensions [time, height, width, extra]
-            while len(mrope_section) < 4:
-                mrope_section.append(0)
-            self.gguf_writer.add_rope_dimension_sections(mrope_section[:4])
-
-            logger.info(f"MRoPE sections: {mrope_section[:4]}")
-
         vision_config = self.hparams.get("vision_config", {})
         deepstack_layer_num = len(vision_config.get("deepstack_visual_indexes", []))
         self.gguf_writer.add_num_deepstack_layers(deepstack_layer_num)
@@ -4416,22 +4432,6 @@ class Qwen3VLMoeTextModel(Qwen3MoeModel):
 
     def set_gguf_parameters(self):
         super().set_gguf_parameters()
-
-        # Handle MRoPE (Multi-axis Rotary Position Embedding) for Qwen3-VL
-        text_config = self.hparams.get("text_config", {})
-        # rope_scaling is deprecated in V5, use rope_parameters instead
-        rope_scaling = text_config.get("rope_scaling") or text_config.get("rope_parameters") or {}
-
-        if rope_scaling.get("mrope_section"):
-            # mrope_section contains [time, height, width] dimensions
-            mrope_section = rope_scaling["mrope_section"]
-            # Pad to 4 dimensions [time, height, width, extra]
-            while len(mrope_section) < 4:
-                mrope_section.append(0)
-            self.gguf_writer.add_rope_dimension_sections(mrope_section[:4])
-
-            logger.info(f"MRoPE sections: {mrope_section[:4]}")
-
         vision_config = self.hparams.get("vision_config", {})
         deepstack_layer_num = len(vision_config.get("deepstack_visual_indexes", []))
         self.gguf_writer.add_num_deepstack_layers(deepstack_layer_num)
@@ -7794,6 +7794,15 @@ class JaisModel(TextModel):
 @ModelBase.register("Glm4ForCausalLM", "Glm4vForConditionalGeneration")
 class Glm4Model(TextModel):
     model_arch = gguf.MODEL_ARCH.GLM4
+    use_mrope = False
+    partial_rotary_factor = 0.5
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.partial_rotary_factor = self.rope_parameters.get("partial_rotary_factor", 0.5)
+        if "mrope_section" in self.rope_parameters:
+            self.use_mrope = True
+            logger.info("Q/K weight will need to be permuted for M-RoPE")
 
     def set_vocab(self):
         from transformers import AutoTokenizer
@@ -7815,17 +7824,49 @@ class Glm4Model(TextModel):
         super().set_gguf_parameters()
         if (rope_dim := self.hparams.get("head_dim")) is None:
             rope_dim = self.hparams["hidden_size"] // self.hparams["num_attention_heads"]
-        self.gguf_writer.add_rope_dimension_count(int(rope_dim * self.hparams.get("partial_rotary_factor", 0.5)))
+        self.gguf_writer.add_rope_dimension_count(int(rope_dim * self.partial_rotary_factor))
+
+    @staticmethod
+    def normal_to_neox(weights: Tensor, n_head: int, n_head_kv: int, head_dim: int, partial_rotary_factor: float) -> Tensor:
+        orig_shape = weights.shape
+        if len(orig_shape) == 1:
+            weights = weights.unsqueeze(1)  # [out_dim, 1]
+        if len(weights.shape) != 2:
+            raise ValueError("Only 1D and 2D tensors are supported.")
+        n_effective_heads = weights.shape[0] // head_dim
+        if n_head_kv is not None and n_effective_heads != n_head:
+            if n_effective_heads != n_head_kv:
+                raise AssertionError(f"Mismatch in effective heads: computed {n_effective_heads}, expected {n_head} or {n_head_kv}")
+        rotary_dim = int(head_dim * partial_rotary_factor)
+        if rotary_dim % 2 != 0:
+            raise ValueError("rotary_dim must be even.")
+        reshaped = weights.reshape(n_effective_heads, head_dim, -1)
+        rot_part = reshaped[:, :rotary_dim, :]
+        non_rot_part = reshaped[:, rotary_dim:, :]
+        permuted_rot = torch.cat((rot_part[:, ::2, :], rot_part[:, 1::2, :]), dim=1)
+        combined = torch.cat((permuted_rot, non_rot_part), dim=1)
+        result = combined.reshape(weights.shape)
+        return result if len(orig_shape) != 1 else result.squeeze(1)
 
     def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
         if name.startswith("model.visual."): # ignore visual part of Glm4v
             return []
         elif name.startswith("model.language_model."):
             name = name.replace("language_model.", "") # for Glm4v
+        if self.use_mrope:
+            n_head = self.hparams["num_attention_heads"]
+            n_kv_head = self.hparams["num_key_value_heads"]
+            n_embd = self.hparams["hidden_size"]
+            head_dim = n_embd // n_head
+            # because llama.cpp M-RoPE kernel only supports Neox ordering, we have to permute the weights here
+            if name.endswith(("q_proj.weight", "q_proj.bias")):
+                data_torch = Glm4Model.normal_to_neox(data_torch, n_head, n_head, head_dim, self.partial_rotary_factor)
+            if name.endswith(("k_proj.weight", "k_proj.bias")):
+                data_torch = Glm4Model.normal_to_neox(data_torch, n_head, n_kv_head, head_dim, self.partial_rotary_factor)
         return super().modify_tensors(data_torch, name, bid)
 
 
-@ModelBase.register("Glm4MoeForCausalLM")
+@ModelBase.register("Glm4MoeForCausalLM", "Glm4vMoeForConditionalGeneration")
 class Glm4MoeModel(TextModel):
     model_arch = gguf.MODEL_ARCH.GLM4_MOE
 
@@ -7892,6 +7933,7 @@ class Glm4MoeModel(TextModel):
 
     _experts: list[dict[str, Tensor]] | None = None
 
+    # note: unlike GLM4V non-MoE, we don't need to permute Q/K here since GLM4V_MOE uses Neox ordering already
     def modify_tensors(
         self, data_torch: Tensor, name: str, bid: int | None
     ) -> Iterable[tuple[str, Tensor]]:
@@ -8489,8 +8531,18 @@ class GraniteHybridModel(Mamba2Model, GraniteMoeModel):
 class NemotronHModel(GraniteHybridModel):
     """Hybrid mamba2/attention model from NVIDIA"""
     model_arch = gguf.MODEL_ARCH.NEMOTRON_H
+    is_moe: bool = False
 
     def __init__(self, *args, **kwargs):
+        # We have to determine the correct model architecture (MoE vs non-MoE) before
+        # calling the parent __init__. This is because the parent constructor
+        # uses self.model_arch to build the tensor name map, and all MoE-specific
+        # mappings would be missed if it were called with the default non-MoE arch.
+        hparams = ModelBase.load_hparams(args[0], self.is_mistral_format)
+        if "num_experts_per_tok" in hparams:
+            self.model_arch = gguf.MODEL_ARCH.NEMOTRON_H_MOE
+            self.is_moe = True
+
         super().__init__(*args, **kwargs)
 
         # Save the top-level head_dim for later
@@ -8502,9 +8554,11 @@ class NemotronHModel(GraniteHybridModel):
 
         # Update the ssm / attn / mlp layers
         # M: Mamba2, *: Attention, -: MLP
+        # MoE:
+        # M: Mamba2, *: Attention, E: Expert
         hybrid_override_pattern = self.hparams["hybrid_override_pattern"]
         self._ssm_layers = [i for i, val in enumerate(hybrid_override_pattern) if val == "M"]
-        self._mlp_layers = [i for i, val in enumerate(hybrid_override_pattern) if val == "-"]
+        self._mlp_layers = [i for i, val in enumerate(hybrid_override_pattern) if val == ("E" if self.is_moe else "-")]
 
     def get_attn_layers(self):
         hybrid_override_pattern = self.hparams["hybrid_override_pattern"]
@@ -8520,10 +8574,28 @@ class NemotronHModel(GraniteHybridModel):
         # Set feed_forward_length
         # NOTE: This will trigger an override warning. This is preferrable to
         #   duplicating all the parent logic
+        if not self.is_moe:
             n_ff = self.find_hparam(["intermediate_size", "n_inner", "hidden_dim"])
             self.gguf_writer.add_feed_forward_length([
                 n_ff if i in self._mlp_layers else 0 for i in range(self.block_count)
             ])
+        else:
+            moe_intermediate_size = self.hparams["moe_intermediate_size"]
+            self.gguf_writer.add_feed_forward_length([
+                moe_intermediate_size if i in self._mlp_layers else 0 for i in range(self.block_count)
+            ])
+            self.gguf_writer.add_expert_used_count(self.hparams["num_experts_per_tok"])
+            self.gguf_writer.add_expert_feed_forward_length(self.hparams["moe_intermediate_size"])
+            self.gguf_writer.add_expert_shared_feed_forward_length(self.hparams["moe_shared_expert_intermediate_size"])
+            self.gguf_writer.add_expert_count(self.hparams["n_routed_experts"])
+            self.gguf_writer.add_expert_shared_count(self.hparams["n_shared_experts"])
+            self.gguf_writer.add_expert_weights_norm(self.hparams["norm_topk_prob"])
+            self.gguf_writer.add_expert_weights_scale(self.hparams["routed_scaling_factor"])
+            self.gguf_writer.add_expert_group_count(self.hparams["n_group"])
+
+            # number of experts used per token (top-k)
+            if (n_experts_used := self.hparams.get("num_experts_per_tok")) is not None:
+                self.gguf_writer.add_expert_used_count(n_experts_used)
 
     def set_vocab(self):
         super().set_vocab()
@@ -8531,8 +8603,82 @@ class NemotronHModel(GraniteHybridModel):
         # The tokenizer _does_ add a BOS token (via post_processor type
         # TemplateProcessing) but does not set add_bos_token to true in the
         # config, so we need to explicitly override it here.
+        if not self.is_moe:
             self.gguf_writer.add_add_bos_token(True)
 
+    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
+        if self.is_moe and bid is not None:
+            if name.endswith("mixer.gate.e_score_correction_bias"):
+                new_name = name.replace("e_score_correction_bias", "e_score_correction.bias")
+                mapped_name = self.map_tensor_name(new_name)
+                return [(mapped_name, data_torch)]
+
+            if name.endswith("mixer.dt_bias"):
+                new_name = name.replace("dt_bias", "dt.bias")
+                mapped_name = self.map_tensor_name(new_name)
+                return [(mapped_name, data_torch)]
+
+            if name.endswith("mixer.conv1d.weight"):
+                squeezed_data = data_torch.squeeze()
+                mapped_name = self.map_tensor_name(name)
+                return [(mapped_name, squeezed_data)]
+
+            if name.endswith("mixer.A_log"):
+                transformed_data = -torch.exp(data_torch)
+                reshaped_data = transformed_data.squeeze().reshape(-1, 1)
+                mapped_name = self.map_tensor_name(name)
+                return [(mapped_name, reshaped_data)]
+
+            if name.endswith("mixer.D"):
+                reshaped_data = data_torch.squeeze().reshape(-1, 1)
+                mapped_name = self.map_tensor_name(name)
+                return [(mapped_name, reshaped_data)]
+
+            if name.endswith("mixer.norm.weight"):
+                reshaped_data = data_torch.reshape(8, 512)
+                mapped_name = self.map_tensor_name(name)
+                return [(mapped_name, reshaped_data)]
+
+            if name.find("mixer.experts") != -1:
+                n_experts = self.hparams["n_routed_experts"]
+                assert bid is not None
+
+                if self._experts is None:
+                    self._experts = [{} for _ in range(self.block_count)]
+
+                self._experts[bid][name] = data_torch
+
+                if len(self._experts[bid]) >= n_experts * 2:
+                    # merge the experts into a single tensor
+                    tensors: list[tuple[str, Tensor]] = []
+                    for w_name in ["down_proj", "up_proj"]:
+                        datas: list[Tensor] = []
+
+                        for xid in range(n_experts):
+                            ename = f"backbone.layers.{bid}.mixer.experts.{xid}.{w_name}.weight"
+                            datas.append(self._experts[bid][ename])
+                            del self._experts[bid][ename]
+
+                        data_torch = torch.stack(datas, dim=0)
+                        merged_name = f"model.layers.{bid}.mlp.experts.{w_name}.weight"
+                        new_name = self.map_tensor_name(merged_name)
+                        tensors.append((new_name, data_torch))
+
+                    return tensors
+                else:
+                    return []
+
+        return super().modify_tensors(data_torch, name, bid)
+
+    def prepare_tensors(self):
+        super().prepare_tensors()
+
+        if self._experts is not None:
+            # flatten `list[dict[str, Tensor]]` into `list[str]`
+            experts = [k for d in self._experts for k in d.keys()]
+            if len(experts) > 0:
+                raise ValueError(f"Unprocessed experts: {experts}")
+
 
 @ModelBase.register("BailingMoeForCausalLM")
 class BailingMoeModel(TextModel):

docs/android.md

@@ -1,6 +1,26 @@
 
 # Android
 
+## Build with Android Studio
+
+Import the `examples/llama.android` directory into Android Studio, then perform a Gradle sync and build the project.
+![Project imported into Android Studio](./android/imported-into-android-studio.png)
+
+This Android binding supports hardware acceleration up to `SME2` for **Arm** and `AMX` for **x86-64** CPUs on Android and ChromeOS devices.
+It automatically detects the host's hardware to load compatible kernels. As a result, it runs seamlessly on both the latest premium devices and older devices that may lack modern CPU features or have limited RAM, without requiring any manual configuration.
+
+A minimal Android app frontend is included to showcase the binding’s core functionalities:
+1.	**Parse GGUF metadata** via `GgufMetadataReader` from either a `ContentResolver` provided `Uri` or a local `File`.
+2.	**Obtain a `TierDetection` or `InferenceEngine`** instance through the high-level facade APIs.
+3.	**Send a raw user prompt** for automatic template formatting, prefill, and decoding. Then collect the generated tokens in a Kotlin `Flow`.
+
+For a production-ready experience that leverages advanced features such as system prompts and benchmarks, check out [Arm AI Chat](https://play.google.com/store/apps/details?id=com.arm.aichat) on Google Play.
+This project is made possible through a collaborative effort by Arm's **CT-ML**, **CE-ML** and **STE** groups:
+
+| ![Home screen](./android/arm-ai-chat-home-screen.png)  | ![System prompt](./android/system-prompt-setup.png)  | !["Haiku"](./android/chat-with-system-prompt-haiku.png)  |
+|:------------------------------------------------------:|:----------------------------------------------------:|:--------------------------------------------------------:|
+|                      Home screen                       |                    System prompt                     |                         "Haiku"                          |
+
 ## Build on Android using Termux
 
 [Termux](https://termux.dev/en/) is an Android terminal emulator and Linux environment app (no root required). As of writing, Termux is available experimentally in the Google Play Store; otherwise, it may be obtained directly from the project repo or on F-Droid.

docs/backend/SYCL.md

@@ -103,6 +103,8 @@ SYCL backend supports Intel GPU Family:
 - Intel Built-in Arc GPU
 - Intel iGPU in Core CPU (11th Generation Core CPU and newer, refer to [oneAPI supported GPU](https://www.intel.com/content/www/us/en/developer/articles/system-requirements/intel-oneapi-base-toolkit-system-requirements.html#inpage-nav-1-1)).
 
+On older Intel GPUs, you may try [OpenCL](/docs/backend/OPENCL.md) although the performance is not optimal, and some GPUs may not support OpenCL nor have any GPGPU capabilities.
+
 #### Verified devices
 
 | Intel GPU                     | Status  | Verified Model                        |

docs/development/HOWTO-add-model.md

@@ -97,7 +97,7 @@ The model params and tensors layout must be defined in `llama.cpp` source files:
 1. Define a new `llm_arch` enum value in `src/llama-arch.h`.
 2. In `src/llama-arch.cpp`:
     - Add the architecture name to the `LLM_ARCH_NAMES` map.
-    - Add the tensor mappings to the `LLM_TENSOR_NAMES` map.
+    - Add the list of model tensors to `llm_get_tensor_names` (you may also need to update `LLM_TENSOR_NAMES`)
 3. Add any non-standard metadata loading in the `llama_model_loader` constructor in `src/llama-model-loader.cpp`.
 4. If the model has a RoPE operation, add a case for the architecture in `llama_model_rope_type` function in `src/llama-model.cpp`.
 

docs/docker.md

@@ -7,9 +7,9 @@
 ## Images
 We have three Docker images available for this project:
 
-1. `ghcr.io/ggml-org/llama.cpp:full`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization. (platforms: `linux/amd64`, `linux/arm64`, `linux/s390x`)
-2. `ghcr.io/ggml-org/llama.cpp:light`: This image only includes the main executable file. (platforms: `linux/amd64`, `linux/arm64`, `linux/s390x`)
-3. `ghcr.io/ggml-org/llama.cpp:server`: This image only includes the server executable file. (platforms: `linux/amd64`, `linux/arm64`, `linux/s390x`)
+1. `ghcr.io/ggml-org/llama.cpp:full`: This image includes both the `llama-cli` and `llama-completion` executables and the tools to convert LLaMA models into ggml and convert into 4-bit quantization. (platforms: `linux/amd64`, `linux/arm64`, `linux/s390x`)
+2. `ghcr.io/ggml-org/llama.cpp:light`: This image only includes the `llama-cli` and `llama-completion` executables. (platforms: `linux/amd64`, `linux/arm64`, `linux/s390x`)
+3. `ghcr.io/ggml-org/llama.cpp:server`: This image only includes the `llama-server` executable. (platforms: `linux/amd64`, `linux/arm64`, `linux/s390x`)
 
 Additionally, there the following images, similar to the above:
 
@@ -44,13 +44,15 @@ docker run -v /path/to/models:/models ghcr.io/ggml-org/llama.cpp:full --all-in-o
 On completion, you are ready to play!
 
 ```bash
-docker run -v /path/to/models:/models ghcr.io/ggml-org/llama.cpp:full --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512
+docker run -v /path/to/models:/models ghcr.io/ggml-org/llama.cpp:full --run -m /models/7B/ggml-model-q4_0.gguf
+docker run -v /path/to/models:/models ghcr.io/ggml-org/llama.cpp:full --run-legacy -m /models/32B/ggml-model-q8_0.gguf -no-cnv -p "Building a mobile app can be done in 15 steps:" -n 512
 ```
 
 or with a light image:
 
 ```bash
-docker run -v /path/to/models:/models ghcr.io/ggml-org/llama.cpp:light -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512
+docker run -v /path/to/models:/models --entrypoint /app/llama-cli ghcr.io/ggml-org/llama.cpp:light -m /models/7B/ggml-model-q4_0.gguf
+docker run -v /path/to/models:/models --entrypoint /app/llama-completion ghcr.io/ggml-org/llama.cpp:light -m /models/32B/ggml-model-q8_0.gguf -no-cnv -p "Building a mobile app can be done in 15 steps:" -n 512
 ```
 
 or with a server image:
@@ -59,6 +61,8 @@ or with a server image:
 docker run -v /path/to/models:/models -p 8080:8080 ghcr.io/ggml-org/llama.cpp:server -m /models/7B/ggml-model-q4_0.gguf --port 8080 --host 0.0.0.0 -n 512
 ```
 
+In the above examples, `--entrypoint /app/llama-cli` is specified for clarity, but you can safely omit it since it's the default entrypoint in the container.
+
 ## Docker With CUDA
 
 Assuming one has the [nvidia-container-toolkit](https://github.com/NVIDIA/nvidia-container-toolkit) properly installed on Linux, or is using a GPU enabled cloud, `cuBLAS` should be accessible inside the container.
@@ -80,9 +84,9 @@ The defaults are:
 
 The resulting images, are essentially the same as the non-CUDA images:
 
-1. `local/llama.cpp:full-cuda`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization.
-2. `local/llama.cpp:light-cuda`: This image only includes the main executable file.
-3. `local/llama.cpp:server-cuda`: This image only includes the server executable file.
+1. `local/llama.cpp:full-cuda`: This image includes both the `llama-cli` and `llama-completion` executables and the tools to convert LLaMA models into ggml and convert into 4-bit quantization.
+2. `local/llama.cpp:light-cuda`: This image only includes the `llama-cli` and `llama-completion` executables.
+3. `local/llama.cpp:server-cuda`: This image only includes the `llama-server` executable.
 
 ## Usage
 
@@ -114,9 +118,9 @@ The defaults are:
 
 The resulting images, are essentially the same as the non-MUSA images:
 
-1. `local/llama.cpp:full-musa`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization.
-2. `local/llama.cpp:light-musa`: This image only includes the main executable file.
-3. `local/llama.cpp:server-musa`: This image only includes the server executable file.
+1. `local/llama.cpp:full-musa`: This image includes both the `llama-cli` and `llama-completion` executables and the tools to convert LLaMA models into ggml and convert into 4-bit quantization.
+2. `local/llama.cpp:light-musa`: This image only includes the `llama-cli` and `llama-completion` executables.
+3. `local/llama.cpp:server-musa`: This image only includes the `llama-server` executable.
 
 ## Usage
 

examples/gen-docs/gen-docs.cpp

@@ -48,7 +48,7 @@ static void write_table(std::ofstream & file, std::vector<common_arg *> & opts)
     }
 }
 
-static void export_md(std::string fname, llama_example ex) {
+static void export_md(std::string fname, llama_example ex, std::string name) {
     std::ofstream file(fname, std::ofstream::out | std::ofstream::trunc);
 
     common_params params;
@@ -72,13 +72,14 @@ static void export_md(std::string fname, llama_example ex) {
     write_table(file, common_options);
     file << "\n\n**Sampling params**\n\n";
     write_table(file, sparam_options);
-    file << "\n\n**Example-specific params**\n\n";
+    file << "\n\n**" << name << "-specific params**\n\n";
     write_table(file, specific_options);
 }
 
 int main(int, char **) {
-    export_md("autogen-main.md", LLAMA_EXAMPLE_COMPLETION);
-    export_md("autogen-server.md", LLAMA_EXAMPLE_SERVER);
+    // TODO: add CLI
+    export_md("autogen-completion.md", LLAMA_EXAMPLE_COMPLETION, "Tool");
+    export_md("autogen-server.md", LLAMA_EXAMPLE_SERVER, "Server");
 
     return 0;
 }

examples/llama.android/app/build.gradle.kts

@@ -1,16 +1,18 @@
 plugins {
-    id("com.android.application")
-    id("org.jetbrains.kotlin.android")
+    alias(libs.plugins.android.application)
+    alias(libs.plugins.jetbrains.kotlin.android)
 }
 
 android {
     namespace = "com.example.llama"
-    compileSdk = 34
+    compileSdk = 36
 
     defaultConfig {
-        applicationId = "com.example.llama"
+        applicationId = "com.example.llama.aichat"
+
         minSdk = 33
-        targetSdk = 34
+        targetSdk = 36
+
         versionCode = 1
         versionName = "1.0"
 
@@ -21,8 +23,17 @@ android {
     }
 
     buildTypes {
+        debug {
+            isMinifyEnabled = true
+            isShrinkResources = true
+            proguardFiles(
+                getDefaultProguardFile("proguard-android.txt"),
+                "proguard-rules.pro"
+            )
+        }
         release {
-            isMinifyEnabled = false
+            isMinifyEnabled = true
+            isShrinkResources = true
             proguardFiles(
                 getDefaultProguardFile("proguard-android-optimize.txt"),
                 "proguard-rules.pro"
@@ -36,30 +47,15 @@ android {
     kotlinOptions {
         jvmTarget = "1.8"
     }
-    buildFeatures {
-        compose = true
-    }
-    composeOptions {
-        kotlinCompilerExtensionVersion = "1.5.1"
-    }
 }
 
 dependencies {
+    implementation(libs.bundles.androidx)
+    implementation(libs.material)
 
-    implementation("androidx.core:core-ktx:1.12.0")
-    implementation("androidx.lifecycle:lifecycle-runtime-ktx:2.6.2")
-    implementation("androidx.activity:activity-compose:1.8.2")
-    implementation(platform("androidx.compose:compose-bom:2023.08.00"))
-    implementation("androidx.compose.ui:ui")
-    implementation("androidx.compose.ui:ui-graphics")
-    implementation("androidx.compose.ui:ui-tooling-preview")
-    implementation("androidx.compose.material3:material3")
-    implementation(project(":llama"))
-    testImplementation("junit:junit:4.13.2")
-    androidTestImplementation("androidx.test.ext:junit:1.1.5")
-    androidTestImplementation("androidx.test.espresso:espresso-core:3.5.1")
-    androidTestImplementation(platform("androidx.compose:compose-bom:2023.08.00"))
-    androidTestImplementation("androidx.compose.ui:ui-test-junit4")
-    debugImplementation("androidx.compose.ui:ui-tooling")
-    debugImplementation("androidx.compose.ui:ui-test-manifest")
+    implementation(project(":lib"))
+
+    testImplementation(libs.junit)
+    androidTestImplementation(libs.androidx.junit)
+    androidTestImplementation(libs.androidx.espresso.core)
 }

examples/llama.android/app/proguard-rules.pro

@@ -19,3 +19,11 @@
 # If you keep the line number information, uncomment this to
 # hide the original source file name.
 #-renamesourcefileattribute SourceFile
+
+-keep class com.arm.aichat.* { *; }
+-keep class com.arm.aichat.gguf.* { *; }
+
+-assumenosideeffects class android.util.Log {
+    public static int v(...);
+    public static int d(...);
+}

examples/llama.android/app/src/main/AndroidManifest.xml

@@ -1,24 +1,21 @@
 <?xml version="1.0" encoding="utf-8"?>
-<manifest xmlns:android="http://schemas.android.com/apk/res/android"
-    xmlns:tools="http://schemas.android.com/tools">
-
-    <uses-permission android:name="android.permission.INTERNET" />
+<manifest xmlns:android="http://schemas.android.com/apk/res/android">
 
     <application
         android:allowBackup="true"
         android:dataExtractionRules="@xml/data_extraction_rules"
+        android:extractNativeLibs="true"
         android:fullBackupContent="@xml/backup_rules"
-        android:icon="@mipmap/ic_launcher"
+        android:icon="@mipmap/ic_launcher_round"
         android:label="@string/app_name"
         android:roundIcon="@mipmap/ic_launcher_round"
         android:supportsRtl="true"
-        android:theme="@style/Theme.LlamaAndroid"
+        android:theme="@style/Theme.AiChatSample"
         >
 
         <activity
             android:name=".MainActivity"
-            android:exported="true"
-            android:theme="@style/Theme.LlamaAndroid">
+            android:exported="true">
             <intent-filter>
                 <action android:name="android.intent.action.MAIN" />
 

examples/llama.android/app/src/main/java/com/example/llama/Downloadable.kt

@@ -1,119 +0,0 @@
-package com.example.llama
-
-import android.app.DownloadManager
-import android.net.Uri
-import android.util.Log
-import androidx.compose.material3.Button
-import androidx.compose.material3.Text
-import androidx.compose.runtime.Composable
-import androidx.compose.runtime.getValue
-import androidx.compose.runtime.mutableDoubleStateOf
-import androidx.compose.runtime.mutableStateOf
-import androidx.compose.runtime.remember
-import androidx.compose.runtime.rememberCoroutineScope
-import androidx.compose.runtime.setValue
-import androidx.core.database.getLongOrNull
-import androidx.core.net.toUri
-import kotlinx.coroutines.delay
-import kotlinx.coroutines.launch
-import java.io.File
-
-data class Downloadable(val name: String, val source: Uri, val destination: File) {
-    companion object {
-        @JvmStatic
-        private val tag: String? = this::class.qualifiedName
-
-        sealed interface State
-        data object Ready: State
-        data class Downloading(val id: Long): State
-        data class Downloaded(val downloadable: Downloadable): State
-        data class Error(val message: String): State
-
-        @JvmStatic
-        @Composable
-        fun Button(viewModel: MainViewModel, dm: DownloadManager, item: Downloadable) {
-            var status: State by remember {
-                mutableStateOf(
-                    if (item.destination.exists()) Downloaded(item)
-                    else Ready
-                )
-            }
-            var progress by remember { mutableDoubleStateOf(0.0) }
-
-            val coroutineScope = rememberCoroutineScope()
-
-            suspend fun waitForDownload(result: Downloading, item: Downloadable): State {
-                while (true) {
-                    val cursor = dm.query(DownloadManager.Query().setFilterById(result.id))
-
-                    if (cursor == null) {
-                        Log.e(tag, "dm.query() returned null")
-                        return Error("dm.query() returned null")
-                    }
-
-                    if (!cursor.moveToFirst() || cursor.count < 1) {
-                        cursor.close()
-                        Log.i(tag, "cursor.moveToFirst() returned false or cursor.count < 1, download canceled?")
-                        return Ready
-                    }
-
-                    val pix = cursor.getColumnIndex(DownloadManager.COLUMN_BYTES_DOWNLOADED_SO_FAR)
-                    val tix = cursor.getColumnIndex(DownloadManager.COLUMN_TOTAL_SIZE_BYTES)
-                    val sofar = cursor.getLongOrNull(pix) ?: 0
-                    val total = cursor.getLongOrNull(tix) ?: 1
-                    cursor.close()
-
-                    if (sofar == total) {
-                        return Downloaded(item)
-                    }
-
-                    progress = (sofar * 1.0) / total
-
-                    delay(1000L)
-                }
-            }
-
-            fun onClick() {
-                when (val s = status) {
-                    is Downloaded -> {
-                        viewModel.load(item.destination.path)
-                    }
-
-                    is Downloading -> {
-                        coroutineScope.launch {
-                            status = waitForDownload(s, item)
-                        }
-                    }
-
-                    else -> {
-                        item.destination.delete()
-
-                        val request = DownloadManager.Request(item.source).apply {
-                            setTitle("Downloading model")
-                            setDescription("Downloading model: ${item.name}")
-                            setAllowedNetworkTypes(DownloadManager.Request.NETWORK_WIFI)
-                            setDestinationUri(item.destination.toUri())
-                        }
-
-                        viewModel.log("Saving ${item.name} to ${item.destination.path}")
-                        Log.i(tag, "Saving ${item.name} to ${item.destination.path}")
-
-                        val id = dm.enqueue(request)
-                        status = Downloading(id)
-                        onClick()
-                    }
-                }
-            }
-
-            Button(onClick = { onClick() }, enabled = status !is Downloading) {
-                when (status) {
-                    is Downloading -> Text(text = "Downloading ${(progress * 100).toInt()}%")
-                    is Downloaded -> Text("Load ${item.name}")
-                    is Ready -> Text("Download ${item.name}")
-                    is Error -> Text("Download ${item.name}")
-                }
-            }
-        }
-
-    }
-}

examples/llama.android/app/src/main/java/com/example/llama/MainActivity.kt

@@ -1,154 +1,257 @@
 package com.example.llama
 
-import android.app.ActivityManager
-import android.app.DownloadManager
-import android.content.ClipData
-import android.content.ClipboardManager
 import android.net.Uri
 import android.os.Bundle
-import android.os.StrictMode
-import android.os.StrictMode.VmPolicy
-import android.text.format.Formatter
-import androidx.activity.ComponentActivity
-import androidx.activity.compose.setContent
-import androidx.activity.viewModels
-import androidx.compose.foundation.layout.Box
-import androidx.compose.foundation.layout.Column
-import androidx.compose.foundation.layout.Row
-import androidx.compose.foundation.layout.fillMaxSize
-import androidx.compose.foundation.layout.padding
-import androidx.compose.foundation.lazy.LazyColumn
-import androidx.compose.foundation.lazy.items
-import androidx.compose.foundation.lazy.rememberLazyListState
-import androidx.compose.material3.Button
-import androidx.compose.material3.LocalContentColor
-import androidx.compose.material3.MaterialTheme
-import androidx.compose.material3.OutlinedTextField
-import androidx.compose.material3.Surface
-import androidx.compose.material3.Text
-import androidx.compose.runtime.Composable
-import androidx.compose.ui.Modifier
-import androidx.compose.ui.unit.dp
-import androidx.core.content.getSystemService
-import com.example.llama.ui.theme.LlamaAndroidTheme
+import android.util.Log
+import android.widget.EditText
+import android.widget.TextView
+import android.widget.Toast
+import androidx.activity.enableEdgeToEdge
+import androidx.activity.result.contract.ActivityResultContracts
+import androidx.appcompat.app.AppCompatActivity
+import androidx.lifecycle.lifecycleScope
+import androidx.recyclerview.widget.LinearLayoutManager
+import androidx.recyclerview.widget.RecyclerView
+import com.arm.aichat.AiChat
+import com.arm.aichat.InferenceEngine
+import com.arm.aichat.gguf.GgufMetadata
+import com.arm.aichat.gguf.GgufMetadataReader
+import com.google.android.material.floatingactionbutton.FloatingActionButton
+import kotlinx.coroutines.Dispatchers
+import kotlinx.coroutines.flow.onCompletion
+import kotlinx.coroutines.launch
+import kotlinx.coroutines.withContext
 import java.io.File
+import java.io.FileOutputStream
+import java.io.InputStream
+import java.util.UUID
 
-class MainActivity(
-    activityManager: ActivityManager? = null,
-    downloadManager: DownloadManager? = null,
-    clipboardManager: ClipboardManager? = null,
-): ComponentActivity() {
-    private val tag: String? = this::class.simpleName
+class MainActivity : AppCompatActivity() {
 
-    private val activityManager by lazy { activityManager ?: getSystemService<ActivityManager>()!! }
-    private val downloadManager by lazy { downloadManager ?: getSystemService<DownloadManager>()!! }
-    private val clipboardManager by lazy { clipboardManager ?: getSystemService<ClipboardManager>()!! }
+    // Android views
+    private lateinit var ggufTv: TextView
+    private lateinit var messagesRv: RecyclerView
+    private lateinit var userInputEt: EditText
+    private lateinit var userActionFab: FloatingActionButton
 
-    private val viewModel: MainViewModel by viewModels()
+    // Arm AI Chat inference engine
+    private lateinit var engine: InferenceEngine
 
-    // Get a MemoryInfo object for the device's current memory status.
-    private fun availableMemory(): ActivityManager.MemoryInfo {
-        return ActivityManager.MemoryInfo().also { memoryInfo ->
-            activityManager.getMemoryInfo(memoryInfo)
-        }
-    }
+    // Conversation states
+    private var isModelReady = false
+    private val messages = mutableListOf<Message>()
+    private val lastAssistantMsg = StringBuilder()
+    private val messageAdapter = MessageAdapter(messages)
 
     override fun onCreate(savedInstanceState: Bundle?) {
         super.onCreate(savedInstanceState)
+        enableEdgeToEdge()
+        setContentView(R.layout.activity_main)
 
-        StrictMode.setVmPolicy(
-            VmPolicy.Builder(StrictMode.getVmPolicy())
-                .detectLeakedClosableObjects()
-                .build()
-        )
+        // Find views
+        ggufTv = findViewById(R.id.gguf)
+        messagesRv = findViewById(R.id.messages)
+        messagesRv.layoutManager = LinearLayoutManager(this)
+        messagesRv.adapter = messageAdapter
+        userInputEt = findViewById(R.id.user_input)
+        userActionFab = findViewById(R.id.fab)
 
-        val free = Formatter.formatFileSize(this, availableMemory().availMem)
-        val total = Formatter.formatFileSize(this, availableMemory().totalMem)
-
-        viewModel.log("Current memory: $free / $total")
-        viewModel.log("Downloads directory: ${getExternalFilesDir(null)}")
-
-        val extFilesDir = getExternalFilesDir(null)
-
-        val models = listOf(
-            Downloadable(
-                "Phi-2 7B (Q4_0, 1.6 GiB)",
-                Uri.parse("https://huggingface.co/ggml-org/models/resolve/main/phi-2/ggml-model-q4_0.gguf?download=true"),
-                File(extFilesDir, "phi-2-q4_0.gguf"),
-            ),
-            Downloadable(
-                "TinyLlama 1.1B (f16, 2.2 GiB)",
-                Uri.parse("https://huggingface.co/ggml-org/models/resolve/main/tinyllama-1.1b/ggml-model-f16.gguf?download=true"),
-                File(extFilesDir, "tinyllama-1.1-f16.gguf"),
-            ),
-            Downloadable(
-                "Phi 2 DPO (Q3_K_M, 1.48 GiB)",
-                Uri.parse("https://huggingface.co/TheBloke/phi-2-dpo-GGUF/resolve/main/phi-2-dpo.Q3_K_M.gguf?download=true"),
-                File(extFilesDir, "phi-2-dpo.Q3_K_M.gguf")
-            ),
-        )
-
-        setContent {
-            LlamaAndroidTheme {
-                // A surface container using the 'background' color from the theme
-                Surface(
-                    modifier = Modifier.fillMaxSize(),
-                    color = MaterialTheme.colorScheme.background
-                ) {
-                    MainCompose(
-                        viewModel,
-                        clipboardManager,
-                        downloadManager,
-                        models,
-                    )
+        // Arm AI Chat initialization
+        lifecycleScope.launch(Dispatchers.Default) {
+            engine = AiChat.getInferenceEngine(applicationContext)
         }
 
+        // Upon CTA button tapped
+        userActionFab.setOnClickListener {
+            if (isModelReady) {
+                // If model is ready, validate input and send to engine
+                handleUserInput()
+            } else {
+                // Otherwise, prompt user to select a GGUF metadata on the device
+                getContent.launch(arrayOf("*/*"))
+            }
+        }
+    }
+
+    private val getContent = registerForActivityResult(
+        ActivityResultContracts.OpenDocument()
+    ) { uri ->
+        Log.i(TAG, "Selected file uri:\n $uri")
+        uri?.let { handleSelectedModel(it) }
+    }
+
+    /**
+     * Handles the file Uri from [getContent] result
+     */
+    private fun handleSelectedModel(uri: Uri) {
+        // Update UI states
+        userActionFab.isEnabled = false
+        userInputEt.hint = "Parsing GGUF..."
+        ggufTv.text = "Parsing metadata from selected file \n$uri"
+
+        lifecycleScope.launch(Dispatchers.IO) {
+            // Parse GGUF metadata
+            Log.i(TAG, "Parsing GGUF metadata...")
+            contentResolver.openInputStream(uri)?.use {
+                GgufMetadataReader.create().readStructuredMetadata(it)
+            }?.let { metadata ->
+                // Update UI to show GGUF metadata to user
+                Log.i(TAG, "GGUF parsed: \n$metadata")
+                withContext(Dispatchers.Main) {
+                    ggufTv.text = metadata.toString()
+                }
+
+                // Ensure the model file is available
+                val modelName = metadata.filename() + FILE_EXTENSION_GGUF
+                contentResolver.openInputStream(uri)?.use { input ->
+                    ensureModelFile(modelName, input)
+                }?.let { modelFile ->
+                    loadModel(modelName, modelFile)
+
+                    withContext(Dispatchers.Main) {
+                        isModelReady = true
+                        userInputEt.hint = "Type and send a message!"
+                        userInputEt.isEnabled = true
+                        userActionFab.setImageResource(R.drawable.outline_send_24)
+                        userActionFab.isEnabled = true
+                    }
                 }
             }
         }
     }
 
-@Composable
-fun MainCompose(
-    viewModel: MainViewModel,
-    clipboard: ClipboardManager,
-    dm: DownloadManager,
-    models: List<Downloadable>
-) {
-    Column {
-        val scrollState = rememberLazyListState()
-
-        Box(modifier = Modifier.weight(1f)) {
-            LazyColumn(state = scrollState) {
-                items(viewModel.messages) {
-                    Text(
-                        it,
-                        style = MaterialTheme.typography.bodyLarge.copy(color = LocalContentColor.current),
-                        modifier = Modifier.padding(16.dp)
-                    )
-                }
-            }
-        }
-        OutlinedTextField(
-            value = viewModel.message,
-            onValueChange = { viewModel.updateMessage(it) },
-            label = { Text("Message") },
-        )
-        Row {
-            Button({ viewModel.send() }) { Text("Send") }
-            Button({ viewModel.bench(8, 4, 1) }) { Text("Bench") }
-            Button({ viewModel.clear() }) { Text("Clear") }
-            Button({
-                viewModel.messages.joinToString("\n").let {
-                    clipboard.setPrimaryClip(ClipData.newPlainText("", it))
-                }
-            }) { Text("Copy") }
+    /**
+     * Prepare the model file within app's private storage
+     */
+    private suspend fun ensureModelFile(modelName: String, input: InputStream) =
+        withContext(Dispatchers.IO) {
+            File(ensureModelsDirectory(), modelName).also { file ->
+                // Copy the file into local storage if not yet done
+                if (!file.exists()) {
+                    Log.i(TAG, "Start copying file to $modelName")
+                    withContext(Dispatchers.Main) {
+                        userInputEt.hint = "Copying file..."
                     }
 
-        Column {
-            for (model in models) {
-                Downloadable.Button(viewModel, dm, model)
+                    FileOutputStream(file).use { input.copyTo(it) }
+                    Log.i(TAG, "Finished copying file to $modelName")
+                } else {
+                    Log.i(TAG, "File already exists $modelName")
+                }
+            }
+        }
+
+    /**
+     * Load the model file from the app private storage
+     */
+    private suspend fun loadModel(modelName: String, modelFile: File) =
+        withContext(Dispatchers.IO) {
+            Log.i(TAG, "Loading model $modelName")
+            withContext(Dispatchers.Main) {
+                userInputEt.hint = "Loading model..."
+            }
+            engine.loadModel(modelFile.path)
+        }
+
+    /**
+     * Validate and send the user message into [InferenceEngine]
+     */
+    private fun handleUserInput() {
+        userInputEt.text.toString().also { userSsg ->
+            if (userSsg.isEmpty()) {
+                Toast.makeText(this, "Input message is empty!", Toast.LENGTH_SHORT).show()
+            } else {
+                userInputEt.text = null
+                userActionFab.isEnabled = false
+
+                // Update message states
+                messages.add(Message(UUID.randomUUID().toString(), userSsg, true))
+                lastAssistantMsg.clear()
+                messages.add(Message(UUID.randomUUID().toString(), lastAssistantMsg.toString(), false))
+
+                lifecycleScope.launch(Dispatchers.Default) {
+                    engine.sendUserPrompt(userSsg)
+                        .onCompletion {
+                            withContext(Dispatchers.Main) {
+                                userActionFab.isEnabled = true
+                            }
+                        }.collect { token ->
+                            val messageCount = messages.size
+                            check(messageCount > 0 && !messages[messageCount - 1].isUser)
+
+                            messages.removeAt(messageCount - 1).copy(
+                                content = lastAssistantMsg.append(token).toString()
+                            ).let { messages.add(it) }
+
+                            withContext(Dispatchers.Main) {
+                                messageAdapter.notifyItemChanged(messages.size - 1)
                             }
                         }
                 }
             }
+        }
+    }
+
+    /**
+     * Run a benchmark with the model file
+     */
+    private suspend fun runBenchmark(modelName: String, modelFile: File) =
+        withContext(Dispatchers.Default) {
+            Log.i(TAG, "Starts benchmarking $modelName")
+            withContext(Dispatchers.Main) {
+                userInputEt.hint = "Running benchmark..."
+            }
+            engine.bench(
+                pp=BENCH_PROMPT_PROCESSING_TOKENS,
+                tg=BENCH_TOKEN_GENERATION_TOKENS,
+                pl=BENCH_SEQUENCE,
+                nr=BENCH_REPETITION
+            ).let { result ->
+                messages.add(Message(UUID.randomUUID().toString(), result, false))
+                withContext(Dispatchers.Main) {
+                    messageAdapter.notifyItemChanged(messages.size - 1)
+                }
+            }
+        }
+
+    /**
+     * Create the `models` directory if not exist.
+     */
+    private fun ensureModelsDirectory() =
+        File(filesDir, DIRECTORY_MODELS).also {
+            if (it.exists() && !it.isDirectory) { it.delete() }
+            if (!it.exists()) { it.mkdir() }
+        }
+
+    companion object {
+        private val TAG = MainActivity::class.java.simpleName
+
+        private const val DIRECTORY_MODELS = "models"
+        private const val FILE_EXTENSION_GGUF = ".gguf"
+
+        private const val BENCH_PROMPT_PROCESSING_TOKENS = 512
+        private const val BENCH_TOKEN_GENERATION_TOKENS = 128
+        private const val BENCH_SEQUENCE = 1
+        private const val BENCH_REPETITION = 3
+    }
+}
+
+fun GgufMetadata.filename() = when {
+    basic.name != null -> {
+        basic.name?.let { name ->
+            basic.sizeLabel?.let { size ->
+                "$name-$size"
+            } ?: name
+        }
+    }
+    architecture?.architecture != null -> {
+        architecture?.architecture?.let { arch ->
+            basic.uuid?.let { uuid ->
+                "$arch-$uuid"
+            } ?: "$arch-${System.currentTimeMillis()}"
+        }
+    }
+    else -> {
+        "model-${System.currentTimeMillis().toHexString()}"
+    }
+}

examples/llama.android/app/src/main/java/com/example/llama/MainViewModel.kt

@@ -1,105 +0,0 @@
-package com.example.llama
-
-import android.llama.cpp.LLamaAndroid
-import android.util.Log
-import androidx.compose.runtime.getValue
-import androidx.compose.runtime.mutableStateOf
-import androidx.compose.runtime.setValue
-import androidx.lifecycle.ViewModel
-import androidx.lifecycle.viewModelScope
-import kotlinx.coroutines.flow.catch
-import kotlinx.coroutines.launch
-
-class MainViewModel(private val llamaAndroid: LLamaAndroid = LLamaAndroid.instance()): ViewModel() {
-    companion object {
-        @JvmStatic
-        private val NanosPerSecond = 1_000_000_000.0
-    }
-
-    private val tag: String? = this::class.simpleName
-
-    var messages by mutableStateOf(listOf("Initializing..."))
-        private set
-
-    var message by mutableStateOf("")
-        private set
-
-    override fun onCleared() {
-        super.onCleared()
-
-        viewModelScope.launch {
-            try {
-                llamaAndroid.unload()
-            } catch (exc: IllegalStateException) {
-                messages += exc.message!!
-            }
-        }
-    }
-
-    fun send() {
-        val text = message
-        message = ""
-
-        // Add to messages console.
-        messages += text
-        messages += ""
-
-        viewModelScope.launch {
-            llamaAndroid.send(text)
-                .catch {
-                    Log.e(tag, "send() failed", it)
-                    messages += it.message!!
-                }
-                .collect { messages = messages.dropLast(1) + (messages.last() + it) }
-        }
-    }
-
-    fun bench(pp: Int, tg: Int, pl: Int, nr: Int = 1) {
-        viewModelScope.launch {
-            try {
-                val start = System.nanoTime()
-                val warmupResult = llamaAndroid.bench(pp, tg, pl, nr)
-                val end = System.nanoTime()
-
-                messages += warmupResult
-
-                val warmup = (end - start).toDouble() / NanosPerSecond
-                messages += "Warm up time: $warmup seconds, please wait..."
-
-                if (warmup > 5.0) {
-                    messages += "Warm up took too long, aborting benchmark"
-                    return@launch
-                }
-
-                messages += llamaAndroid.bench(512, 128, 1, 3)
-            } catch (exc: IllegalStateException) {
-                Log.e(tag, "bench() failed", exc)
-                messages += exc.message!!
-            }
-        }
-    }
-
-    fun load(pathToModel: String) {
-        viewModelScope.launch {
-            try {
-                llamaAndroid.load(pathToModel)
-                messages += "Loaded $pathToModel"
-            } catch (exc: IllegalStateException) {
-                Log.e(tag, "load() failed", exc)
-                messages += exc.message!!
-            }
-        }
-    }
-
-    fun updateMessage(newMessage: String) {
-        message = newMessage
-    }
-
-    fun clear() {
-        messages = listOf()
-    }
-
-    fun log(message: String) {
-        messages += message
-    }
-}

examples/llama.android/app/src/main/java/com/example/llama/MessageAdapter.kt

@@ -0,0 +1,51 @@
+package com.example.llama
+
+import android.view.LayoutInflater
+import android.view.View
+import android.view.ViewGroup
+import android.widget.TextView
+import androidx.recyclerview.widget.RecyclerView
+
+data class Message(
+    val id: String,
+    val content: String,
+    val isUser: Boolean
+)
+
+class MessageAdapter(
+    private val messages: List<Message>
+) : RecyclerView.Adapter<RecyclerView.ViewHolder>() {
+
+    companion object {
+        private const val VIEW_TYPE_USER = 1
+        private const val VIEW_TYPE_ASSISTANT = 2
+    }
+
+    override fun getItemViewType(position: Int): Int {
+        return if (messages[position].isUser) VIEW_TYPE_USER else VIEW_TYPE_ASSISTANT
+    }
+
+    override fun onCreateViewHolder(parent: ViewGroup, viewType: Int): RecyclerView.ViewHolder {
+        val layoutInflater = LayoutInflater.from(parent.context)
+        return if (viewType == VIEW_TYPE_USER) {
+            val view = layoutInflater.inflate(R.layout.item_message_user, parent, false)
+            UserMessageViewHolder(view)
+        } else {
+            val view = layoutInflater.inflate(R.layout.item_message_assistant, parent, false)
+            AssistantMessageViewHolder(view)
+        }
+    }
+
+    override fun onBindViewHolder(holder: RecyclerView.ViewHolder, position: Int) {
+        val message = messages[position]
+        if (holder is UserMessageViewHolder || holder is AssistantMessageViewHolder) {
+            val textView = holder.itemView.findViewById<TextView>(R.id.msg_content)
+            textView.text = message.content
+        }
+    }
+
+    override fun getItemCount(): Int = messages.size
+
+    class UserMessageViewHolder(view: View) : RecyclerView.ViewHolder(view)
+    class AssistantMessageViewHolder(view: View) : RecyclerView.ViewHolder(view)
+}

examples/llama.android/app/src/main/java/com/example/llama/ui/theme/Color.kt

@@ -1,11 +0,0 @@
-package com.example.llama.ui.theme
-
-import androidx.compose.ui.graphics.Color
-
-val Purple80 = Color(0xFFD0BCFF)
-val PurpleGrey80 = Color(0xFFCCC2DC)
-val Pink80 = Color(0xFFEFB8C8)
-
-val Purple40 = Color(0xFF6650a4)
-val PurpleGrey40 = Color(0xFF625b71)
-val Pink40 = Color(0xFF7D5260)

examples/llama.android/app/src/main/java/com/example/llama/ui/theme/Theme.kt

@@ -1,70 +0,0 @@
-package com.example.llama.ui.theme
-
-import android.app.Activity
-import android.os.Build
-import androidx.compose.foundation.isSystemInDarkTheme
-import androidx.compose.material3.MaterialTheme
-import androidx.compose.material3.darkColorScheme
-import androidx.compose.material3.dynamicDarkColorScheme
-import androidx.compose.material3.dynamicLightColorScheme
-import androidx.compose.material3.lightColorScheme
-import androidx.compose.runtime.Composable
-import androidx.compose.runtime.SideEffect
-import androidx.compose.ui.graphics.toArgb
-import androidx.compose.ui.platform.LocalContext
-import androidx.compose.ui.platform.LocalView
-import androidx.core.view.WindowCompat
-
-private val DarkColorScheme = darkColorScheme(
-    primary = Purple80,
-    secondary = PurpleGrey80,
-    tertiary = Pink80
-)
-
-private val LightColorScheme = lightColorScheme(
-    primary = Purple40,
-    secondary = PurpleGrey40,
-    tertiary = Pink40
-
-    /* Other default colors to override
-    background = Color(0xFFFFFBFE),
-    surface = Color(0xFFFFFBFE),
-    onPrimary = Color.White,
-    onSecondary = Color.White,
-    onTertiary = Color.White,
-    onBackground = Color(0xFF1C1B1F),
-    onSurface = Color(0xFF1C1B1F),
-    */
-)
-
-@Composable
-fun LlamaAndroidTheme(
-    darkTheme: Boolean = isSystemInDarkTheme(),
-    // Dynamic color is available on Android 12+
-    dynamicColor: Boolean = true,
-    content: @Composable () -> Unit
-) {
-    val colorScheme = when {
-        dynamicColor && Build.VERSION.SDK_INT >= Build.VERSION_CODES.S -> {
-            val context = LocalContext.current
-            if (darkTheme) dynamicDarkColorScheme(context) else dynamicLightColorScheme(context)
-        }
-
-        darkTheme -> DarkColorScheme
-        else -> LightColorScheme
-    }
-    val view = LocalView.current
-    if (!view.isInEditMode) {
-        SideEffect {
-            val window = (view.context as Activity).window
-            window.statusBarColor = colorScheme.primary.toArgb()
-            WindowCompat.getInsetsController(window, view).isAppearanceLightStatusBars = darkTheme
-        }
-    }
-
-    MaterialTheme(
-        colorScheme = colorScheme,
-        typography = Typography,
-        content = content
-    )
-}

examples/llama.android/app/src/main/java/com/example/llama/ui/theme/Type.kt

@@ -1,34 +0,0 @@
-package com.example.llama.ui.theme
-
-import androidx.compose.material3.Typography
-import androidx.compose.ui.text.TextStyle
-import androidx.compose.ui.text.font.FontFamily
-import androidx.compose.ui.text.font.FontWeight
-import androidx.compose.ui.unit.sp
-
-// Set of Material typography styles to start with
-val Typography = Typography(
-    bodyLarge = TextStyle(
-        fontFamily = FontFamily.Default,
-        fontWeight = FontWeight.Normal,
-        fontSize = 16.sp,
-        lineHeight = 24.sp,
-        letterSpacing = 0.5.sp
-    )
-    /* Other default text styles to override
-    titleLarge = TextStyle(
-        fontFamily = FontFamily.Default,
-        fontWeight = FontWeight.Normal,
-        fontSize = 22.sp,
-        lineHeight = 28.sp,
-        letterSpacing = 0.sp
-    ),
-    labelSmall = TextStyle(
-        fontFamily = FontFamily.Default,
-        fontWeight = FontWeight.Medium,
-        fontSize = 11.sp,
-        lineHeight = 16.sp,
-        letterSpacing = 0.5.sp
-    )
-    */
-)

examples/llama.android/app/src/main/res/drawable/bg_assistant_message.xml

@@ -0,0 +1,4 @@
+<shape xmlns:android="http://schemas.android.com/apk/res/android" android:shape="rectangle">
+    <solid android:color="#E5E5EA" />
+    <corners android:radius="16dp" />
+</shape>

examples/llama.android/app/src/main/res/drawable/bg_user_message.xml

@@ -0,0 +1,4 @@
+<shape xmlns:android="http://schemas.android.com/apk/res/android" android:shape="rectangle">
+    <solid android:color="#4285F4" />
+    <corners android:radius="16dp" />
+</shape>

examples/llama.android/app/src/main/res/drawable/outline_folder_open_24.xml

@@ -0,0 +1,10 @@
+<vector xmlns:android="http://schemas.android.com/apk/res/android"
+    android:width="24dp"
+    android:height="24dp"
+    android:viewportWidth="24"
+    android:viewportHeight="24"
+    android:tint="?attr/colorControlNormal">
+  <path
+      android:fillColor="@android:color/white"
+      android:pathData="M20,6h-8l-2,-2L4,4c-1.1,0 -1.99,0.9 -1.99,2L2,18c0,1.1 0.9,2 2,2h16c1.1,0 2,-0.9 2,-2L22,8c0,-1.1 -0.9,-2 -2,-2zM20,18L4,18L4,8h16v10z"/>
+</vector>

examples/llama.android/app/src/main/res/drawable/outline_send_24.xml

@@ -0,0 +1,11 @@
+<vector xmlns:android="http://schemas.android.com/apk/res/android"
+    android:width="24dp"
+    android:height="24dp"
+    android:viewportWidth="24"
+    android:viewportHeight="24"
+    android:tint="?attr/colorControlNormal"
+    android:autoMirrored="true">
+  <path
+      android:fillColor="@android:color/white"
+      android:pathData="M4.01,6.03l7.51,3.22 -7.52,-1 0.01,-2.22m7.5,8.72L4,17.97v-2.22l7.51,-1M2.01,3L2,10l15,2 -15,2 0.01,7L23,12 2.01,3z"/>
+</vector>

examples/llama.android/app/src/main/res/layout/activity_main.xml

@@ -0,0 +1,76 @@
+<?xml version="1.0" encoding="utf-8"?>
+<androidx.constraintlayout.widget.ConstraintLayout xmlns:android="http://schemas.android.com/apk/res/android"
+    xmlns:app="http://schemas.android.com/apk/res-auto"
+        xmlns:tools="http://schemas.android.com/tools"
+        android:id="@+id/main"
+        android:layout_height="match_parent"
+        android:layout_width="match_parent">
+
+    <LinearLayout
+        android:fitsSystemWindows="true"
+        android:layout_width="match_parent"
+        android:layout_height="match_parent"
+        android:orientation="vertical"
+        tools:context=".MainActivity">
+
+        <FrameLayout
+            android:layout_width="match_parent"
+            android:layout_height="0dp"
+            android:layout_weight="1">
+
+            <ScrollView
+                android:layout_width="match_parent"
+                android:layout_height="wrap_content"
+                android:fadeScrollbars="false">
+
+                <TextView
+                    android:id="@+id/gguf"
+                    android:layout_width="match_parent"
+                    android:layout_height="wrap_content"
+                    android:layout_margin="16dp"
+                    android:text="Selected GGUF model's metadata will show here."
+                    style="@style/TextAppearance.MaterialComponents.Body2"
+                    android:maxLines="100" />
+
+            </ScrollView>
+
+        </FrameLayout>
+
+        <androidx.recyclerview.widget.RecyclerView
+            android:id="@+id/messages"
+            android:layout_width="match_parent"
+            android:layout_height="0dp"
+            android:layout_weight="4"
+            android:padding="16dp"
+            android:fadeScrollbars="false"
+            app:reverseLayout="true"
+            tools:listitem="@layout/item_message_assistant"/>
+
+        <LinearLayout
+            android:layout_width="match_parent"
+            android:layout_height="wrap_content"
+            android:orientation="horizontal">
+
+            <EditText
+                android:id="@+id/user_input"
+                android:enabled="false"
+                android:layout_width="0dp"
+                android:layout_weight="1"
+                android:layout_height="match_parent"
+                android:padding="8dp"
+                style="@style/TextAppearance.MaterialComponents.Body2"
+                android:hint="Please first pick a GGUF model file to import." />
+
+            <com.google.android.material.floatingactionbutton.FloatingActionButton
+                android:id="@+id/fab"
+                android:enabled="true"
+                style="@style/Widget.Material3.FloatingActionButton.Primary"
+                android:layout_width="wrap_content"
+                android:layout_height="wrap_content"
+                android:layout_margin="8dp"
+                android:src="@drawable/outline_folder_open_24" />
+
+        </LinearLayout>
+
+    </LinearLayout>
+</androidx.constraintlayout.widget.ConstraintLayout>

examples/llama.android/app/src/main/res/layout/item_message_assistant.xml

@@ -0,0 +1,15 @@
+<?xml version="1.0" encoding="utf-8"?>
+<LinearLayout xmlns:android="http://schemas.android.com/apk/res/android"
+    android:layout_width="match_parent"
+    android:layout_height="wrap_content"
+    android:padding="8dp"
+    android:gravity="start">
+
+    <TextView
+        android:id="@+id/msg_content"
+        android:layout_width="wrap_content"
+        android:layout_height="wrap_content"
+        android:background="@drawable/bg_assistant_message"
+        android:padding="12dp"
+        android:textColor="@android:color/black" />
+</LinearLayout>

examples/llama.android/app/src/main/res/layout/item_message_user.xml

@@ -0,0 +1,15 @@
+<?xml version="1.0" encoding="utf-8"?>
+<LinearLayout xmlns:android="http://schemas.android.com/apk/res/android"
+    android:layout_width="match_parent"
+    android:layout_height="wrap_content"
+    android:padding="8dp"
+    android:gravity="end">
+
+    <TextView
+        android:id="@+id/msg_content"
+        android:layout_width="wrap_content"
+        android:layout_height="wrap_content"
+        android:background="@drawable/bg_user_message"
+        android:padding="12dp"
+        android:textColor="@android:color/white" />
+</LinearLayout>

examples/llama.android/app/src/main/res/values/strings.xml

@@ -1,3 +1,3 @@
 <resources>
-    <string name="app_name">LlamaAndroid</string>
+    <string name="app_name">AI Chat basic sample</string>
 </resources>

examples/llama.android/app/src/main/res/values/themes.xml

@@ -1,5 +1,10 @@
 <?xml version="1.0" encoding="utf-8"?>
 <resources>
 
-    <style name="Theme.LlamaAndroid" parent="android:Theme.Material.Light.NoActionBar" />
+    <style name="Base.Theme.AiChatSample" parent="Theme.Material3.DayNight.NoActionBar">
+        <!-- Customize your light theme here. -->
+        <!-- <item name="colorPrimary">@color/my_light_primary</item> -->
+    </style>
+
+    <style name="Theme.AiChatSample" parent="Base.Theme.AiChatSample" />
 </resources>

examples/llama.android/build.gradle.kts

@@ -1,6 +1,6 @@
 // Top-level build file where you can add configuration options common to all sub-projects/modules.
 plugins {
-    id("com.android.application") version "8.2.0" apply false
-    id("org.jetbrains.kotlin.android") version "1.9.0" apply false
-    id("com.android.library") version "8.2.0" apply false
+    alias(libs.plugins.android.application) apply false
+    alias(libs.plugins.android.library) apply false
+    alias(libs.plugins.jetbrains.kotlin.android) apply false
 }

examples/llama.android/gradle.properties

@@ -21,3 +21,4 @@ kotlin.code.style=official
 # resources declared in the library itself and none from the library's dependencies,
 # thereby reducing the size of the R class for that library
 android.nonTransitiveRClass=true
+android.native.buildOutput=verbose

examples/llama.android/gradle/libs.versions.toml

@@ -0,0 +1,53 @@
+[versions]
+
+# Plugins
+agp = "8.13.0"
+kotlin = "2.2.20"
+
+# AndroidX
+activity = "1.11.0"
+appcompat = "1.7.1"
+core-ktx = "1.17.0"
+constraint-layout = "2.2.1"
+datastore-preferences = "1.1.7"
+
+# Material
+material = "1.13.0"
+
+# Testing
+espresso-core = "3.7.0"
+androidx-junit = "1.3.0"
+junit = "4.13.2"
+
+
+[plugins]
+android-application = { id = "com.android.application", version.ref = "agp" }
+android-library = { id = "com.android.library", version.ref = "agp" }
+jetbrains-kotlin-android = { id = "org.jetbrains.kotlin.android", version.ref = "kotlin" }
+
+
+[libraries]
+
+# AndroidX
+androidx-activity = { group = "androidx.activity", name = "activity", version.ref = "activity" }
+androidx-appcompat = { group = "androidx.appcompat", name = "appcompat", version.ref = "appcompat" }
+androidx-constraintlayout = { group = "androidx.constraintlayout", name = "constraintlayout", version.ref = "constraint-layout" }
+androidx-core-ktx = { group = "androidx.core", name = "core-ktx", version.ref = "core-ktx" }
+androidx-datastore-preferences = { group = "androidx.datastore", name = "datastore-preferences", version.ref = "datastore-preferences" }
+
+#Material
+material = { group = "com.google.android.material", name = "material", version.ref = "material" }
+
+# Testing
+androidx-espresso-core = { group = "androidx.test.espresso", name = "espresso-core", version.ref = "espresso-core" }
+androidx-junit = { group = "androidx.test.ext", name = "junit", version.ref = "androidx-junit" }
+junit = { group = "junit", name = "junit", version.ref = "junit" }
+
+[bundles]
+androidx = [
+    "androidx-activity",
+    "androidx-appcompat",
+    "androidx-constraintlayout",
+    "androidx-core-ktx",
+    "androidx-datastore-preferences",
+]

examples/llama.android/gradle/wrapper/gradle-wrapper.properties

@@ -1,6 +1,6 @@
-#Thu Dec 21 14:31:09 AEDT 2023
+#Tue Apr 01 11:15:06 PDT 2025
 distributionBase=GRADLE_USER_HOME
 distributionPath=wrapper/dists
-distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.14.3-bin.zip
 zipStoreBase=GRADLE_USER_HOME
 zipStorePath=wrapper/dists

examples/llama.android/lib/build.gradle.kts

@@ -0,0 +1,78 @@
+plugins {
+    alias(libs.plugins.android.library)
+    alias(libs.plugins.jetbrains.kotlin.android)
+}
+
+android {
+    namespace = "com.arm.aichat"
+    compileSdk = 36
+
+    ndkVersion = "29.0.13113456"
+
+    defaultConfig {
+        minSdk = 33
+
+        testInstrumentationRunner = "androidx.test.runner.AndroidJUnitRunner"
+        consumerProguardFiles("consumer-rules.pro")
+
+        ndk {
+             abiFilters += listOf("arm64-v8a", "x86_64")
+        }
+        externalNativeBuild {
+            cmake {
+                arguments += "-DCMAKE_BUILD_TYPE=Release"
+                arguments += "-DCMAKE_MESSAGE_LOG_LEVEL=DEBUG"
+                arguments += "-DCMAKE_VERBOSE_MAKEFILE=ON"
+
+                arguments += "-DBUILD_SHARED_LIBS=ON"
+                arguments += "-DLLAMA_BUILD_COMMON=ON"
+                arguments += "-DLLAMA_CURL=OFF"
+
+                arguments += "-DGGML_NATIVE=OFF"
+                arguments += "-DGGML_BACKEND_DL=ON"
+                arguments += "-DGGML_CPU_ALL_VARIANTS=ON"
+                arguments += "-DGGML_LLAMAFILE=OFF"
+            }
+        }
+        aarMetadata {
+            minCompileSdk = 35
+        }
+    }
+    externalNativeBuild {
+        cmake {
+            path("src/main/cpp/CMakeLists.txt")
+            version = "3.31.6"
+        }
+    }
+    compileOptions {
+        sourceCompatibility = JavaVersion.VERSION_17
+        targetCompatibility = JavaVersion.VERSION_17
+    }
+    kotlin {
+        jvmToolchain(17)
+
+        compileOptions {
+            targetCompatibility = JavaVersion.VERSION_17
+        }
+    }
+
+    packaging {
+        resources {
+            excludes += "/META-INF/{AL2.0,LGPL2.1}"
+        }
+    }
+
+    publishing {
+        singleVariant("release") {
+            withJavadocJar()
+        }
+    }
+}
+
+dependencies {
+    implementation(libs.androidx.core.ktx)
+    implementation(libs.androidx.datastore.preferences)
+
+    testImplementation(libs.junit)
+    androidTestImplementation(libs.androidx.junit)
+}

examples/llama.android/lib/consumer-rules.pro

@@ -0,0 +1,8 @@
+-keep class com.arm.aichat.* { *; }
+-keep class com.arm.aichat.gguf.* { *; }
+
+-keepclasseswithmembernames class * {
+    native <methods>;
+}
+
+-keep class kotlin.Metadata { *; }

examples/llama.android/lib/src/main/cpp/CMakeLists.txt

@@ -0,0 +1,56 @@
+cmake_minimum_required(VERSION 3.31.6)
+
+project("ai-chat" VERSION 1.0.0 LANGUAGES C CXX)
+
+set(CMAKE_C_STANDARD 11)
+set(CMAKE_C_STANDARD_REQUIRED true)
+
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_STANDARD_REQUIRED true)
+
+set(CMAKE_C_FLAGS   "${CMAKE_C_FLAGS}"   CACHE STRING "" FORCE)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}" CACHE STRING "" FORCE)
+
+# --------------------------------------------------------------------------
+# AI Chat library
+# --------------------------------------------------------------------------
+
+if(DEFINED ANDROID_ABI)
+    message(STATUS "Detected Android ABI: ${ANDROID_ABI}")
+    if(ANDROID_ABI STREQUAL "arm64-v8a")
+        set(GGML_SYSTEM_ARCH "ARM")
+        set(GGML_CPU_KLEIDIAI ON)
+        set(GGML_OPENMP ON)
+    elseif(ANDROID_ABI STREQUAL "x86_64")
+        set(GGML_SYSTEM_ARCH "x86")
+        set(GGML_CPU_KLEIDIAI OFF)
+        set(GGML_OPENMP OFF)
+    else()
+        message(FATAL_ERROR "Unsupported ABI: ${ANDROID_ABI}")
+    endif()
+endif()
+
+set(LLAMA_SRC ${CMAKE_CURRENT_LIST_DIR}/../../../../../../)
+add_subdirectory(${LLAMA_SRC} build-llama)
+
+add_library(${CMAKE_PROJECT_NAME} SHARED
+        ai_chat.cpp)
+
+target_compile_definitions(${CMAKE_PROJECT_NAME} PRIVATE
+        GGML_SYSTEM_ARCH=${GGML_SYSTEM_ARCH}
+        GGML_CPU_KLEIDIAI=$<BOOL:${GGML_CPU_KLEIDIAI}>
+        GGML_OPENMP=$<BOOL:${GGML_OPENMP}>
+)
+
+target_include_directories(${CMAKE_PROJECT_NAME} PRIVATE
+        ${LLAMA_SRC}
+        ${LLAMA_SRC}/common
+        ${LLAMA_SRC}/include
+        ${LLAMA_SRC}/ggml/include
+        ${LLAMA_SRC}/ggml/src)
+
+target_link_libraries(${CMAKE_PROJECT_NAME}
+        llama
+        common
+        android
+        log)

examples/llama.android/lib/src/main/cpp/ai_chat.cpp

@@ -0,0 +1,565 @@
+#include <android/log.h>
+#include <jni.h>
+#include <iomanip>
+#include <cmath>
+#include <string>
+#include <unistd.h>
+#include <sampling.h>
+
+#include "logging.h"
+#include "chat.h"
+#include "common.h"
+#include "llama.h"
+
+template<class T>
+static std::string join(const std::vector<T> &values, const std::string &delim) {
+    std::ostringstream str;
+    for (size_t i = 0; i < values.size(); i++) {
+        str << values[i];
+        if (i < values.size() - 1) { str << delim; }
+    }
+    return str.str();
+}
+
+/**
+ * LLama resources: context, model, batch and sampler
+ */
+constexpr int   N_THREADS_MIN           = 2;
+constexpr int   N_THREADS_MAX           = 4;
+constexpr int   N_THREADS_HEADROOM      = 2;
+
+constexpr int   DEFAULT_CONTEXT_SIZE    = 8192;
+constexpr int   OVERFLOW_HEADROOM       = 4;
+constexpr int   BATCH_SIZE              = 512;
+constexpr float DEFAULT_SAMPLER_TEMP    = 0.3f;
+
+static llama_model                      * g_model;
+static llama_context                    * g_context;
+static llama_batch                        g_batch;
+static common_chat_templates_ptr          g_chat_templates;
+static common_sampler                   * g_sampler;
+
+extern "C"
+JNIEXPORT void JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_init(JNIEnv *env, jobject /*unused*/, jstring nativeLibDir) {
+    // Set llama log handler to Android
+    llama_log_set(aichat_android_log_callback, nullptr);
+
+    // Loading all CPU backend variants
+    const auto *path_to_backend = env->GetStringUTFChars(nativeLibDir, 0);
+    LOGi("Loading backends from %s", path_to_backend);
+    ggml_backend_load_all_from_path(path_to_backend);
+    env->ReleaseStringUTFChars(nativeLibDir, path_to_backend);
+
+    // Initialize backends
+    llama_backend_init();
+    LOGi("Backend initiated; Log handler set.");
+}
+
+extern "C"
+JNIEXPORT jint JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_load(JNIEnv *env, jobject, jstring jmodel_path) {
+    llama_model_params model_params = llama_model_default_params();
+
+    const auto *model_path = env->GetStringUTFChars(jmodel_path, 0);
+    LOGd("%s: Loading model from: \n%s\n", __func__, model_path);
+
+    auto *model = llama_model_load_from_file(model_path, model_params);
+    env->ReleaseStringUTFChars(jmodel_path, model_path);
+    if (!model) {
+        return 1;
+    }
+    g_model = model;
+    return 0;
+}
+
+static llama_context *init_context(llama_model *model, const int n_ctx = DEFAULT_CONTEXT_SIZE) {
+    if (!model) {
+        LOGe("%s: model cannot be null", __func__);
+        return nullptr;
+    }
+
+    // Multi-threading setup
+    const int n_threads = std::max(N_THREADS_MIN, std::min(N_THREADS_MAX,
+                                                     (int) sysconf(_SC_NPROCESSORS_ONLN) -
+                                                     N_THREADS_HEADROOM));
+    LOGi("%s: Using %d threads", __func__, n_threads);
+
+    // Context parameters setup
+    llama_context_params ctx_params = llama_context_default_params();
+    const int trained_context_size = llama_model_n_ctx_train(model);
+    if (n_ctx > trained_context_size) {
+        LOGw("%s: Model was trained with only %d context size! Enforcing %d context size...",
+             __func__, trained_context_size, n_ctx);
+    }
+    ctx_params.n_ctx = n_ctx;
+    ctx_params.n_batch = BATCH_SIZE;
+    ctx_params.n_ubatch = BATCH_SIZE;
+    ctx_params.n_threads = n_threads;
+    ctx_params.n_threads_batch = n_threads;
+    auto *context = llama_init_from_model(g_model, ctx_params);
+    if (context == nullptr) {
+        LOGe("%s: llama_new_context_with_model() returned null)", __func__);
+    }
+    return context;
+}
+
+static common_sampler *new_sampler(float temp) {
+    common_params_sampling sparams;
+    sparams.temp = temp;
+    return common_sampler_init(g_model, sparams);
+}
+
+extern "C"
+JNIEXPORT jint JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_prepare(JNIEnv * /*env*/, jobject /*unused*/) {
+    auto *context = init_context(g_model);
+    if (!context) { return 1; }
+    g_context = context;
+    g_batch = llama_batch_init(BATCH_SIZE, 0, 1);
+    g_chat_templates = common_chat_templates_init(g_model, "");
+    g_sampler = new_sampler(DEFAULT_SAMPLER_TEMP);
+    return 0;
+}
+
+static std::string get_backend() {
+    std::vector<std::string> backends;
+    for (size_t i = 0; i < ggml_backend_reg_count(); i++) {
+        auto *reg = ggml_backend_reg_get(i);
+        std::string name = ggml_backend_reg_name(reg);
+        if (name != "CPU") {
+            backends.push_back(ggml_backend_reg_name(reg));
+        }
+    }
+    return backends.empty() ? "CPU" : join(backends, ",");
+}
+
+extern "C"
+JNIEXPORT jstring JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_systemInfo(JNIEnv *env, jobject /*unused*/) {
+    return env->NewStringUTF(llama_print_system_info());
+}
+
+extern "C"
+JNIEXPORT jstring JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_benchModel(JNIEnv *env, jobject /*unused*/, jint pp, jint tg,
+                                                      jint pl, jint nr) {
+    auto *context = init_context(g_model, pp);
+    if (!context) {
+        const auto *const err_msg = "Fail to init_context! Bench aborted.";
+        LOGe(err_msg);
+        return env->NewStringUTF(err_msg);
+    }
+
+    auto pp_avg = 0.0;
+    auto tg_avg = 0.0;
+    auto pp_std = 0.0;
+    auto tg_std = 0.0;
+
+    const uint32_t n_ctx = llama_n_ctx(context);
+    LOGi("n_ctx = %d", n_ctx);
+
+    int i, j;
+    int nri;
+    for (nri = 0; nri < nr; nri++) {
+        LOGi("Benchmark prompt processing (pp = %d)", pp);
+
+        common_batch_clear(g_batch);
+
+        const int n_tokens = pp;
+        for (i = 0; i < n_tokens; i++) {
+            common_batch_add(g_batch, 0, i, {0}, false);
+        }
+
+        g_batch.logits[g_batch.n_tokens - 1] = true;
+        llama_memory_clear(llama_get_memory(context), false);
+
+        const auto t_pp_start = ggml_time_us();
+        if (llama_decode(context, g_batch) != 0) {
+            LOGe("llama_decode() failed during prompt processing");
+        }
+        const auto t_pp_end = ggml_time_us();
+
+        // bench text generation
+
+        LOGi("Benchmark text generation (tg = %d)", tg);
+
+        llama_memory_clear(llama_get_memory(context), false);
+        const auto t_tg_start = ggml_time_us();
+        for (i = 0; i < tg; i++) {
+            common_batch_clear(g_batch);
+            for (j = 0; j < pl; j++) {
+                common_batch_add(g_batch, 0, i, {j}, true);
+            }
+
+            if (llama_decode(context, g_batch) != 0) {
+                LOGe("llama_decode() failed during text generation");
+            }
+        }
+        const auto t_tg_end = ggml_time_us();
+
+        llama_memory_clear(llama_get_memory(context), false);
+
+        const auto t_pp = double(t_pp_end - t_pp_start) / 1000000.0;
+        const auto t_tg = double(t_tg_end - t_tg_start) / 1000000.0;
+
+        const auto speed_pp = double(pp) / t_pp;
+        const auto speed_tg = double(pl * tg) / t_tg;
+
+        pp_avg += speed_pp;
+        tg_avg += speed_tg;
+
+        pp_std += speed_pp * speed_pp;
+        tg_std += speed_tg * speed_tg;
+
+        LOGi("pp %f t/s, tg %f t/s", speed_pp, speed_tg);
+    }
+
+    llama_free(context);
+
+    pp_avg /= double(nr);
+    tg_avg /= double(nr);
+
+    if (nr > 1) {
+        pp_std = sqrt(pp_std / double(nr - 1) - pp_avg * pp_avg * double(nr) / double(nr - 1));
+        tg_std = sqrt(tg_std / double(nr - 1) - tg_avg * tg_avg * double(nr) / double(nr - 1));
+    } else {
+        pp_std = 0;
+        tg_std = 0;
+    }
+
+    char model_desc[128];
+    llama_model_desc(g_model, model_desc, sizeof(model_desc));
+
+    const auto model_size = double(llama_model_size(g_model)) / 1024.0 / 1024.0 / 1024.0;
+    const auto model_n_params = double(llama_model_n_params(g_model)) / 1e9;
+
+    const auto backend = get_backend();
+    std::stringstream result;
+    result << std::setprecision(3);
+    result << "| model | size | params | backend | test | t/s |\n";
+    result << "| --- | --- | --- | --- | --- | --- |\n";
+    result << "| " << model_desc << " | " << model_size << "GiB | " << model_n_params << "B | "
+           << backend << " | pp " << pp << " | " << pp_avg << " ± " << pp_std << " |\n";
+    result << "| " << model_desc << " | " << model_size << "GiB | " << model_n_params << "B | "
+           << backend << " | tg " << tg << " | " << tg_avg << " ± " << tg_std << " |\n";
+    return env->NewStringUTF(result.str().c_str());
+}
+
+
+/**
+ * Completion loop's long-term states:
+ * - chat management
+ * - position tracking
+ */
+constexpr const char *ROLE_SYSTEM       = "system";
+constexpr const char *ROLE_USER         = "user";
+constexpr const char *ROLE_ASSISTANT    = "assistant";
+
+static std::vector<common_chat_msg> chat_msgs;
+static llama_pos system_prompt_position;
+static llama_pos current_position;
+
+static void reset_long_term_states(const bool clear_kv_cache = true) {
+    chat_msgs.clear();
+    system_prompt_position = 0;
+    current_position = 0;
+
+    if (clear_kv_cache)
+        llama_memory_clear(llama_get_memory(g_context), false);
+}
+
+/**
+ * TODO-hyin: implement sliding-window version as a better alternative
+ *
+ * Context shifting by discarding the older half of the tokens appended after system prompt:
+ * - take the [system_prompt_position] first tokens from the original prompt
+ * - take half of the last (system_prompt_position - system_prompt_position) tokens
+ * - recompute the logits in batches
+ */
+static void shift_context() {
+    const int n_discard = (current_position - system_prompt_position) / 2;
+    LOGi("%s: Discarding %d tokens", __func__, n_discard);
+    llama_memory_seq_rm(llama_get_memory(g_context), 0, system_prompt_position, system_prompt_position + n_discard);
+    llama_memory_seq_add(llama_get_memory(g_context), 0, system_prompt_position + n_discard, current_position, -n_discard);
+    current_position -= n_discard;
+    LOGi("%s: Context shifting done! Current position: %d", __func__, current_position);
+}
+
+static std::string chat_add_and_format(const std::string &role, const std::string &content) {
+    common_chat_msg new_msg;
+    new_msg.role = role;
+    new_msg.content = content;
+    auto formatted = common_chat_format_single(
+            g_chat_templates.get(), chat_msgs, new_msg, role == ROLE_USER, /* use_jinja */ false);
+    chat_msgs.push_back(new_msg);
+    LOGi("%s: Formatted and added %s message: \n%s\n", __func__, role.c_str(), formatted.c_str());
+    return formatted;
+}
+
+/**
+ * Completion loop's short-term states:
+ * - stop generation position
+ * - token chars caching
+ * - current assistant message being generated
+ */
+static llama_pos stop_generation_position;
+static std::string cached_token_chars;
+static std::ostringstream assistant_ss;
+
+static void reset_short_term_states() {
+    stop_generation_position = 0;
+    cached_token_chars.clear();
+    assistant_ss.str("");
+}
+
+static int decode_tokens_in_batches(
+        llama_context *context,
+        llama_batch &batch,
+        const llama_tokens &tokens,
+        const llama_pos start_pos,
+        const bool compute_last_logit = false) {
+    // Process tokens in batches using the global batch
+    LOGd("%s: Decode %d tokens starting at position %d", __func__, (int) tokens.size(), start_pos);
+    for (int i = 0; i < (int) tokens.size(); i += BATCH_SIZE) {
+        const int cur_batch_size = std::min((int) tokens.size() - i, BATCH_SIZE);
+        common_batch_clear(batch);
+        LOGv("%s: Preparing a batch size of %d starting at: %d", __func__, cur_batch_size, i);
+
+        // Shift context if current batch cannot fit into the context
+        if (start_pos + i + cur_batch_size >= DEFAULT_CONTEXT_SIZE - OVERFLOW_HEADROOM) {
+            LOGw("%s: Current batch won't fit into context! Shifting...", __func__);
+            shift_context();
+        }
+
+        // Add tokens to the batch with proper positions
+        for (int j = 0; j < cur_batch_size; j++) {
+            const llama_token token_id = tokens[i + j];
+            const llama_pos position = start_pos + i + j;
+            const bool want_logit = compute_last_logit && (i + j == tokens.size() - 1);
+            common_batch_add(batch, token_id, position, {0}, want_logit);
+        }
+
+        // Decode this batch
+        const int decode_result = llama_decode(context, batch);
+        if (decode_result) {
+            LOGe("%s: llama_decode failed w/ %d", __func__, decode_result);
+            return 1;
+        }
+    }
+    return 0;
+}
+
+extern "C"
+JNIEXPORT jint JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_processSystemPrompt(
+        JNIEnv *env,
+        jobject /*unused*/,
+        jstring jsystem_prompt
+) {
+    // Reset long-term & short-term states
+    reset_long_term_states();
+    reset_short_term_states();
+
+    // Obtain system prompt from JEnv
+    const auto *system_prompt = env->GetStringUTFChars(jsystem_prompt, nullptr);
+    LOGd("%s: System prompt received: \n%s", __func__, system_prompt);
+    std::string formatted_system_prompt(system_prompt);
+    env->ReleaseStringUTFChars(jsystem_prompt, system_prompt);
+
+    // Format system prompt if applicable
+    const bool has_chat_template = common_chat_templates_was_explicit(g_chat_templates.get());
+    if (has_chat_template) {
+        formatted_system_prompt = chat_add_and_format(ROLE_SYSTEM, system_prompt);
+    }
+
+    // Tokenize system prompt
+    const auto system_tokens = common_tokenize(g_context, formatted_system_prompt,
+                                               has_chat_template, has_chat_template);
+    for (auto id: system_tokens) {
+        LOGv("token: `%s`\t -> `%d`", common_token_to_piece(g_context, id).c_str(), id);
+    }
+
+    // Handle context overflow
+    const int max_batch_size = DEFAULT_CONTEXT_SIZE - OVERFLOW_HEADROOM;
+    if ((int) system_tokens.size() > max_batch_size) {
+        LOGe("%s: System prompt too long for context! %d tokens, max: %d",
+             __func__, (int) system_tokens.size(), max_batch_size);
+        return 1;
+    }
+
+    // Decode system tokens in batches
+    if (decode_tokens_in_batches(g_context, g_batch, system_tokens, current_position)) {
+        LOGe("%s: llama_decode() failed!", __func__);
+        return 2;
+    }
+
+    // Update position
+    system_prompt_position = current_position = (int) system_tokens.size();
+    return 0;
+}
+
+extern "C"
+JNIEXPORT jint JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_processUserPrompt(
+        JNIEnv *env,
+        jobject /*unused*/,
+        jstring juser_prompt,
+        jint n_predict
+) {
+    // Reset short-term states
+    reset_short_term_states();
+
+    // Obtain and tokenize user prompt
+    const auto *const user_prompt = env->GetStringUTFChars(juser_prompt, nullptr);
+    LOGd("%s: User prompt received: \n%s", __func__, user_prompt);
+    std::string formatted_user_prompt(user_prompt);
+    env->ReleaseStringUTFChars(juser_prompt, user_prompt);
+
+    // Format user prompt if applicable
+    const bool has_chat_template = common_chat_templates_was_explicit(g_chat_templates.get());
+    if (has_chat_template) {
+        formatted_user_prompt = chat_add_and_format(ROLE_USER, user_prompt);
+    }
+
+    // Decode formatted user prompts
+    auto user_tokens = common_tokenize(g_context, formatted_user_prompt, has_chat_template, has_chat_template);
+    for (auto id: user_tokens) {
+        LOGv("token: `%s`\t -> `%d`", common_token_to_piece(g_context, id).c_str(), id);
+    }
+
+    // Ensure user prompt doesn't exceed the context size by truncating if necessary.
+    const int user_prompt_size = (int) user_tokens.size();
+    const int max_batch_size = DEFAULT_CONTEXT_SIZE - OVERFLOW_HEADROOM;
+    if (user_prompt_size > max_batch_size) {
+        const int skipped_tokens = user_prompt_size - max_batch_size;
+        user_tokens.resize(max_batch_size);
+        LOGw("%s: User prompt too long! Skipped %d tokens!", __func__, skipped_tokens);
+    }
+
+    // Decode user tokens in batches
+    if (decode_tokens_in_batches(g_context, g_batch, user_tokens, current_position, true)) {
+        LOGe("%s: llama_decode() failed!", __func__);
+        return 2;
+    }
+
+    // Update position
+    current_position += user_prompt_size;
+    stop_generation_position = current_position + user_prompt_size + n_predict;
+    return 0;
+}
+
+static bool is_valid_utf8(const char *string) {
+    if (!string) { return true; }
+
+    const auto *bytes = (const unsigned char *) string;
+    int num;
+
+    while (*bytes != 0x00) {
+        if ((*bytes & 0x80) == 0x00) {
+            // U+0000 to U+007F
+            num = 1;
+        } else if ((*bytes & 0xE0) == 0xC0) {
+            // U+0080 to U+07FF
+            num = 2;
+        } else if ((*bytes & 0xF0) == 0xE0) {
+            // U+0800 to U+FFFF
+            num = 3;
+        } else if ((*bytes & 0xF8) == 0xF0) {
+            // U+10000 to U+10FFFF
+            num = 4;
+        } else {
+            return false;
+        }
+
+        bytes += 1;
+        for (int i = 1; i < num; ++i) {
+            if ((*bytes & 0xC0) != 0x80) {
+                return false;
+            }
+            bytes += 1;
+        }
+    }
+    return true;
+}
+
+extern "C"
+JNIEXPORT jstring JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_generateNextToken(
+        JNIEnv *env,
+        jobject /*unused*/
+) {
+    // Infinite text generation via context shifting
+    if (current_position >= DEFAULT_CONTEXT_SIZE - OVERFLOW_HEADROOM) {
+        LOGw("%s: Context full! Shifting...", __func__);
+        shift_context();
+    }
+
+    // Stop if reaching the marked position
+    if (current_position >= stop_generation_position) {
+        LOGw("%s: STOP: hitting stop position: %d", __func__, stop_generation_position);
+        return nullptr;
+    }
+
+    // Sample next token
+    const auto new_token_id = common_sampler_sample(g_sampler, g_context, -1);
+    common_sampler_accept(g_sampler, new_token_id, true);
+
+    // Populate the batch with new token, then decode
+    common_batch_clear(g_batch);
+    common_batch_add(g_batch, new_token_id, current_position, {0}, true);
+    if (llama_decode(g_context, g_batch) != 0) {
+        LOGe("%s: llama_decode() failed for generated token", __func__);
+        return nullptr;
+    }
+
+    // Update position
+    current_position++;
+
+    // Stop if next token is EOG
+    if (llama_vocab_is_eog(llama_model_get_vocab(g_model), new_token_id)) {
+        LOGd("id: %d,\tIS EOG!\nSTOP.", new_token_id);
+        chat_add_and_format(ROLE_ASSISTANT, assistant_ss.str());
+        return nullptr;
+    }
+
+    // If not EOG, convert to text
+    auto new_token_chars = common_token_to_piece(g_context, new_token_id);
+    cached_token_chars += new_token_chars;
+
+    // Create and return a valid UTF-8 Java string
+    jstring result = nullptr;
+    if (is_valid_utf8(cached_token_chars.c_str())) {
+        result = env->NewStringUTF(cached_token_chars.c_str());
+        LOGv("id: %d,\tcached: `%s`,\tnew: `%s`", new_token_id, cached_token_chars.c_str(), new_token_chars.c_str());
+
+        assistant_ss << cached_token_chars;
+        cached_token_chars.clear();
+    } else {
+        LOGv("id: %d,\tappend to cache", new_token_id);
+        result = env->NewStringUTF("");
+    }
+    return result;
+}
+
+
+extern "C"
+JNIEXPORT void JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_unload(JNIEnv * /*unused*/, jobject /*unused*/) {
+    // Reset long-term & short-term states
+    reset_long_term_states();
+    reset_short_term_states();
+
+    // Free up resources
+    common_sampler_free(g_sampler);
+    g_chat_templates.reset();
+    llama_batch_free(g_batch);
+    llama_free(g_context);
+    llama_model_free(g_model);
+}
+
+extern "C"
+JNIEXPORT void JNICALL
+Java_com_arm_aichat_internal_InferenceEngineImpl_shutdown(JNIEnv *env, jobject /*unused*/) {
+    llama_backend_free();
+}

examples/llama.android/lib/src/main/cpp/logging.h

@@ -0,0 +1,61 @@
+//
+// Created by Han Yin on 10/31/25.
+//
+
+#ifndef AICHAT_LOGGING_H
+#define AICHAT_LOGGING_H
+
+#endif //AICHAT_LOGGING_H
+
+#pragma once
+#include <android/log.h>
+
+#ifndef LOG_TAG
+#define LOG_TAG "ai-chat"
+#endif
+
+#ifndef LOG_MIN_LEVEL
+#if defined(NDEBUG)
+#define LOG_MIN_LEVEL ANDROID_LOG_INFO
+#else
+#define LOG_MIN_LEVEL ANDROID_LOG_VERBOSE
+#endif
+#endif
+
+static inline int ai_should_log(int prio) {
+    return __android_log_is_loggable(prio, LOG_TAG, LOG_MIN_LEVEL);
+}
+
+#if LOG_MIN_LEVEL <= ANDROID_LOG_VERBOSE
+#define LOGv(...) do { if (ai_should_log(ANDROID_LOG_VERBOSE)) __android_log_print(ANDROID_LOG_VERBOSE, LOG_TAG, __VA_ARGS__); } while (0)
+#else
+#define LOGv(...) ((void)0)
+#endif
+
+#if LOG_MIN_LEVEL <= ANDROID_LOG_DEBUG
+#define LOGd(...) do { if (ai_should_log(ANDROID_LOG_DEBUG)) __android_log_print(ANDROID_LOG_DEBUG, LOG_TAG, __VA_ARGS__); } while (0)
+#else
+#define LOGd(...) ((void)0)
+#endif
+
+#define LOGi(...)   do { if (ai_should_log(ANDROID_LOG_INFO )) __android_log_print(ANDROID_LOG_INFO , LOG_TAG, __VA_ARGS__); } while (0)
+#define LOGw(...)   do { if (ai_should_log(ANDROID_LOG_WARN )) __android_log_print(ANDROID_LOG_WARN , LOG_TAG, __VA_ARGS__); } while (0)
+#define LOGe(...)   do { if (ai_should_log(ANDROID_LOG_ERROR)) __android_log_print(ANDROID_LOG_ERROR, LOG_TAG, __VA_ARGS__); } while (0)
+
+static inline int android_log_prio_from_ggml(enum ggml_log_level level) {
+    switch (level) {
+        case GGML_LOG_LEVEL_ERROR: return ANDROID_LOG_ERROR;
+        case GGML_LOG_LEVEL_WARN:  return ANDROID_LOG_WARN;
+        case GGML_LOG_LEVEL_INFO:  return ANDROID_LOG_INFO;
+        case GGML_LOG_LEVEL_DEBUG: return ANDROID_LOG_DEBUG;
+        default:                   return ANDROID_LOG_DEFAULT;
+    }
+}
+
+static inline void aichat_android_log_callback(enum ggml_log_level level,
+                                              const char* text,
+                                              void* /*user*/) {
+    const int prio = android_log_prio_from_ggml(level);
+    if (!ai_should_log(prio)) return;
+    __android_log_write(prio, LOG_TAG, text);
+}

examples/llama.android/lib/src/main/java/com/arm/aichat/AiChat.kt

@@ -0,0 +1,14 @@
+package com.arm.aichat
+
+import android.content.Context
+import com.arm.aichat.internal.InferenceEngineImpl
+
+/**
+ * Main entry point for Arm's AI Chat library.
+ */
+object AiChat {
+    /**
+     * Get the inference engine single instance.
+     */
+    fun getInferenceEngine(context: Context) = InferenceEngineImpl.getInstance(context)
+}

examples/llama.android/lib/src/main/java/com/arm/aichat/InferenceEngine.kt

@@ -0,0 +1,89 @@
+package com.arm.aichat
+
+import com.arm.aichat.InferenceEngine.State
+import kotlinx.coroutines.flow.Flow
+import kotlinx.coroutines.flow.StateFlow
+
+/**
+ * Interface defining the core LLM inference operations.
+ */
+interface InferenceEngine {
+    /**
+     * Current state of the inference engine
+     */
+    val state: StateFlow<State>
+
+    /**
+     * Load a model from the given path.
+     *
+     * @throws UnsupportedArchitectureException if model architecture not supported
+     */
+    suspend fun loadModel(pathToModel: String)
+
+    /**
+     * Sends a system prompt to the loaded model
+     */
+    suspend fun setSystemPrompt(systemPrompt: String)
+
+    /**
+     * Sends a user prompt to the loaded model and returns a Flow of generated tokens.
+     */
+    fun sendUserPrompt(message: String, predictLength: Int = DEFAULT_PREDICT_LENGTH): Flow<String>
+
+    /**
+     * Runs a benchmark with the specified parameters.
+     */
+    suspend fun bench(pp: Int, tg: Int, pl: Int, nr: Int = 1): String
+
+    /**
+     * Unloads the currently loaded model.
+     */
+    suspend fun cleanUp()
+
+    /**
+     * Cleans up resources when the engine is no longer needed.
+     */
+    fun destroy()
+
+    /**
+     * States of the inference engine
+     */
+    sealed class State {
+        object Uninitialized : State()
+        object Initializing : State()
+        object Initialized : State()
+
+        object LoadingModel : State()
+        object UnloadingModel : State()
+        object ModelReady : State()
+
+        object Benchmarking : State()
+        object ProcessingSystemPrompt : State()
+        object ProcessingUserPrompt : State()
+
+        object Generating : State()
+
+        data class Error(val exception: Exception) : State()
+    }
+
+    companion object {
+        const val DEFAULT_PREDICT_LENGTH = 1024
+    }
+}
+
+val State.isUninterruptible
+    get() = this is State.Initializing ||
+        this is State.LoadingModel ||
+        this is State.UnloadingModel ||
+        this is State.Benchmarking ||
+        this is State.ProcessingSystemPrompt ||
+        this is State.ProcessingUserPrompt
+
+val State.isModelLoaded: Boolean
+    get() = this is State.ModelReady ||
+        this is State.Benchmarking ||
+        this is State.ProcessingSystemPrompt ||
+        this is State.ProcessingUserPrompt ||
+        this is State.Generating
+
+class UnsupportedArchitectureException : Exception()

examples/llama.android/lib/src/main/java/com/arm/aichat/gguf/FileType.kt

@@ -0,0 +1,61 @@
+package com.arm.aichat.gguf
+
+import kotlin.collections.get
+
+
+/**
+ * Numerical codes used by `general.file_type` (see llama.cpp repo's `constants.py`).
+ * The `label` matches what llama‑cli prints.
+ */
+enum class FileType(val code: Int, val label: String) {
+    ALL_F32(0, "all F32"),
+    MOSTLY_F16(1, "F16"),
+    MOSTLY_Q4_0(2, "Q4_0"),
+    MOSTLY_Q4_1(3, "Q4_1"),
+    // 4 removed
+    MOSTLY_Q8_0(7, "Q8_0"),
+    MOSTLY_Q5_0(8, "Q5_0"),
+    MOSTLY_Q5_1(9, "Q5_1"),
+
+    /* K‑quants ------------------------------------------------------------ */
+    MOSTLY_Q2_K      (10, "Q2_K - Medium"),
+    MOSTLY_Q3_K_S    (11, "Q3_K - Small"),
+    MOSTLY_Q3_K_M    (12, "Q3_K - Medium"),
+    MOSTLY_Q3_K_L    (13, "Q3_K - Large"),
+    MOSTLY_Q4_K_S    (14, "Q4_K - Small"),
+    MOSTLY_Q4_K_M    (15, "Q4_K - Medium"),
+    MOSTLY_Q5_K_S    (16, "Q5_K - Small"),
+    MOSTLY_Q5_K_M    (17, "Q5_K - Medium"),
+    MOSTLY_Q6_K      (18, "Q6_K"),
+
+    /* IQ quants ----------------------------------------------------------- */
+    MOSTLY_IQ2_XXS   (19, "IQ2_XXS - 2.06 bpw"),
+    MOSTLY_IQ2_XS    (20, "IQ2_XS - 2.31 bpw"),
+    MOSTLY_Q2_K_S    (21, "Q2_K - Small"),
+    MOSTLY_IQ3_XS    (22, "IQ3_XS - 3.30 bpw"),
+    MOSTLY_IQ3_XXS   (23, "IQ3_XXS - 3.06 bpw"),
+    MOSTLY_IQ1_S     (24, "IQ1_S - 1.56 bpw"),
+    MOSTLY_IQ4_NL    (25, "IQ4_NL - 4.5 bpw"),
+    MOSTLY_IQ3_S     (26, "IQ3_S - 3.44 bpw"),
+    MOSTLY_IQ3_M     (27, "IQ3_M - 3.66 bpw"),
+    MOSTLY_IQ2_S     (28, "IQ2_S - 2.50 bpw"),
+    MOSTLY_IQ2_M     (29, "IQ2_M - 2.70 bpw"),
+    MOSTLY_IQ4_XS    (30, "IQ4_XS - 4.25 bpw"),
+    MOSTLY_IQ1_M     (31, "IQ1_M - 1.75 bpw"),
+
+    /* BF16 & Ternary ------------------------------------------------------ */
+    MOSTLY_BF16      (32, "BF16"),
+    MOSTLY_TQ1_0     (36, "TQ1_0 - 1.69 bpw ternary"),
+    MOSTLY_TQ2_0     (37, "TQ2_0 - 2.06 bpw ternary"),
+
+    /* Special flag -------------------------------------------------------- */
+    GUESSED(1024, "(guessed)"),
+
+    UNKNOWN(-1, "unknown");
+
+    companion object {
+        private val map = entries.associateBy(FileType::code)
+
+        fun fromCode(code: Int?): FileType = map[code] ?: UNKNOWN
+    }
+}

examples/llama.android/lib/src/main/java/com/arm/aichat/gguf/GgufMetadata.kt

@@ -0,0 +1,132 @@
+package com.arm.aichat.gguf
+
+import java.io.IOException
+
+
+/**
+ * Structured metadata of GGUF
+ */
+data class GgufMetadata(
+    // Basic file info
+    val version: GgufVersion,
+    val tensorCount: Long,
+    val kvCount: Long,
+
+    // General info
+    val basic: BasicInfo,
+    val author: AuthorInfo? = null,
+    val additional: AdditionalInfo? = null,
+    val architecture: ArchitectureInfo? = null,
+    val baseModels: List<BaseModelInfo>? = null,
+    val tokenizer: TokenizerInfo? = null,
+
+    // Derivative info
+    val dimensions: DimensionsInfo? = null,
+    val attention: AttentionInfo? = null,
+    val rope: RopeInfo? = null,
+    val experts: ExpertsInfo? = null
+) {
+    enum class GgufVersion(val code: Int, val label: String) {
+        /** First public draft; little‑endian only, no alignment key. */
+        LEGACY_V1(1, "Legacy v1"),
+
+        /** Added split‑file support and some extra metadata keys. */
+        EXTENDED_V2(2, "Extended v2"),
+
+        /** Current spec: endian‑aware, mandatory alignment, fully validated. */
+        VALIDATED_V3(3, "Validated v3");
+
+        companion object {
+            fun fromCode(code: Int): GgufVersion =
+                entries.firstOrNull { it.code == code }
+                    ?: throw IOException("Unknown GGUF version code $code")
+        }
+
+        override fun toString(): String = "$label (code=$code)"
+    }
+
+    data class BasicInfo(
+        val uuid: String? = null,
+        val name: String? = null,
+        val nameLabel: String? = null,
+        val sizeLabel: String? = null,  // Size label like "7B"
+    )
+
+    data class AuthorInfo(
+        val organization: String? = null,
+        val author: String? = null,
+        val doi: String? = null,
+        val url: String? = null,
+        val repoUrl: String? = null,
+        val license: String? = null,
+        val licenseLink: String? = null,
+    )
+
+    data class AdditionalInfo(
+        val type: String? = null,
+        val description: String? = null,
+        val tags: List<String>? = null,
+        val languages: List<String>? = null,
+    )
+
+    data class ArchitectureInfo(
+        val architecture: String? = null,
+        val fileType: Int? = null,
+        val vocabSize: Int? = null,
+        val finetune: String? = null,
+        val quantizationVersion: Int? = null,
+    )
+
+    data class BaseModelInfo(
+        val name: String? = null,
+        val author: String? = null,
+        val version: String? = null,
+        val organization: String? = null,
+        val url: String? = null,
+        val doi: String? = null,
+        val uuid: String? = null,
+        val repoUrl: String? = null,
+    )
+
+    data class TokenizerInfo(
+        val model: String? = null,
+        val bosTokenId: Int? = null,
+        val eosTokenId: Int? = null,
+        val unknownTokenId: Int? = null,
+        val paddingTokenId: Int? = null,
+        val addBosToken: Boolean? = null,
+        val addEosToken: Boolean? = null,
+        val chatTemplate: String? = null,
+    )
+
+    data class DimensionsInfo(
+        val contextLength: Int? = null,
+        val embeddingSize: Int? = null,
+        val blockCount: Int? = null,
+        val feedForwardSize: Int? = null,
+    )
+
+    data class AttentionInfo(
+        val headCount: Int? = null,
+        val headCountKv: Int? = null,
+        val keyLength: Int? = null,
+        val valueLength: Int? = null,
+        val layerNormEpsilon: Float? = null,
+        val layerNormRmsEpsilon: Float? = null,
+    )
+
+    data class RopeInfo(
+        val frequencyBase: Float? = null,
+        val dimensionCount: Int? = null,
+        val scalingType: String? = null,
+        val scalingFactor: Float? = null,
+        val attnFactor: Float? = null,
+        val originalContextLength: Int? = null,
+        val finetuned: Boolean? = null,
+    )
+
+    data class ExpertsInfo(
+        val count: Int? = null,
+        val usedCount: Int? = null,
+    )
+}

examples/llama.android/lib/src/main/java/com/arm/aichat/gguf/GgufMetadataReader.kt

@@ -0,0 +1,77 @@
+package com.arm.aichat.gguf
+
+import android.content.Context
+import android.net.Uri
+import com.arm.aichat.internal.gguf.GgufMetadataReaderImpl
+import java.io.File
+import java.io.IOException
+import java.io.InputStream
+
+/**
+ * Interface for reading GGUF metadata from model files.
+ * Use `GgufMetadataReader.create()` to get an instance.
+ */
+interface GgufMetadataReader {
+    /**
+     * Reads the magic number from the specified file path.
+     *
+     * @param file Java File to the GGUF file with absolute path
+     * @return true if file is valid GGUF, otherwise false
+     * @throws InvalidFileFormatException if file format is invalid
+     */
+    suspend fun ensureSourceFileFormat(file: File): Boolean
+
+    /**
+     * Reads the magic number from the specified file path.
+     *
+     * @param context Context for obtaining [android.content.ContentProvider]
+     * @param uri Uri to the GGUF file provided by [android.content.ContentProvider]
+     * @return true if file is valid GGUF, otherwise false
+     * @throws InvalidFileFormatException if file format is invalid
+     */
+    suspend fun ensureSourceFileFormat(context: Context, uri: Uri): Boolean
+
+    /**
+     * Reads and parses GGUF metadata from the specified file path.
+     *
+     * @param input the [InputStream] obtained from a readable file or content
+     * @return Structured metadata extracted from the file
+     * @throws IOException if file is damaged or cannot be read
+     * @throws InvalidFileFormatException if file format is invalid
+     */
+    suspend fun readStructuredMetadata(input: InputStream): GgufMetadata
+
+    companion object {
+        private val DEFAULT_SKIP_KEYS = setOf(
+            "tokenizer.chat_template",
+            "tokenizer.ggml.scores",
+            "tokenizer.ggml.tokens",
+            "tokenizer.ggml.token_type"
+        )
+
+        /**
+         * Creates a default GgufMetadataReader instance
+         */
+        fun create(): GgufMetadataReader = GgufMetadataReaderImpl(
+            skipKeys = DEFAULT_SKIP_KEYS,
+            arraySummariseThreshold = 1_000
+        )
+
+        /**
+         * Creates a GgufMetadataReader with custom configuration
+         *
+         * @param skipKeys Keys whose value should be skipped entirely (not kept in the result map)
+         * @param arraySummariseThreshold If ≥0, arrays longer get summarised, not materialised;
+         *                                If -1, never summarise.
+         */
+        fun create(
+            skipKeys: Set<String> = DEFAULT_SKIP_KEYS,
+            arraySummariseThreshold: Int = 1_000
+        ): GgufMetadataReader = GgufMetadataReaderImpl(
+            skipKeys = skipKeys,
+            arraySummariseThreshold = arraySummariseThreshold
+        )
+    }
+}
+
+class InvalidFileFormatException : IOException()

examples/llama.android/lib/src/main/java/com/arm/aichat/internal/InferenceEngineImpl.kt

@@ -0,0 +1,309 @@
+package com.arm.aichat.internal
+
+import android.content.Context
+import android.util.Log
+import com.arm.aichat.InferenceEngine
+import com.arm.aichat.UnsupportedArchitectureException
+import com.arm.aichat.internal.InferenceEngineImpl.Companion.getInstance
+import dalvik.annotation.optimization.FastNative
+import kotlinx.coroutines.CancellationException
+import kotlinx.coroutines.CoroutineScope
+import kotlinx.coroutines.Dispatchers
+import kotlinx.coroutines.ExperimentalCoroutinesApi
+import kotlinx.coroutines.SupervisorJob
+import kotlinx.coroutines.cancel
+import kotlinx.coroutines.flow.Flow
+import kotlinx.coroutines.flow.MutableStateFlow
+import kotlinx.coroutines.flow.StateFlow
+import kotlinx.coroutines.flow.flow
+import kotlinx.coroutines.flow.flowOn
+import kotlinx.coroutines.launch
+import kotlinx.coroutines.withContext
+import java.io.File
+import java.io.IOException
+
+/**
+ * JNI wrapper for the llama.cpp library providing Android-friendly access to large language models.
+ *
+ * This class implements a singleton pattern for managing the lifecycle of a single LLM instance.
+ * All operations are executed on a dedicated single-threaded dispatcher to ensure thread safety
+ * with the underlying C++ native code.
+ *
+ * The typical usage flow is:
+ * 1. Get instance via [getInstance]
+ * 2. Load a model with [loadModel]
+ * 3. Send prompts with [sendUserPrompt]
+ * 4. Generate responses as token streams
+ * 5. Perform [cleanUp] when done with a model
+ * 6. Properly [destroy] when completely done
+ *
+ * State transitions are managed automatically and validated at each operation.
+ *
+ * @see ai_chat.cpp for the native implementation details
+ */
+internal class InferenceEngineImpl private constructor(
+    private val nativeLibDir: String
+) : InferenceEngine {
+
+    companion object {
+        private val TAG = InferenceEngineImpl::class.java.simpleName
+
+        @Volatile
+        private var instance: InferenceEngine? = null
+
+        /**
+         * Create or obtain [InferenceEngineImpl]'s single instance.
+         *
+         * @param Context for obtaining native library directory
+         * @throws IllegalArgumentException if native library path is invalid
+         * @throws UnsatisfiedLinkError if library failed to load
+         */
+        internal fun getInstance(context: Context) =
+            instance ?: synchronized(this) {
+                val nativeLibDir = context.applicationInfo.nativeLibraryDir
+                require(nativeLibDir.isNotBlank()) { "Expected a valid native library path!" }
+
+                try {
+                    Log.i(TAG, "Instantiating InferenceEngineImpl,,,")
+                    InferenceEngineImpl(nativeLibDir).also { instance = it }
+                } catch (e: UnsatisfiedLinkError) {
+                    Log.e(TAG, "Failed to load native library from $nativeLibDir", e)
+                    throw e
+                }
+            }
+    }
+
+    /**
+     * JNI methods
+     * @see ai_chat.cpp
+     */
+    @FastNative
+    private external fun init(nativeLibDir: String)
+
+    @FastNative
+    private external fun load(modelPath: String): Int
+
+    @FastNative
+    private external fun prepare(): Int
+
+    @FastNative
+    private external fun systemInfo(): String
+
+    @FastNative
+    private external fun benchModel(pp: Int, tg: Int, pl: Int, nr: Int): String
+
+    @FastNative
+    private external fun processSystemPrompt(systemPrompt: String): Int
+
+    @FastNative
+    private external fun processUserPrompt(userPrompt: String, predictLength: Int): Int
+
+    @FastNative
+    private external fun generateNextToken(): String?
+
+    @FastNative
+    private external fun unload()
+
+    @FastNative
+    private external fun shutdown()
+
+    private val _state =
+        MutableStateFlow<InferenceEngine.State>(InferenceEngine.State.Uninitialized)
+    override val state: StateFlow<InferenceEngine.State> = _state
+
+    private var _readyForSystemPrompt = false
+
+    /**
+     * Single-threaded coroutine dispatcher & scope for LLama asynchronous operations
+     */
+    @OptIn(ExperimentalCoroutinesApi::class)
+    private val llamaDispatcher = Dispatchers.IO.limitedParallelism(1)
+    private val llamaScope = CoroutineScope(llamaDispatcher + SupervisorJob())
+
+    init {
+        llamaScope.launch {
+            try {
+                check(_state.value is InferenceEngine.State.Uninitialized) {
+                    "Cannot load native library in ${_state.value.javaClass.simpleName}!"
+                }
+                _state.value = InferenceEngine.State.Initializing
+                Log.i(TAG, "Loading native library...")
+                System.loadLibrary("ai-chat")
+                init(nativeLibDir)
+                _state.value = InferenceEngine.State.Initialized
+                Log.i(TAG, "Native library loaded! System info: \n${systemInfo()}")
+
+            } catch (e: Exception) {
+                Log.e(TAG, "Failed to load native library", e)
+                throw e
+            }
+        }
+    }
+
+    /**
+     * Load the LLM
+     */
+    override suspend fun loadModel(pathToModel: String) =
+        withContext(llamaDispatcher) {
+            check(_state.value is InferenceEngine.State.Initialized) {
+                "Cannot load model in ${_state.value.javaClass.simpleName}!"
+            }
+
+            try {
+                Log.i(TAG, "Checking access to model file... \n$pathToModel")
+                File(pathToModel).let {
+                    require(it.exists()) { "File not found" }
+                    require(it.isFile) { "Not a valid file" }
+                    require(it.canRead()) { "Cannot read file" }
+                }
+
+                Log.i(TAG, "Loading model... \n$pathToModel")
+                _readyForSystemPrompt = false
+                _state.value = InferenceEngine.State.LoadingModel
+                load(pathToModel).let {
+                    // TODO-han.yin: find a better way to pass other error codes
+                    if (it != 0) throw UnsupportedArchitectureException()
+                }
+                prepare().let {
+                    if (it != 0) throw IOException("Failed to prepare resources")
+                }
+                Log.i(TAG, "Model loaded!")
+                _readyForSystemPrompt = true
+                _state.value = InferenceEngine.State.ModelReady
+            } catch (e: Exception) {
+                Log.e(TAG, (e.message ?: "Error loading model") + "\n" + pathToModel, e)
+                _state.value = InferenceEngine.State.Error(e)
+                throw e
+            }
+        }
+
+    /**
+     * Process the plain text system prompt
+     *
+     * TODO-han.yin: return error code if system prompt not correct processed?
+     */
+    override suspend fun setSystemPrompt(prompt: String) =
+        withContext(llamaDispatcher) {
+            require(prompt.isNotBlank()) { "Cannot process empty system prompt!" }
+            check(_readyForSystemPrompt) { "System prompt must be set ** RIGHT AFTER ** model loaded!" }
+            check(_state.value is InferenceEngine.State.ModelReady) {
+                "Cannot process system prompt in ${_state.value.javaClass.simpleName}!"
+            }
+
+            Log.i(TAG, "Sending system prompt...")
+            _readyForSystemPrompt = false
+            _state.value = InferenceEngine.State.ProcessingSystemPrompt
+            processSystemPrompt(prompt).let { result ->
+                if (result != 0) {
+                    RuntimeException("Failed to process system prompt: $result").also {
+                        _state.value = InferenceEngine.State.Error(it)
+                        throw it
+                    }
+                }
+            }
+            Log.i(TAG, "System prompt processed! Awaiting user prompt...")
+            _state.value = InferenceEngine.State.ModelReady
+        }
+
+    /**
+     * Send plain text user prompt to LLM, which starts generating tokens in a [Flow]
+     */
+    override fun sendUserPrompt(
+        message: String,
+        predictLength: Int,
+    ): Flow<String> = flow {
+        require(message.isNotEmpty()) { "User prompt discarded due to being empty!" }
+        check(_state.value is InferenceEngine.State.ModelReady) {
+            "User prompt discarded due to: ${_state.value.javaClass.simpleName}"
+        }
+
+        try {
+            Log.i(TAG, "Sending user prompt...")
+            _readyForSystemPrompt = false
+            _state.value = InferenceEngine.State.ProcessingUserPrompt
+
+            processUserPrompt(message, predictLength).let { result ->
+                if (result != 0) {
+                    Log.e(TAG, "Failed to process user prompt: $result")
+                    return@flow
+                }
+            }
+
+            Log.i(TAG, "User prompt processed. Generating assistant prompt...")
+            _state.value = InferenceEngine.State.Generating
+            while (true) {
+                generateNextToken()?.let { utf8token ->
+                    if (utf8token.isNotEmpty()) emit(utf8token)
+                } ?: break
+            }
+            Log.i(TAG, "Assistant generation complete. Awaiting user prompt...")
+            _state.value = InferenceEngine.State.ModelReady
+        } catch (e: CancellationException) {
+            Log.i(TAG, "Generation cancelled by user.")
+            _state.value = InferenceEngine.State.ModelReady
+            throw e
+        } catch (e: Exception) {
+            Log.e(TAG, "Error during generation!", e)
+            _state.value = InferenceEngine.State.Error(e)
+            throw e
+        }
+    }.flowOn(llamaDispatcher)
+
+    /**
+     * Benchmark the model
+     */
+    override suspend fun bench(pp: Int, tg: Int, pl: Int, nr: Int): String =
+        withContext(llamaDispatcher) {
+            check(_state.value is InferenceEngine.State.ModelReady) {
+                "Benchmark request discarded due to: $state"
+            }
+            Log.i(TAG, "Start benchmark (pp: $pp, tg: $tg, pl: $pl, nr: $nr)")
+            _readyForSystemPrompt = false   // Just to be safe
+            _state.value = InferenceEngine.State.Benchmarking
+            benchModel(pp, tg, pl, nr).also {
+                _state.value = InferenceEngine.State.ModelReady
+            }
+        }
+
+    /**
+     * Unloads the model and frees resources, or reset error states
+     */
+    override suspend fun cleanUp() =
+        withContext(llamaDispatcher) {
+            when (val state = _state.value) {
+                is InferenceEngine.State.ModelReady -> {
+                    Log.i(TAG, "Unloading model and free resources...")
+                    _readyForSystemPrompt = false
+                    _state.value = InferenceEngine.State.UnloadingModel
+
+                    unload()
+
+                    _state.value = InferenceEngine.State.Initialized
+                    Log.i(TAG, "Model unloaded!")
+                    Unit
+                }
+
+                is InferenceEngine.State.Error -> {
+                    Log.i(TAG, "Resetting error states...")
+                    _state.value = InferenceEngine.State.Initialized
+                    Log.i(TAG, "States reset!")
+                    Unit
+                }
+
+                else -> throw IllegalStateException("Cannot unload model in ${state.javaClass.simpleName}")
+            }
+        }
+
+    /**
+     * Cancel all ongoing coroutines and free GGML backends
+     */
+    override fun destroy() {
+        _readyForSystemPrompt = false
+        llamaScope.cancel()
+        when(_state.value) {
+            is InferenceEngine.State.Uninitialized -> {}
+            is InferenceEngine.State.Initialized -> shutdown()
+            else -> { unload(); shutdown() }
+        }
+    }
+}

examples/llama.android/lib/src/main/java/com/arm/aichat/internal/gguf/GgufMetadataReaderImpl.kt

@@ -0,0 +1,590 @@
+package com.arm.aichat.internal.gguf
+
+import android.content.Context
+import android.net.Uri
+import com.arm.aichat.gguf.GgufMetadata
+import com.arm.aichat.gguf.GgufMetadataReader
+import com.arm.aichat.gguf.InvalidFileFormatException
+import java.io.File
+import java.io.IOException
+import java.io.InputStream
+
+
+/**
+ * Utility class to read GGUF model files and extract metadata key-value pairs.
+ * This parser reads the header and metadata of a GGUF v3 file (little-endian) and skips tensor data.
+ */
+internal class GgufMetadataReaderImpl(
+    private val skipKeys: Set<String>,
+    private val arraySummariseThreshold: Int,
+) : GgufMetadataReader {
+    companion object {
+        private const val ARCH_LLAMA = "llama"
+    }
+
+    /** Enum corresponding to GGUF metadata value types (for convenience and array element typing). */
+    enum class MetadataType(val code: Int) {
+        UINT8(0), INT8(1), UINT16(2), INT16(3),
+        UINT32(4), INT32(5), FLOAT32(6), BOOL(7),
+        STRING(8), ARRAY(9), UINT64(10), INT64(11), FLOAT64(12);
+        companion object {
+            private val codeMap = entries.associateBy(MetadataType::code)
+            fun fromCode(code: Int): MetadataType = codeMap[code]
+                ?: throw IOException("Unknown metadata value type code: $code")
+        }
+    }
+
+    /** Sealed class hierarchy for metadata values, providing type-safe representations for each GGUF metadata type. */
+    sealed class MetadataValue {
+        data class UInt8(val value: UByte) : MetadataValue()       // 0:  8-bit unsigned int
+        data class Int8(val value: Byte) : MetadataValue()         // 1:  8-bit signed int
+        data class UInt16(val value: UShort) : MetadataValue()     // 2:  16-bit unsigned int (little-endian)
+        data class Int16(val value: Short) : MetadataValue()       // 3:  16-bit signed int (little-endian)
+        data class UInt32(val value: UInt) : MetadataValue()       // 4:  32-bit unsigned int (little-endian)
+        data class Int32(val value: Int) : MetadataValue()         // 5:  32-bit signed int (little-endian)
+        data class Float32(val value: Float) : MetadataValue()     // 6:  32-bit IEEE754 float
+        data class Bool(val value: Boolean) : MetadataValue()      // 7:  Boolean (1-byte, 0=false, 1=true)
+        data class StringVal(val value: String) : MetadataValue()  // 8:  UTF-8 string (length-prefixed)
+        data class ArrayVal(val elementType: MetadataType, val elements: List<MetadataValue>) : MetadataValue()
+        data class UInt64(val value: ULong) : MetadataValue()      // 10: 64-bit unsigned int (little-endian)
+        data class Int64(val value: Long) : MetadataValue()        // 11: 64-bit signed int (little-endian)
+        data class Float64(val value: Double) : MetadataValue()    // 12: 64-bit IEEE754 double
+    }
+
+    /* Convert MetadataValue to plain Kotlin primitives for allMetadata map */
+    private fun MetadataValue.toPrimitive(): Any = when (this) {
+        is MetadataValue.UInt8     -> value
+        is MetadataValue.Int8      -> value
+        is MetadataValue.UInt16    -> value
+        is MetadataValue.Int16     -> value
+        is MetadataValue.UInt32    -> value
+        is MetadataValue.Int32     -> value
+        is MetadataValue.Float32   -> value
+        is MetadataValue.Bool      -> value
+        is MetadataValue.StringVal -> value
+        is MetadataValue.UInt64    -> value
+        is MetadataValue.Int64     -> value
+        is MetadataValue.Float64   -> value
+        is MetadataValue.ArrayVal  -> elements.map { it.toPrimitive() }
+    }
+
+    /**
+     * Reads the magic number from the specified file path.
+     *
+     * @param context Context for obtaining ContentResolver
+     * @param uri Uri to the GGUF file provided by ContentProvider
+     * @return true if file is valid GGUF, otherwise false
+     */
+    override suspend fun ensureSourceFileFormat(file: File): Boolean =
+        file.inputStream().buffered().use { ensureMagic(it) }
+
+    /**
+     * Reads the magic number from the specified file path.
+     *
+     * @param context Context for obtaining ContentResolver
+     * @param uri Uri to the GGUF file provided by ContentProvider
+     * @return true if file is valid GGUF, otherwise false
+     */
+    override suspend fun ensureSourceFileFormat(context: Context, uri: Uri): Boolean =
+        context.contentResolver.openInputStream(uri)?.buffered()?.use { ensureMagic(it) } == true
+
+    /** Reads the 4‑byte magic; throws if magic ≠ "GGUF". */
+    private fun ensureMagic(input: InputStream): Boolean =
+        ByteArray(4).let {
+            if (input.read(it) != 4) throw IOException("Not a valid file!")
+            it.contentEquals(byteArrayOf(0x47, 0x47, 0x55, 0x46)) // "GGUF"
+        }
+
+    /**
+     * High‑level entry point: parses a `.gguf` file on disk and returns the fully
+     * populated [GgufMetadata] tree.
+     *
+     * Steps performed internally:
+     * 1.  Reads and validates the 8‑byte header (`"GGUF"` magic + version).
+     * 2.  Streams through the key‑value section, skipping large blobs if the key
+     *     appears in [skipKeys] or if an array exceeds [arraySummariseThreshold].
+     * 3.  Converts the resulting raw map into strongly‑typed sub‑structures
+     *     (basic info, tokenizer, rope, etc.).
+     *
+     * The method is STREAMING‑ONLY: tensors are never mapped or loaded into
+     * memory, so even multi‑GB model files can be processed in < 50 ms.
+     *
+     * @param path Absolute or relative filesystem path to a `.gguf` file.
+     * @return A [GgufMetadata] instance containing all recognised metadata plus
+     *         an `allMetadata` map with any keys that were not given a dedicated
+     *         field.
+     * @throws IOException if the file is not GGUF, the version is unsupported,
+     *         or the metadata block is truncated / corrupt.
+     */
+    override suspend fun readStructuredMetadata(input: InputStream): GgufMetadata {
+        // ── 1. header ──────────────────────────────────────────────────────────
+        // throws on mismatch
+        val version       = ensureMagicAndVersion(input)
+        val tensorCount   = readLittleLong(input)
+        val kvCount       = readLittleLong(input)
+
+        // ── 2. metadata map (reuse our raw parser, but we need access to the stream) ──
+        val meta = readMetaMap(input, kvCount)    // <String, MetadataValue>
+
+        // ── 3. build structured object ────────────────────────────────────────
+        return buildStructured(meta, version, tensorCount, kvCount)
+    }
+
+    /** Reads the 4‑byte magic + 4‑byte version; throws if magic ≠ "GGUF". */
+    private fun ensureMagicAndVersion(input: InputStream): GgufMetadata.GgufVersion {
+        if (!ensureMagic(input)) throw InvalidFileFormatException()
+        return GgufMetadata.GgufVersion.fromCode(readLEUInt32(input))
+    }
+
+    /**
+     * Read an unsigned 32‑bit little‑endian integer.
+     *
+     * @throws IOException if fewer than four bytes are available.
+     */
+    private fun readLEUInt32(input: InputStream): Int {
+        val b0 = input.read(); val b1 = input.read(); val b2 = input.read(); val b3 = input.read()
+        if (b3 == -1) throw IOException("Unexpected EOF while reading UInt32")
+        return (b3 and 0xFF shl 24) or
+            (b2 and 0xFF shl 16) or
+            (b1 and 0xFF shl  8) or
+            (b0 and 0xFF)
+    }
+
+    /**
+     * Low‑level helper that reads the entire “key-value” section from the current
+     * stream position.
+     *
+     * @param input  Open stream positioned JUST AFTER the header.
+     * @param kvCnt  Number of key‑value pairs (taken from the header).
+     * @return       Mutable map with one [MetadataValue] for every key that is NOT skipped.
+     *
+     * The function honours [skipKeys] and [arraySummariseThreshold] by invoking
+     * [skipValue] or [parseValue] accordingly.
+     */
+    private fun readMetaMap(input: InputStream, kvCnt: Long): Map<String, MetadataValue> =
+        mutableMapOf<String, MetadataValue>().apply {
+             repeat(kvCnt.toInt()) {
+                 val key = readString(input)
+                 val valueT = MetadataType.fromCode(littleEndianBytesToInt(input.readNBytesExact(4)))
+                 if (key in skipKeys) {
+                     skipValue(input, valueT)
+                 } else {
+                     this[key] = parseValue(input, valueT)
+                 }
+             }
+         }
+
+    /**
+     * Converts a flat [Map]<[String], [MetadataValue]> into the strongly‑typed
+     * [GgufMetadata] tree used by the rest of the app.
+     *
+     * Only the keys listed in the spec are copied into dedicated data classes;
+     * everything else is preserved in `GgufMetadata.allMetadata`.
+     *
+     * @param m            Raw key/value map.
+     * @param version      GGUF file‑format version (enum).
+     * @param tensorCnt    Number of tensors (from the header).
+     * @param kvCnt        Total metadata pair count (from the header).
+     */
+    private fun buildStructured(
+        m: Map<String, MetadataValue>,
+        version: GgufMetadata.GgufVersion,
+        tensorCnt: Long,
+        kvCnt: Long
+    ): GgufMetadata {
+        // ---------- helpers ----------
+        fun String.str()  = (m[this] as? MetadataValue.StringVal)?.value
+        fun String.bool() = (m[this] as? MetadataValue.Bool)?.value
+        fun String.i32()  = (m[this] as? MetadataValue.Int32)?.value
+        fun String.u32()  = (m[this] as? MetadataValue.UInt32)?.value?.toInt()
+        fun String.f32()  = (m[this] as? MetadataValue.Float32)?.value
+        fun String.f64()  = (m[this] as? MetadataValue.Float64)?.value?.toFloat()
+        fun String.strList(): List<String>? =
+            (m[this] as? MetadataValue.ArrayVal)
+                ?.elements
+                ?.mapNotNull { (it as? MetadataValue.StringVal)?.value }
+
+        val arch = "general.architecture".str() ?: ARCH_LLAMA
+
+        // -------------- populate sections ----------------
+        val basic = GgufMetadata.BasicInfo(
+            uuid      = "general.uuid".str(),
+            name      = "general.basename".str(),
+            nameLabel = "general.name".str(),
+            sizeLabel = "general.size_label".str()
+        )
+
+        val author = GgufMetadata.AuthorInfo(
+            organization = "general.organization".str(),
+            author       = "general.author".str(),
+            doi          = "general.doi".str(),
+            url          = "general.url".str(),
+            repoUrl      = "general.repo_url".str(),
+            license      = "general.license".str(),
+            licenseLink  = "general.license.link".str()
+        ).takeUnless {
+            organization == null && author == null && doi == null &&
+                url == null && repoUrl == null && license == null && licenseLink == null
+        }
+
+        val additional = GgufMetadata.AdditionalInfo(
+            type        = "general.type".str(),
+            description = "general.description".str(),
+            tags        = "general.tags".strList(),
+            languages   = "general.languages".strList()
+        ).takeUnless {
+            type == null && description == null && tags == null && languages == null
+        }
+
+        val architectureInfo = GgufMetadata.ArchitectureInfo(
+            architecture        = arch,
+            fileType            = "general.file_type".u32(),
+            vocabSize           = "$arch.vocab_size".u32(),
+            finetune            = "general.finetune".str(),
+            quantizationVersion = "general.quantization_version".u32()
+        ).takeUnless { fileType == null && vocabSize == null && finetune == null && quantizationVersion == null }
+
+        val baseModels = buildList {
+            val n = "general.base_model.count".u32() ?: 0
+            for (i in 0 until n) {
+                fun k(s: String) = "general.base_model.$i.$s"
+                add(
+                    GgufMetadata.BaseModelInfo(
+                        name         = k("name").str(),
+                        author       = k("author").str(),
+                        version      = k("version").str(),
+                        organization = k("organization").str(),
+                        url          = k("url").str(),
+                        doi          = k("doi").str(),
+                        uuid         = k("uuid").str(),
+                        repoUrl      = k("repo_url").str(),
+                    )
+                )
+            }
+        }.takeIf { it.isNotEmpty() }
+
+        val tokenizer = GgufMetadata.TokenizerInfo(
+            model            = "tokenizer.ggml.model".str(),
+            bosTokenId       = "tokenizer.ggml.bos_token_id".u32(),
+            eosTokenId       = "tokenizer.ggml.eos_token_id".u32(),
+            unknownTokenId   = "tokenizer.ggml.unknown_token_id".u32(),
+            paddingTokenId   = "tokenizer.ggml.padding_token_id".u32(),
+            addBosToken      = "tokenizer.ggml.add_bos_token".bool(),
+            addEosToken      = "tokenizer.ggml.add_eos_token".bool(),
+            chatTemplate     = "tokenizer.chat_template".str()
+        ).takeUnless { model == null && bosTokenId == null && eosTokenId == null &&
+            unknownTokenId == null && paddingTokenId == null &&
+            addBosToken == null && addEosToken == null && chatTemplate == null
+        }
+
+        val dimensions = GgufMetadata.DimensionsInfo(
+            contextLength    = "$arch.context_length".u32(),
+            embeddingSize    = "$arch.embedding_length".u32(),
+            blockCount       = "$arch.block_count".u32(),
+            feedForwardSize  = "$arch.feed_forward_length".u32()
+        ).takeUnless { contextLength == null && embeddingSize == null && blockCount == null && feedForwardSize == null }
+
+        val attention = GgufMetadata.AttentionInfo(
+            headCount             = "$arch.attention.head_count".u32(),
+            headCountKv           = "$arch.attention.head_count_kv".u32(),
+            keyLength             = "$arch.attention.key_length".u32(),
+            valueLength           = "$arch.attention.value_length".u32(),
+            layerNormEpsilon      = "$arch.attention.layer_norm_epsilon".f32(),
+            layerNormRmsEpsilon   = "$arch.attention.layer_norm_rms_epsilon".f32(),
+        ).takeUnless { headCount == null && headCountKv == null && keyLength == null && valueLength == null &&
+            layerNormEpsilon == null && layerNormRmsEpsilon == null
+        }
+
+        val rope = GgufMetadata.RopeInfo(
+            frequencyBase          = "$arch.rope.freq_base".f32(),
+            dimensionCount         = "$arch.rope.dimension_count".u32(),
+            scalingType            = "$arch.rope.scaling.type".str(),
+            scalingFactor          = "$arch.rope.scaling.factor".f32(),
+            attnFactor             = "$arch.rope.scaling.attn_factor".f32(),
+            originalContextLength  = "$arch.rope.scaling.original_context_length".u32(),
+            finetuned              = "$arch.rope.scaling.finetuned".bool()
+        ).takeUnless { frequencyBase == null && dimensionCount == null &&
+            scalingType == null && scalingFactor == null && attnFactor == null &&
+            originalContextLength == null && finetuned == null
+        }
+
+        val experts = GgufMetadata.ExpertsInfo(
+            count      = "$arch.expert_count".u32(),
+            usedCount  = "$arch.expert_used_count".u32()
+        ).takeUnless { count == null && usedCount == null }
+
+        return GgufMetadata(
+            version = version,
+            tensorCount = tensorCnt,
+            kvCount = kvCnt,
+            basic = basic,
+            author = author,
+            additional = additional,
+            architecture = architectureInfo,
+            baseModels = baseModels,
+            tokenizer = tokenizer,
+            dimensions = dimensions,
+            attention = attention,
+            rope = rope,
+            experts = experts
+        )
+    }
+
+    /**
+     * Recursively parses a metadata value of the given type from the input stream.
+     * @param input The input stream positioned at the start of the value.
+     * @param type The metadata value type to parse.
+     */
+    private fun parseValue(input: InputStream, type: MetadataType): MetadataValue = when (type) {
+        MetadataType.UINT8 -> {
+            // 1-byte unsigned integer
+            val byteVal = input.read()
+            if (byteVal == -1) throw IOException("Unexpected EOF while reading uint8 value.")
+            MetadataValue.UInt8(byteVal.toUByte())
+        }
+        MetadataType.INT8 -> {
+            // 1-byte signed integer
+            val byteVal = input.read()
+            if (byteVal == -1) throw IOException("Unexpected EOF while reading int8 value.")
+            MetadataValue.Int8(byteVal.toByte())
+        }
+        MetadataType.UINT16 -> {
+            // 2-byte unsigned integer (little-endian)
+            val bytes = ByteArray(2)
+            if (input.read(bytes) != 2) throw IOException("Unexpected EOF while reading uint16 value.")
+            // Combine two bytes (little-endian) into an unsigned 16-bit value
+            val u16 = ((bytes[1].toInt() and 0xFF) shl 8) or (bytes[0].toInt() and 0xFF)
+            MetadataValue.UInt16(u16.toUShort())
+        }
+        MetadataType.INT16 -> {
+            // 2-byte signed integer (little-endian)
+            val bytes = ByteArray(2)
+            if (input.read(bytes) != 2) throw IOException("Unexpected EOF while reading int16 value.")
+            // Combine to 16-bit and interpret as signed
+            val i16 = ((bytes[1].toInt() and 0xFF) shl 8) or (bytes[0].toInt() and 0xFF)
+            MetadataValue.Int16(i16.toShort())
+        }
+        MetadataType.UINT32 -> {
+            // 4-byte unsigned integer (little-endian)
+            val bytes = ByteArray(4)
+            if (input.read(bytes) != 4) throw IOException("Unexpected EOF while reading uint32 value.")
+            // Combine four bytes into a 32-bit value (as Long to avoid overflow), then convert to UInt
+            val u32 = (bytes[3].toLong() and 0xFFL shl 24) or
+                (bytes[2].toLong() and 0xFFL shl 16) or
+                (bytes[1].toLong() and 0xFFL shl 8) or
+                (bytes[0].toLong() and 0xFFL)
+            MetadataValue.UInt32(u32.toUInt())
+        }
+        MetadataType.INT32 -> {
+            // 4-byte signed integer (little-endian)
+            val bytes = ByteArray(4)
+            if (input.read(bytes) != 4) throw IOException("Unexpected EOF while reading int32 value.")
+            // Combine four bytes into a 32-bit signed int
+            val i32 = (bytes[3].toInt() and 0xFF shl 24) or
+                (bytes[2].toInt() and 0xFF shl 16) or
+                (bytes[1].toInt() and 0xFF shl 8) or
+                (bytes[0].toInt() and 0xFF)
+            MetadataValue.Int32(i32)
+        }
+        MetadataType.FLOAT32 -> {
+            // 4-byte IEEE 754 float (little-endian)
+            val bytes = ByteArray(4)
+            if (input.read(bytes) != 4) throw IOException("Unexpected EOF while reading float32 value.")
+            // Assemble 4 bytes into a 32-bit int bit-pattern, then convert to Float
+            val bits = (bytes[3].toInt() and 0xFF shl 24) or
+                (bytes[2].toInt() and 0xFF shl 16) or
+                (bytes[1].toInt() and 0xFF shl 8) or
+                (bytes[0].toInt() and 0xFF)
+            val floatVal = Float.fromBits(bits)
+            MetadataValue.Float32(floatVal)
+        }
+        MetadataType.BOOL -> {
+            // 1-byte boolean (0 = false, 1 = true)
+            val byteVal = input.read()
+            if (byteVal == -1) throw IOException("Unexpected EOF while reading boolean value.")
+            if (byteVal != 0 && byteVal != 1) {
+                throw IOException("Invalid boolean value: $byteVal (must be 0 or 1).")
+            }
+            MetadataValue.Bool(byteVal != 0)
+        }
+        MetadataType.STRING -> {
+            // UTF-8 string (length-prefixed with 8-byte length)
+            val str = readString(input)
+            MetadataValue.StringVal(str)
+        }
+        MetadataType.ARRAY -> {
+            val elemType = MetadataType.fromCode(littleEndianBytesToInt(input.readNBytesExact(4)))
+            val len      = readLittleLong(input)
+            val count    = len.toInt()
+
+            if (arraySummariseThreshold >= 0 && count > arraySummariseThreshold) {
+                // fast‑forward without allocation
+                repeat(count) { skipValue(input, elemType) }
+                MetadataValue.StringVal("Array($elemType, $count items) /* summarised */")
+            } else {
+                val list = ArrayList<MetadataValue>(count)
+                repeat(count) { list += parseValue(input, elemType) }
+                MetadataValue.ArrayVal(elemType, list)
+            }
+        }
+        MetadataType.UINT64 -> {
+            // 8-byte unsigned integer (little-endian)
+            val bytes = ByteArray(8)
+            if (input.read(bytes) != 8) throw IOException("Unexpected EOF while reading uint64 value.")
+            // Combine 8 bytes into an unsigned 64-bit (ULong). Use ULong for full 0 to 2^64-1 range.
+            val u64 = (bytes[7].toULong() and 0xFFuL shl 56) or
+                (bytes[6].toULong() and 0xFFuL shl 48) or
+                (bytes[5].toULong() and 0xFFuL shl 40) or
+                (bytes[4].toULong() and 0xFFuL shl 32) or
+                (bytes[3].toULong() and 0xFFuL shl 24) or
+                (bytes[2].toULong() and 0xFFuL shl 16) or
+                (bytes[1].toULong() and 0xFFuL shl 8) or
+                (bytes[0].toULong() and 0xFFuL)
+            MetadataValue.UInt64(u64)
+        }
+        MetadataType.INT64 -> {
+            // 8-byte signed integer (little-endian)
+            val bytes = ByteArray(8)
+            if (input.read(bytes) != 8) throw IOException("Unexpected EOF while reading int64 value.")
+            // Combine 8 bytes into a signed 64-bit value (Long)
+            val i64 = (bytes[7].toLong() and 0xFFL shl 56) or
+                (bytes[6].toLong() and 0xFFL shl 48) or
+                (bytes[5].toLong() and 0xFFL shl 40) or
+                (bytes[4].toLong() and 0xFFL shl 32) or
+                (bytes[3].toLong() and 0xFFL shl 24) or
+                (bytes[2].toLong() and 0xFFL shl 16) or
+                (bytes[1].toLong() and 0xFFL shl 8) or
+                (bytes[0].toLong() and 0xFFL)
+            MetadataValue.Int64(i64)
+        }
+        MetadataType.FLOAT64 -> {
+            // 8-byte IEEE 754 double (little-endian)
+            val bytes = ByteArray(8)
+            if (input.read(bytes) != 8) throw IOException("Unexpected EOF while reading float64 value.")
+            // Assemble 8 bytes into a 64-bit bit-pattern, then convert to Double
+            val bits = (bytes[7].toLong() and 0xFFL shl 56) or
+                (bytes[6].toLong() and 0xFFL shl 48) or
+                (bytes[5].toLong() and 0xFFL shl 40) or
+                (bytes[4].toLong() and 0xFFL shl 32) or
+                (bytes[3].toLong() and 0xFFL shl 24) or
+                (bytes[2].toLong() and 0xFFL shl 16) or
+                (bytes[1].toLong() and 0xFFL shl 8) or
+                (bytes[0].toLong() and 0xFFL)
+            val doubleVal = Double.fromBits(bits)
+            MetadataValue.Float64(doubleVal)
+        }
+    }
+
+
+    private fun <T> T?.takeUnless(check: T.() -> Boolean): T? =
+        this?.takeIf { !it.check() }
+
+    /** Helper: Skip a value in the stream without storing it (still maintains pointer). */
+    private fun skipValue(input: InputStream, type: MetadataType) {
+        when (type) {
+            MetadataType.UINT8, MetadataType.INT8, MetadataType.BOOL -> input.skipFully(1)
+            MetadataType.UINT16, MetadataType.INT16                  -> input.skipFully(2)
+            MetadataType.UINT32, MetadataType.INT32, MetadataType.FLOAT32 -> input.skipFully(4)
+            MetadataType.UINT64, MetadataType.INT64, MetadataType.FLOAT64 -> input.skipFully(8)
+            MetadataType.STRING -> {
+                val len = readLittleLong(input); input.skipFully(len)
+            }
+            MetadataType.ARRAY -> {
+                val elemType = MetadataType.fromCode(littleEndianBytesToInt(input.readNBytesExact(4)))
+                val len      = readLittleLong(input)
+                repeat(len.toInt()) { skipValue(input, elemType) }   // recursive skip
+            }
+        }
+    }
+
+    /** Helper: Read an 8-byte little-endian unsigned value and return it as a signed Long (assuming it fits in 63 bits). */
+    private fun readLittleLong(input: InputStream): Long {
+        val bytes = ByteArray(8)
+        input.readFully(bytes)
+
+        // Combine 8 bytes into a 64-bit value (Little Endian).
+        // Note: If the value exceeds Long.MAX_VALUE (bit 63 is 1), this will produce a negative Long (two's complement).
+        // In our context (lengths/counts), such extremely large values are not expected.
+        return (bytes[7].toLong() and 0xFFL shl 56) or
+            (bytes[6].toLong() and 0xFFL shl 48) or
+            (bytes[5].toLong() and 0xFFL shl 40) or
+            (bytes[4].toLong() and 0xFFL shl 32) or
+            (bytes[3].toLong() and 0xFFL shl 24) or
+            (bytes[2].toLong() and 0xFFL shl 16) or
+            (bytes[1].toLong() and 0xFFL shl 8) or
+            (bytes[0].toLong() and 0xFFL)
+    }
+
+    /** Helper: Read a GGUF string from the stream (8-byte length followed by UTF-8 bytes). */
+    private fun readString(input: InputStream): String =
+        // Read 8-byte little-endian length (number of bytes in the string).
+        readLittleLong(input).let { len ->
+            if (len < 0 || len > Int.MAX_VALUE) throw IOException("String too long: $len")
+
+            // Read the UTF-8 bytes of the given length.
+            ByteArray(len.toInt()).let {
+                if (it.isNotEmpty()) input.readFully(it)
+                String(it, Charsets.UTF_8)
+            }
+        }
+
+    /** Helper: Convert a 4-byte little-endian byte array to a 32-bit integer. */
+    private fun littleEndianBytesToInt(bytes: ByteArray): Int =
+        // Note: assumes bytes length is 4.
+        (bytes[3].toInt() and 0xFF shl 24) or
+            (bytes[2].toInt() and 0xFF shl 16) or
+            (bytes[1].toInt() and 0xFF shl 8) or
+            (bytes[0].toInt() and 0xFF)
+
+    /**
+     * Robust skip that works the same on JDK 11 and Android’s desugared runtime.
+     *
+     * @param n  Number of bytes to advance in the stream.
+     * @throws IOException on premature EOF.
+     */
+    private fun InputStream.skipFully(n: Long) {
+        var remaining = n
+        val scratch = ByteArray(8192)                 // read‑and‑toss buffer
+        while (remaining > 0) {
+            val skipped = skip(remaining)
+            when {
+                skipped > 0      -> remaining -= skipped               // normal fast path
+                skipped == 0L    -> {
+                    // fallback: read and discard
+                    val read = read(scratch, 0, minOf(remaining, scratch.size.toLong()).toInt())
+                    if (read == -1) throw IOException("EOF while skipping $n bytes")
+                    remaining -= read
+                }
+                else             -> throw IOException("Skip returned negative value")
+            }
+        }
+    }
+
+    /**
+     * Extension that keeps reading until the requested number of bytes are filled.
+     * Falls back to `read()` when `skip()` returns 0, which happens on some Android
+     * streams.
+     *
+     * @param buf  Destination buffer.
+     * @param len  Number of bytes to fill (defaults to `buf.size`).
+     * @throws IOException on premature EOF.
+     */
+    private fun InputStream.readFully(buf: ByteArray, len: Int = buf.size) {
+        var off = 0
+        while (off < len) {
+            val n = read(buf, off, len - off)
+            if (n == -1) throw IOException("EOF after $off of $len bytes")
+            off += n
+        }
+    }
+
+    /**
+     * Read EXACTLY `n` bytes or throw – never returns a partially‑filled array.
+     * This is used for small fixed‑length reads (e.g. 4‑byte type codes).
+     *
+     * @throws IOException on premature EOF.
+     */
+    private fun InputStream.readNBytesExact(n: Int) = ByteArray(n).also {
+        if (read(it) != n) throw IOException("Unexpected EOF")
+    }
+}

examples/llama.android/llama/build.gradle.kts

@@ -1,71 +0,0 @@
-plugins {
-    id("com.android.library")
-    id("org.jetbrains.kotlin.android")
-}
-
-android {
-    namespace = "android.llama.cpp"
-    compileSdk = 34
-
-    defaultConfig {
-        minSdk = 33
-
-        testInstrumentationRunner = "androidx.test.runner.AndroidJUnitRunner"
-        consumerProguardFiles("consumer-rules.pro")
-        ndk {
-            // Add NDK properties if wanted, e.g.
-            // abiFilters += listOf("arm64-v8a")
-        }
-        externalNativeBuild {
-            cmake {
-                arguments += "-DLLAMA_CURL=OFF"
-                arguments += "-DLLAMA_BUILD_COMMON=ON"
-                arguments += "-DGGML_LLAMAFILE=OFF"
-                arguments += "-DCMAKE_BUILD_TYPE=Release"
-                cppFlags += listOf()
-                arguments += listOf()
-
-                cppFlags("")
-            }
-        }
-    }
-
-    buildTypes {
-        release {
-            isMinifyEnabled = false
-            proguardFiles(
-                getDefaultProguardFile("proguard-android-optimize.txt"),
-                "proguard-rules.pro"
-            )
-        }
-    }
-    externalNativeBuild {
-        cmake {
-            path("src/main/cpp/CMakeLists.txt")
-            version = "3.22.1"
-        }
-    }
-    compileOptions {
-        sourceCompatibility = JavaVersion.VERSION_1_8
-        targetCompatibility = JavaVersion.VERSION_1_8
-    }
-    kotlinOptions {
-        jvmTarget = "1.8"
-    }
-
-    packaging {
-        resources {
-            excludes += "/META-INF/{AL2.0,LGPL2.1}"
-        }
-    }
-}
-
-dependencies {
-
-    implementation("androidx.core:core-ktx:1.12.0")
-    implementation("androidx.appcompat:appcompat:1.6.1")
-    implementation("com.google.android.material:material:1.11.0")
-    testImplementation("junit:junit:4.13.2")
-    androidTestImplementation("androidx.test.ext:junit:1.1.5")
-    androidTestImplementation("androidx.test.espresso:espresso-core:3.5.1")
-}

examples/llama.android/llama/src/main/cpp/CMakeLists.txt

@@ -1,53 +0,0 @@
-# For more information about using CMake with Android Studio, read the
-# documentation: https://d.android.com/studio/projects/add-native-code.html.
-# For more examples on how to use CMake, see https://github.com/android/ndk-samples.
-
-# Sets the minimum CMake version required for this project.
-cmake_minimum_required(VERSION 3.22.1)
-
-# Declares the project name. The project name can be accessed via ${ PROJECT_NAME},
-# Since this is the top level CMakeLists.txt, the project name is also accessible
-# with ${CMAKE_PROJECT_NAME} (both CMake variables are in-sync within the top level
-# build script scope).
-project("llama-android")
-
-#include(FetchContent)
-#FetchContent_Declare(
-#        llama
-#        GIT_REPOSITORY https://github.com/ggml-org/llama.cpp
-#        GIT_TAG        master
-#)
-
-# Also provides "common"
-#FetchContent_MakeAvailable(llama)
-
-# Creates and names a library, sets it as either STATIC
-# or SHARED, and provides the relative paths to its source code.
-# You can define multiple libraries, and CMake builds them for you.
-# Gradle automatically packages shared libraries with your APK.
-#
-# In this top level CMakeLists.txt, ${CMAKE_PROJECT_NAME} is used to define
-# the target library name; in the sub-module's CMakeLists.txt, ${PROJECT_NAME}
-# is preferred for the same purpose.
-#
-
-#load local llama.cpp
-add_subdirectory(../../../../../../ build-llama)
-
-# In order to load a library into your app from Java/Kotlin, you must call
-# System.loadLibrary() and pass the name of the library defined here;
-# for GameActivity/NativeActivity derived applications, the same library name must be
-# used in the AndroidManifest.xml file.
-add_library(${CMAKE_PROJECT_NAME} SHARED
-        # List C/C++ source files with relative paths to this CMakeLists.txt.
-        llama-android.cpp)
-
-# Specifies libraries CMake should link to your target library. You
-# can link libraries from various origins, such as libraries defined in this
-# build script, prebuilt third-party libraries, or Android system libraries.
-target_link_libraries(${CMAKE_PROJECT_NAME}
-        # List libraries link to the target library
-        llama
-        common
-        android
-        log)

examples/llama.android/llama/src/main/cpp/llama-android.cpp

@@ -1,452 +0,0 @@
-#include <android/log.h>
-#include <jni.h>
-#include <iomanip>
-#include <math.h>
-#include <string>
-#include <unistd.h>
-#include "llama.h"
-#include "common.h"
-
-// Write C++ code here.
-//
-// Do not forget to dynamically load the C++ library into your application.
-//
-// For instance,
-//
-// In MainActivity.java:
-//    static {
-//       System.loadLibrary("llama-android");
-//    }
-//
-// Or, in MainActivity.kt:
-//    companion object {
-//      init {
-//         System.loadLibrary("llama-android")
-//      }
-//    }
-
-#define TAG "llama-android.cpp"
-#define LOGi(...) __android_log_print(ANDROID_LOG_INFO, TAG, __VA_ARGS__)
-#define LOGe(...) __android_log_print(ANDROID_LOG_ERROR, TAG, __VA_ARGS__)
-
-jclass la_int_var;
-jmethodID la_int_var_value;
-jmethodID la_int_var_inc;
-
-std::string cached_token_chars;
-
-bool is_valid_utf8(const char * string) {
-    if (!string) {
-        return true;
-    }
-
-    const unsigned char * bytes = (const unsigned char *)string;
-    int num;
-
-    while (*bytes != 0x00) {
-        if ((*bytes & 0x80) == 0x00) {
-            // U+0000 to U+007F
-            num = 1;
-        } else if ((*bytes & 0xE0) == 0xC0) {
-            // U+0080 to U+07FF
-            num = 2;
-        } else if ((*bytes & 0xF0) == 0xE0) {
-            // U+0800 to U+FFFF
-            num = 3;
-        } else if ((*bytes & 0xF8) == 0xF0) {
-            // U+10000 to U+10FFFF
-            num = 4;
-        } else {
-            return false;
-        }
-
-        bytes += 1;
-        for (int i = 1; i < num; ++i) {
-            if ((*bytes & 0xC0) != 0x80) {
-                return false;
-            }
-            bytes += 1;
-        }
-    }
-
-    return true;
-}
-
-static void log_callback(ggml_log_level level, const char * fmt, void * data) {
-    if (level == GGML_LOG_LEVEL_ERROR)     __android_log_print(ANDROID_LOG_ERROR, TAG, fmt, data);
-    else if (level == GGML_LOG_LEVEL_INFO) __android_log_print(ANDROID_LOG_INFO, TAG, fmt, data);
-    else if (level == GGML_LOG_LEVEL_WARN) __android_log_print(ANDROID_LOG_WARN, TAG, fmt, data);
-    else __android_log_print(ANDROID_LOG_DEFAULT, TAG, fmt, data);
-}
-
-extern "C"
-JNIEXPORT jlong JNICALL
-Java_android_llama_cpp_LLamaAndroid_load_1model(JNIEnv *env, jobject, jstring filename) {
-    llama_model_params model_params = llama_model_default_params();
-
-    auto path_to_model = env->GetStringUTFChars(filename, 0);
-    LOGi("Loading model from %s", path_to_model);
-
-    auto model = llama_model_load_from_file(path_to_model, model_params);
-    env->ReleaseStringUTFChars(filename, path_to_model);
-
-    if (!model) {
-        LOGe("load_model() failed");
-        env->ThrowNew(env->FindClass("java/lang/IllegalStateException"), "load_model() failed");
-        return 0;
-    }
-
-    return reinterpret_cast<jlong>(model);
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_free_1model(JNIEnv *, jobject, jlong model) {
-    llama_model_free(reinterpret_cast<llama_model *>(model));
-}
-
-extern "C"
-JNIEXPORT jlong JNICALL
-Java_android_llama_cpp_LLamaAndroid_new_1context(JNIEnv *env, jobject, jlong jmodel) {
-    auto model = reinterpret_cast<llama_model *>(jmodel);
-
-    if (!model) {
-        LOGe("new_context(): model cannot be null");
-        env->ThrowNew(env->FindClass("java/lang/IllegalArgumentException"), "Model cannot be null");
-        return 0;
-    }
-
-    int n_threads = std::max(1, std::min(8, (int) sysconf(_SC_NPROCESSORS_ONLN) - 2));
-    LOGi("Using %d threads", n_threads);
-
-    llama_context_params ctx_params = llama_context_default_params();
-
-    ctx_params.n_ctx           = 2048;
-    ctx_params.n_threads       = n_threads;
-    ctx_params.n_threads_batch = n_threads;
-
-    llama_context * context = llama_new_context_with_model(model, ctx_params);
-
-    if (!context) {
-        LOGe("llama_new_context_with_model() returned null)");
-        env->ThrowNew(env->FindClass("java/lang/IllegalStateException"),
-                      "llama_new_context_with_model() returned null)");
-        return 0;
-    }
-
-    return reinterpret_cast<jlong>(context);
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_free_1context(JNIEnv *, jobject, jlong context) {
-    llama_free(reinterpret_cast<llama_context *>(context));
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_backend_1free(JNIEnv *, jobject) {
-    llama_backend_free();
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_log_1to_1android(JNIEnv *, jobject) {
-    llama_log_set(log_callback, NULL);
-}
-
-extern "C"
-JNIEXPORT jstring JNICALL
-Java_android_llama_cpp_LLamaAndroid_bench_1model(
-        JNIEnv *env,
-        jobject,
-        jlong context_pointer,
-        jlong model_pointer,
-        jlong batch_pointer,
-        jint pp,
-        jint tg,
-        jint pl,
-        jint nr
-        ) {
-    auto pp_avg = 0.0;
-    auto tg_avg = 0.0;
-    auto pp_std = 0.0;
-    auto tg_std = 0.0;
-
-    const auto context = reinterpret_cast<llama_context *>(context_pointer);
-    const auto model = reinterpret_cast<llama_model *>(model_pointer);
-    const auto batch = reinterpret_cast<llama_batch *>(batch_pointer);
-
-    const int n_ctx = llama_n_ctx(context);
-
-    LOGi("n_ctx = %d", n_ctx);
-
-    int i, j;
-    int nri;
-    for (nri = 0; nri < nr; nri++) {
-        LOGi("Benchmark prompt processing (pp)");
-
-        common_batch_clear(*batch);
-
-        const int n_tokens = pp;
-        for (i = 0; i < n_tokens; i++) {
-            common_batch_add(*batch, 0, i, { 0 }, false);
-        }
-
-        batch->logits[batch->n_tokens - 1] = true;
-        llama_memory_clear(llama_get_memory(context), false);
-
-        const auto t_pp_start = ggml_time_us();
-        if (llama_decode(context, *batch) != 0) {
-            LOGi("llama_decode() failed during prompt processing");
-        }
-        const auto t_pp_end = ggml_time_us();
-
-        // bench text generation
-
-        LOGi("Benchmark text generation (tg)");
-
-        llama_memory_clear(llama_get_memory(context), false);
-        const auto t_tg_start = ggml_time_us();
-        for (i = 0; i < tg; i++) {
-
-            common_batch_clear(*batch);
-            for (j = 0; j < pl; j++) {
-                common_batch_add(*batch, 0, i, { j }, true);
-            }
-
-            LOGi("llama_decode() text generation: %d", i);
-            if (llama_decode(context, *batch) != 0) {
-                LOGi("llama_decode() failed during text generation");
-            }
-        }
-
-        const auto t_tg_end = ggml_time_us();
-
-        llama_memory_clear(llama_get_memory(context), false);
-
-        const auto t_pp = double(t_pp_end - t_pp_start) / 1000000.0;
-        const auto t_tg = double(t_tg_end - t_tg_start) / 1000000.0;
-
-        const auto speed_pp = double(pp) / t_pp;
-        const auto speed_tg = double(pl * tg) / t_tg;
-
-        pp_avg += speed_pp;
-        tg_avg += speed_tg;
-
-        pp_std += speed_pp * speed_pp;
-        tg_std += speed_tg * speed_tg;
-
-        LOGi("pp %f t/s, tg %f t/s", speed_pp, speed_tg);
-    }
-
-    pp_avg /= double(nr);
-    tg_avg /= double(nr);
-
-    if (nr > 1) {
-        pp_std = sqrt(pp_std / double(nr - 1) - pp_avg * pp_avg * double(nr) / double(nr - 1));
-        tg_std = sqrt(tg_std / double(nr - 1) - tg_avg * tg_avg * double(nr) / double(nr - 1));
-    } else {
-        pp_std = 0;
-        tg_std = 0;
-    }
-
-    char model_desc[128];
-    llama_model_desc(model, model_desc, sizeof(model_desc));
-
-    const auto model_size     = double(llama_model_size(model)) / 1024.0 / 1024.0 / 1024.0;
-    const auto model_n_params = double(llama_model_n_params(model)) / 1e9;
-
-    const auto backend    = "(Android)"; // TODO: What should this be?
-
-    std::stringstream result;
-    result << std::setprecision(2);
-    result << "| model | size | params | backend | test | t/s |\n";
-    result << "| --- | --- | --- | --- | --- | --- |\n";
-    result << "| " << model_desc << " | " << model_size << "GiB | " << model_n_params << "B | " << backend << " | pp " << pp << " | " << pp_avg << " ± " << pp_std << " |\n";
-    result << "| " << model_desc << " | " << model_size << "GiB | " << model_n_params << "B | " << backend << " | tg " << tg << " | " << tg_avg << " ± " << tg_std << " |\n";
-
-    return env->NewStringUTF(result.str().c_str());
-}
-
-extern "C"
-JNIEXPORT jlong JNICALL
-Java_android_llama_cpp_LLamaAndroid_new_1batch(JNIEnv *, jobject, jint n_tokens, jint embd, jint n_seq_max) {
-
-    // Source: Copy of llama.cpp:llama_batch_init but heap-allocated.
-
-    llama_batch *batch = new llama_batch {
-        0,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-    };
-
-    if (embd) {
-        batch->embd = (float *) malloc(sizeof(float) * n_tokens * embd);
-    } else {
-        batch->token = (llama_token *) malloc(sizeof(llama_token) * n_tokens);
-    }
-
-    batch->pos      = (llama_pos *)     malloc(sizeof(llama_pos)      * n_tokens);
-    batch->n_seq_id = (int32_t *)       malloc(sizeof(int32_t)        * n_tokens);
-    batch->seq_id   = (llama_seq_id **) malloc(sizeof(llama_seq_id *) * n_tokens);
-    for (int i = 0; i < n_tokens; ++i) {
-        batch->seq_id[i] = (llama_seq_id *) malloc(sizeof(llama_seq_id) * n_seq_max);
-    }
-    batch->logits   = (int8_t *)        malloc(sizeof(int8_t)         * n_tokens);
-
-    return reinterpret_cast<jlong>(batch);
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_free_1batch(JNIEnv *, jobject, jlong batch_pointer) {
-    //llama_batch_free(*reinterpret_cast<llama_batch *>(batch_pointer));
-    const auto batch = reinterpret_cast<llama_batch *>(batch_pointer);
-    delete batch;
-}
-
-extern "C"
-JNIEXPORT jlong JNICALL
-Java_android_llama_cpp_LLamaAndroid_new_1sampler(JNIEnv *, jobject) {
-    auto sparams = llama_sampler_chain_default_params();
-    sparams.no_perf = true;
-    llama_sampler * smpl = llama_sampler_chain_init(sparams);
-    llama_sampler_chain_add(smpl, llama_sampler_init_greedy());
-
-    return reinterpret_cast<jlong>(smpl);
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_free_1sampler(JNIEnv *, jobject, jlong sampler_pointer) {
-    llama_sampler_free(reinterpret_cast<llama_sampler *>(sampler_pointer));
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_backend_1init(JNIEnv *, jobject) {
-    llama_backend_init();
-}
-
-extern "C"
-JNIEXPORT jstring JNICALL
-Java_android_llama_cpp_LLamaAndroid_system_1info(JNIEnv *env, jobject) {
-    return env->NewStringUTF(llama_print_system_info());
-}
-
-extern "C"
-JNIEXPORT jint JNICALL
-Java_android_llama_cpp_LLamaAndroid_completion_1init(
-        JNIEnv *env,
-        jobject,
-        jlong context_pointer,
-        jlong batch_pointer,
-        jstring jtext,
-        jboolean format_chat,
-        jint n_len
-    ) {
-
-    cached_token_chars.clear();
-
-    const auto text = env->GetStringUTFChars(jtext, 0);
-    const auto context = reinterpret_cast<llama_context *>(context_pointer);
-    const auto batch = reinterpret_cast<llama_batch *>(batch_pointer);
-
-    bool parse_special = (format_chat == JNI_TRUE);
-    const auto tokens_list = common_tokenize(context, text, true, parse_special);
-
-    auto n_ctx = llama_n_ctx(context);
-    auto n_kv_req = tokens_list.size() + n_len;
-
-    LOGi("n_len = %d, n_ctx = %d, n_kv_req = %d", n_len, n_ctx, n_kv_req);
-
-    if (n_kv_req > n_ctx) {
-        LOGe("error: n_kv_req > n_ctx, the required KV cache size is not big enough");
-    }
-
-    for (auto id : tokens_list) {
-        LOGi("token: `%s`-> %d ", common_token_to_piece(context, id).c_str(), id);
-    }
-
-    common_batch_clear(*batch);
-
-    // evaluate the initial prompt
-    for (auto i = 0; i < tokens_list.size(); i++) {
-        common_batch_add(*batch, tokens_list[i], i, { 0 }, false);
-    }
-
-    // llama_decode will output logits only for the last token of the prompt
-    batch->logits[batch->n_tokens - 1] = true;
-
-    if (llama_decode(context, *batch) != 0) {
-        LOGe("llama_decode() failed");
-    }
-
-    env->ReleaseStringUTFChars(jtext, text);
-
-    return batch->n_tokens;
-}
-
-extern "C"
-JNIEXPORT jstring JNICALL
-Java_android_llama_cpp_LLamaAndroid_completion_1loop(
-        JNIEnv * env,
-        jobject,
-        jlong context_pointer,
-        jlong batch_pointer,
-        jlong sampler_pointer,
-        jint n_len,
-        jobject intvar_ncur
-) {
-    const auto context = reinterpret_cast<llama_context *>(context_pointer);
-    const auto batch   = reinterpret_cast<llama_batch   *>(batch_pointer);
-    const auto sampler = reinterpret_cast<llama_sampler *>(sampler_pointer);
-    const auto model = llama_get_model(context);
-    const auto vocab = llama_model_get_vocab(model);
-
-    if (!la_int_var) la_int_var = env->GetObjectClass(intvar_ncur);
-    if (!la_int_var_value) la_int_var_value = env->GetMethodID(la_int_var, "getValue", "()I");
-    if (!la_int_var_inc) la_int_var_inc = env->GetMethodID(la_int_var, "inc", "()V");
-
-    // sample the most likely token
-    const auto new_token_id = llama_sampler_sample(sampler, context, -1);
-
-    const auto n_cur = env->CallIntMethod(intvar_ncur, la_int_var_value);
-    if (llama_vocab_is_eog(vocab, new_token_id) || n_cur == n_len) {
-        return nullptr;
-    }
-
-    auto new_token_chars = common_token_to_piece(context, new_token_id);
-    cached_token_chars += new_token_chars;
-
-    jstring new_token = nullptr;
-    if (is_valid_utf8(cached_token_chars.c_str())) {
-        new_token = env->NewStringUTF(cached_token_chars.c_str());
-        LOGi("cached: %s, new_token_chars: `%s`, id: %d", cached_token_chars.c_str(), new_token_chars.c_str(), new_token_id);
-        cached_token_chars.clear();
-    } else {
-        new_token = env->NewStringUTF("");
-    }
-
-    common_batch_clear(*batch);
-    common_batch_add(*batch, new_token_id, n_cur, { 0 }, true);
-
-    env->CallVoidMethod(intvar_ncur, la_int_var_inc);
-
-    if (llama_decode(context, *batch) != 0) {
-        LOGe("llama_decode() returned null");
-    }
-
-    return new_token;
-}
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_android_llama_cpp_LLamaAndroid_kv_1cache_1clear(JNIEnv *, jobject, jlong context) {
-    llama_memory_clear(llama_get_memory(reinterpret_cast<llama_context *>(context)), true);
-}

examples/llama.android/llama/src/main/java/android/llama/cpp/LLamaAndroid.kt

@@ -1,180 +0,0 @@
-package android.llama.cpp
-
-import android.util.Log
-import kotlinx.coroutines.CoroutineDispatcher
-import kotlinx.coroutines.asCoroutineDispatcher
-import kotlinx.coroutines.flow.Flow
-import kotlinx.coroutines.flow.flow
-import kotlinx.coroutines.flow.flowOn
-import kotlinx.coroutines.withContext
-import java.util.concurrent.Executors
-import kotlin.concurrent.thread
-
-class LLamaAndroid {
-    private val tag: String? = this::class.simpleName
-
-    private val threadLocalState: ThreadLocal<State> = ThreadLocal.withInitial { State.Idle }
-
-    private val runLoop: CoroutineDispatcher = Executors.newSingleThreadExecutor {
-        thread(start = false, name = "Llm-RunLoop") {
-            Log.d(tag, "Dedicated thread for native code: ${Thread.currentThread().name}")
-
-            // No-op if called more than once.
-            System.loadLibrary("llama-android")
-
-            // Set llama log handler to Android
-            log_to_android()
-            backend_init(false)
-
-            Log.d(tag, system_info())
-
-            it.run()
-        }.apply {
-            uncaughtExceptionHandler = Thread.UncaughtExceptionHandler { _, exception: Throwable ->
-                Log.e(tag, "Unhandled exception", exception)
-            }
-        }
-    }.asCoroutineDispatcher()
-
-    private val nlen: Int = 64
-
-    private external fun log_to_android()
-    private external fun load_model(filename: String): Long
-    private external fun free_model(model: Long)
-    private external fun new_context(model: Long): Long
-    private external fun free_context(context: Long)
-    private external fun backend_init(numa: Boolean)
-    private external fun backend_free()
-    private external fun new_batch(nTokens: Int, embd: Int, nSeqMax: Int): Long
-    private external fun free_batch(batch: Long)
-    private external fun new_sampler(): Long
-    private external fun free_sampler(sampler: Long)
-    private external fun bench_model(
-        context: Long,
-        model: Long,
-        batch: Long,
-        pp: Int,
-        tg: Int,
-        pl: Int,
-        nr: Int
-    ): String
-
-    private external fun system_info(): String
-
-    private external fun completion_init(
-        context: Long,
-        batch: Long,
-        text: String,
-        formatChat: Boolean,
-        nLen: Int
-    ): Int
-
-    private external fun completion_loop(
-        context: Long,
-        batch: Long,
-        sampler: Long,
-        nLen: Int,
-        ncur: IntVar
-    ): String?
-
-    private external fun kv_cache_clear(context: Long)
-
-    suspend fun bench(pp: Int, tg: Int, pl: Int, nr: Int = 1): String {
-        return withContext(runLoop) {
-            when (val state = threadLocalState.get()) {
-                is State.Loaded -> {
-                    Log.d(tag, "bench(): $state")
-                    bench_model(state.context, state.model, state.batch, pp, tg, pl, nr)
-                }
-
-                else -> throw IllegalStateException("No model loaded")
-            }
-        }
-    }
-
-    suspend fun load(pathToModel: String) {
-        withContext(runLoop) {
-            when (threadLocalState.get()) {
-                is State.Idle -> {
-                    val model = load_model(pathToModel)
-                    if (model == 0L)  throw IllegalStateException("load_model() failed")
-
-                    val context = new_context(model)
-                    if (context == 0L) throw IllegalStateException("new_context() failed")
-
-                    val batch = new_batch(512, 0, 1)
-                    if (batch == 0L) throw IllegalStateException("new_batch() failed")
-
-                    val sampler = new_sampler()
-                    if (sampler == 0L) throw IllegalStateException("new_sampler() failed")
-
-                    Log.i(tag, "Loaded model $pathToModel")
-                    threadLocalState.set(State.Loaded(model, context, batch, sampler))
-                }
-                else -> throw IllegalStateException("Model already loaded")
-            }
-        }
-    }
-
-    fun send(message: String, formatChat: Boolean = false): Flow<String> = flow {
-        when (val state = threadLocalState.get()) {
-            is State.Loaded -> {
-                val ncur = IntVar(completion_init(state.context, state.batch, message, formatChat, nlen))
-                while (ncur.value <= nlen) {
-                    val str = completion_loop(state.context, state.batch, state.sampler, nlen, ncur)
-                    if (str == null) {
-                        break
-                    }
-                    emit(str)
-                }
-                kv_cache_clear(state.context)
-            }
-            else -> {}
-        }
-    }.flowOn(runLoop)
-
-    /**
-     * Unloads the model and frees resources.
-     *
-     * This is a no-op if there's no model loaded.
-     */
-    suspend fun unload() {
-        withContext(runLoop) {
-            when (val state = threadLocalState.get()) {
-                is State.Loaded -> {
-                    free_context(state.context)
-                    free_model(state.model)
-                    free_batch(state.batch)
-                    free_sampler(state.sampler);
-
-                    threadLocalState.set(State.Idle)
-                }
-                else -> {}
-            }
-        }
-    }
-
-    companion object {
-        private class IntVar(value: Int) {
-            @Volatile
-            var value: Int = value
-                private set
-
-            fun inc() {
-                synchronized(this) {
-                    value += 1
-                }
-            }
-        }
-
-        private sealed interface State {
-            data object Idle: State
-            data class Loaded(val model: Long, val context: Long, val batch: Long, val sampler: Long): State
-        }
-
-        // Enforce only one instance of Llm.
-        private val _instance: LLamaAndroid = LLamaAndroid()
-
-        fun instance(): LLamaAndroid = _instance
-    }
-}

examples/llama.android/settings.gradle.kts

@@ -8,11 +8,11 @@ pluginManagement {
 dependencyResolutionManagement {
     repositoriesMode.set(RepositoriesMode.FAIL_ON_PROJECT_REPOS)
     repositories {
-        google()
         mavenCentral()
+        google()
     }
 }
 
-rootProject.name = "LlamaAndroid"
+rootProject.name = "AiChat"
 include(":app")
-include(":llama")
+include(":lib")

examples/model-conversion/README.md

@@ -10,6 +10,13 @@ and in some cases perplexity checked of the quantized model. And finally the
 model/models need to the ggml-org on Hugging Face. This tool/example tries to
 help with this process.
 
+> 📝 **Note:** When adding a new model from an existing family, verify the
+> previous version passes logits verification first. Existing models can have
+> subtle numerical differences that don't affect generation quality but cause
+> logits mismatches. Identifying these upfront whether they exist in llama.cpp,
+> the conversion script, or in an upstream implementation, can save significant
+> debugging time.
+
 ### Overview
 The idea is that the makefile targets and scripts here can be used in the
 development/conversion process assisting with things like:

examples/model-conversion/scripts/causal/modelcard.template

@@ -7,7 +7,7 @@ base_model:
 Recommended way to run this model:
 
 ```sh
-llama-server -hf {namespace}/{model_name}-GGUF -c 0 -fa
+llama-server -hf {namespace}/{model_name}-GGUF -c 0
 ```
 
 Then, access http://localhost:8080

examples/model-conversion/scripts/causal/run-org-model.py

@@ -5,7 +5,7 @@ import os
 import importlib
 from pathlib import Path
 
-from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig
+from transformers import AutoTokenizer, AutoModelForCausalLM, AutoModelForImageTextToText, AutoConfig
 import torch
 import numpy as np
 
@@ -116,11 +116,11 @@ def debug_hook(name):
     def fn(_m, input, output):
         if isinstance(input, torch.Tensor):
             summarize(input, name + "_in")
-        elif isinstance(input, (tuple, list)) and isinstance(input[0], torch.Tensor):
+        elif isinstance(input, (tuple, list)) and len(input) > 0 and isinstance(input[0], torch.Tensor):
             summarize(input[0], name + "_in")
         if isinstance(output, torch.Tensor):
             summarize(output, name + "_out")
-        elif isinstance(output, (tuple, list)) and isinstance(output[0], torch.Tensor):
+        elif isinstance(output, (tuple, list)) and len(output) > 0 and isinstance(output[0], torch.Tensor):
             summarize(output[0], name + "_out")
 
     return fn
@@ -130,6 +130,7 @@ unreleased_model_name = os.getenv("UNRELEASED_MODEL_NAME")
 
 parser = argparse.ArgumentParser(description="Process model with specified path")
 parser.add_argument("--model-path", "-m", help="Path to the model")
+parser.add_argument("--prompt-file", "-f", help="Optional prompt file", required=False)
 args = parser.parse_args()
 
 model_path = os.environ.get("MODEL_PATH", args.model_path)
@@ -142,8 +143,13 @@ if model_path is None:
 print("Loading model and tokenizer using AutoTokenizer:", model_path)
 tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
 config = AutoConfig.from_pretrained(model_path, trust_remote_code=True)
+multimodal = False
+full_config = config
 
 print("Model type:       ", config.model_type)
+if "vocab_size" not in config and "text_config" in config:
+    config = config.text_config
+    multimodal = True
 print("Vocab size:       ", config.vocab_size)
 print("Hidden size:      ", config.hidden_size)
 print("Number of layers: ", config.num_hidden_layers)
@@ -168,6 +174,11 @@ if unreleased_model_name:
     except (ImportError, AttributeError) as e:
         print(f"Failed to import or load model: {e}")
         exit(1)
+else:
+    if multimodal:
+        model = AutoModelForImageTextToText.from_pretrained(
+            model_path, device_map="auto", offload_folder="offload", trust_remote_code=True, config=full_config
+        )
     else:
         model = AutoModelForCausalLM.from_pretrained(
             model_path, device_map="auto", offload_folder="offload", trust_remote_code=True, config=config
@@ -185,7 +196,10 @@ model_name = os.path.basename(model_path)
 print(f"Model class: {model.__class__.__name__}")
 
 device = next(model.parameters()).device
-if os.getenv("MODEL_TESTING_PROMPT"):
+if args.prompt_file:
+    with open(args.prompt_file, encoding='utf-8') as f:
+        prompt = f.read()
+elif os.getenv("MODEL_TESTING_PROMPT"):
     prompt = os.getenv("MODEL_TESTING_PROMPT")
 else:
     prompt = "Hello, my name is"
@@ -195,9 +209,18 @@ print(f"Input tokens: {input_ids}")
 print(f"Input text: {repr(prompt)}")
 print(f"Tokenized: {tokenizer.convert_ids_to_tokens(input_ids[0])}")
 
+batch_size = 512
+
 with torch.no_grad():
-    outputs = model(input_ids.to(model.device))
-    logits = outputs.logits
+    past = None
+    outputs = None
+    for i in range(0, input_ids.size(1), batch_size):
+        print(f"Processing chunk with tokens {i} to {i + batch_size}")
+        chunk = input_ids[:, i:i + batch_size]
+        outputs = model(chunk.to(model.device), past_key_values=past, use_cache=True)
+        past = outputs.past_key_values
+
+    logits = outputs.logits # type: ignore
 
     # Extract logits for the last token (next token prediction)
     last_logits = logits[0, -1, :].float().cpu().numpy()

examples/model-conversion/scripts/embedding/compare-embeddings-logits.sh

@@ -34,8 +34,11 @@ done
 MODEL_PATH="${MODEL_PATH:-"$EMBEDDING_MODEL_PATH"}"
 MODEL_NAME="${MODEL_NAME:-$(basename "$MODEL_PATH")}"
 
+CONVERTED_MODEL_PATH="${CONVERTED_EMBEDDING_PATH:-"$CONVERTED_EMBEDDING_MODEL"}"
+CONVERTED_MODEL_NAME="${CONVERTED_MODEL_NAME:-$(basename "$CONVERTED_MODEL_PATH" .gguf)}"
+
 if [ -t 0 ]; then
-    CPP_EMBEDDINGS="data/llamacpp-${MODEL_NAME}-embeddings.bin"
+    CPP_EMBEDDINGS="data/llamacpp-${CONVERTED_MODEL_NAME}-embeddings.bin"
 else
     # Process piped JSON data and convert to binary (matching logits.cpp format)
     TEMP_FILE=$(mktemp /tmp/tmp.XXXXXX.binn)

examples/speculative/speculative.cpp

@@ -242,7 +242,7 @@ int main(int argc, char ** argv) {
                 bool accept = false;
                 if (params.sampling.temp > 0) {
                     // stochastic verification
-                    common_sampler_sample(smpl, ctx_tgt, drafts[s_keep].i_batch_tgt[i_dft]);
+                    common_sampler_sample(smpl, ctx_tgt, drafts[s_keep].i_batch_tgt[i_dft], true);
 
                     auto & dist_tgt = *common_sampler_get_candidates(smpl, true);
 
@@ -491,7 +491,7 @@ int main(int argc, char ** argv) {
                     continue;
                 }
 
-                common_sampler_sample(drafts[s].smpl, ctx_dft, drafts[s].i_batch_dft);
+                common_sampler_sample(drafts[s].smpl, ctx_dft, drafts[s].i_batch_dft, true);
 
                 const auto * cur_p = common_sampler_get_candidates(drafts[s].smpl, true);
 

ggml/src/CMakeLists.txt

@@ -386,6 +386,9 @@ if (GGML_CPU_ALL_VARIANTS)
             ggml_add_cpu_backend_variant(android_armv8.2_1    DOTPROD)
             ggml_add_cpu_backend_variant(android_armv8.2_2    DOTPROD FP16_VECTOR_ARITHMETIC)
             ggml_add_cpu_backend_variant(android_armv8.6_1    DOTPROD FP16_VECTOR_ARITHMETIC MATMUL_INT8)
+            ggml_add_cpu_backend_variant(android_armv9.0_1    DOTPROD MATMUL_INT8 FP16_VECTOR_ARITHMETIC SVE2)
+            ggml_add_cpu_backend_variant(android_armv9.2_1    DOTPROD MATMUL_INT8 FP16_VECTOR_ARITHMETIC SME)
+            ggml_add_cpu_backend_variant(android_armv9.2_2    DOTPROD MATMUL_INT8 FP16_VECTOR_ARITHMETIC SVE SME)
         elseif (APPLE)
             ggml_add_cpu_backend_variant(apple_m1             DOTPROD)
             ggml_add_cpu_backend_variant(apple_m2_m3          DOTPROD MATMUL_INT8)

ggml/src/ggml-cpu/arch-fallback.h

@@ -43,6 +43,8 @@
 #define ggml_gemv_q2_K_8x8_q8_K_generic ggml_gemv_q2_K_8x8_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
 #define ggml_gemv_iq4_nl_8x8_q8_0_generic ggml_gemv_iq4_nl_8x8_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_0_8x8_q8_0_generic ggml_gemm_q4_0_8x8_q8_0
@@ -51,6 +53,8 @@
 #define ggml_gemm_q2_K_8x8_q8_K_generic ggml_gemm_q2_K_8x8_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
 #define ggml_gemm_iq4_nl_8x8_q8_0_generic ggml_gemm_iq4_nl_8x8_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #elif defined(__aarch64__) || defined(__arm__) || defined(_M_ARM) || defined(_M_ARM64)
 // repack.cpp
 #define ggml_quantize_mat_q8_K_4x4_generic ggml_quantize_mat_q8_K_4x4
@@ -67,10 +71,14 @@
 #define ggml_gemv_q4_0_4x8_q8_0_generic ggml_gemv_q4_0_4x8_q8_0
 #define ggml_gemv_q4_K_8x4_q8_K_generic ggml_gemv_q4_K_8x4_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_K_8x4_q8_K_generic ggml_gemm_q4_K_8x4_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #elif defined(__POWERPC__) || defined(__powerpc__)
 // ref: https://github.com/ggml-org/llama.cpp/pull/14146#issuecomment-2972561679
 // quants.c
@@ -91,6 +99,8 @@
 #define ggml_gemv_q2_K_8x8_q8_K_generic ggml_gemv_q2_K_8x8_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
 #define ggml_gemv_iq4_nl_8x8_q8_0_generic ggml_gemv_iq4_nl_8x8_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_0_8x8_q8_0_generic ggml_gemm_q4_0_8x8_q8_0
@@ -99,6 +109,8 @@
 #define ggml_gemm_q2_K_8x8_q8_K_generic ggml_gemm_q2_K_8x8_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
 #define ggml_gemm_iq4_nl_8x8_q8_0_generic ggml_gemm_iq4_nl_8x8_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #elif defined(__loongarch64)
 // quants.c
 #define quantize_row_q8_K_generic quantize_row_q8_K
@@ -119,6 +131,8 @@
 #define ggml_gemv_q2_K_8x8_q8_K_generic ggml_gemv_q2_K_8x8_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
 #define ggml_gemv_iq4_nl_8x8_q8_0_generic ggml_gemv_iq4_nl_8x8_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_0_8x8_q8_0_generic ggml_gemm_q4_0_8x8_q8_0
@@ -127,6 +141,8 @@
 #define ggml_gemm_q2_K_8x8_q8_K_generic ggml_gemm_q2_K_8x8_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
 #define ggml_gemm_iq4_nl_8x8_q8_0_generic ggml_gemm_iq4_nl_8x8_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #elif defined(__riscv)
 // quants.c
 #define quantize_row_q8_K_generic quantize_row_q8_K
@@ -154,6 +170,8 @@
 #define ggml_gemv_q2_K_8x8_q8_K_generic ggml_gemv_q2_K_8x8_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
 #define ggml_gemv_iq4_nl_8x8_q8_0_generic ggml_gemv_iq4_nl_8x8_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_K_8x4_q8_K_generic ggml_gemm_q4_K_8x4_q8_K
@@ -161,6 +179,8 @@
 #define ggml_gemm_q2_K_8x8_q8_K_generic ggml_gemm_q2_K_8x8_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
 #define ggml_gemm_iq4_nl_8x8_q8_0_generic ggml_gemm_iq4_nl_8x8_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #elif defined(__s390x__)
 // quants.c
 #define quantize_row_q8_K_generic quantize_row_q8_K
@@ -187,6 +207,8 @@
 #define ggml_gemv_q2_K_8x8_q8_K_generic ggml_gemv_q2_K_8x8_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
 #define ggml_gemv_iq4_nl_8x8_q8_0_generic ggml_gemv_iq4_nl_8x8_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_0_8x8_q8_0_generic ggml_gemm_q4_0_8x8_q8_0
@@ -195,6 +217,8 @@
 #define ggml_gemm_q2_K_8x8_q8_K_generic ggml_gemm_q2_K_8x8_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
 #define ggml_gemm_iq4_nl_8x8_q8_0_generic ggml_gemm_iq4_nl_8x8_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #elif defined(__wasm__)
 // quants.c
 #define ggml_vec_dot_q4_1_q8_1_generic ggml_vec_dot_q4_1_q8_1
@@ -223,6 +247,8 @@
 #define ggml_gemv_q2_K_8x8_q8_K_generic ggml_gemv_q2_K_8x8_q8_K
 #define ggml_gemv_iq4_nl_4x4_q8_0_generic ggml_gemv_iq4_nl_4x4_q8_0
 #define ggml_gemv_iq4_nl_8x8_q8_0_generic ggml_gemv_iq4_nl_8x8_q8_0
+#define ggml_gemv_q8_0_4x4_q8_0_generic ggml_gemv_q8_0_4x4_q8_0
+#define ggml_gemv_q8_0_4x8_q8_0_generic ggml_gemv_q8_0_4x8_q8_0
 #define ggml_gemm_q4_0_4x4_q8_0_generic ggml_gemm_q4_0_4x4_q8_0
 #define ggml_gemm_q4_0_4x8_q8_0_generic ggml_gemm_q4_0_4x8_q8_0
 #define ggml_gemm_q4_0_8x8_q8_0_generic ggml_gemm_q4_0_8x8_q8_0
@@ -231,4 +257,6 @@
 #define ggml_gemm_q2_K_8x8_q8_K_generic ggml_gemm_q2_K_8x8_q8_K
 #define ggml_gemm_iq4_nl_4x4_q8_0_generic ggml_gemm_iq4_nl_4x4_q8_0
 #define ggml_gemm_iq4_nl_8x8_q8_0_generic ggml_gemm_iq4_nl_8x8_q8_0
+#define ggml_gemm_q8_0_4x4_q8_0_generic ggml_gemm_q8_0_4x4_q8_0
+#define ggml_gemm_q8_0_4x8_q8_0_generic ggml_gemm_q8_0_4x8_q8_0
 #endif

ggml/src/ggml-cpu/arch/arm/repack.cpp

@@ -786,6 +786,133 @@ void ggml_gemv_q4_K_8x8_q8_K(int                        n,
     ggml_gemv_q4_K_8x8_q8_K_generic(n, s, bs, vx, vy, nr, nc);
 }
 
+void ggml_gemv_q8_0_4x4_q8_0(int                        n,
+                             float * GGML_RESTRICT      s,
+                             size_t                     bs,
+                             const void * GGML_RESTRICT vx,
+                             const void * GGML_RESTRICT vy,
+                             int                        nr,
+                             int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 4;
+
+    assert(n % qk == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    UNUSED(nb);
+    UNUSED(ncols_interleaved);
+    UNUSED(blocklen);
+
+#if defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
+    const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx;
+
+    for (int c = 0; c < nc; c += ncols_interleaved) {
+        const block_q8_0 * a_ptr = (const block_q8_0 *) vy;
+        float32x4_t        acc   = vdupq_n_f32(0);
+        for (int b = 0; b < nb; b++) {
+            int8x16x4_t b_low  = vld1q_s8_x4((const int8_t *) b_ptr->qs);
+            int8x16x4_t b_high = vld1q_s8_x4((const int8_t *) b_ptr->qs + 64);
+            float16x4_t bd     = vld1_f16((const __fp16 *) b_ptr->d);
+
+            int8x16x2_t a  = vld1q_s8_x2(a_ptr->qs);
+            float16x4_t ad = vld1_dup_f16((const __fp16 *) &a_ptr->d);
+
+            int32x4_t ret = vdupq_n_s32(0);
+
+            ret = vdotq_laneq_s32(ret, b_low.val[0], a.val[0], 0);
+            ret = vdotq_laneq_s32(ret, b_low.val[1], a.val[0], 1);
+            ret = vdotq_laneq_s32(ret, b_low.val[2], a.val[0], 2);
+            ret = vdotq_laneq_s32(ret, b_low.val[3], a.val[0], 3);
+
+            ret = vdotq_laneq_s32(ret, b_high.val[0], a.val[1], 0);
+            ret = vdotq_laneq_s32(ret, b_high.val[1], a.val[1], 1);
+            ret = vdotq_laneq_s32(ret, b_high.val[2], a.val[1], 2);
+            ret = vdotq_laneq_s32(ret, b_high.val[3], a.val[1], 3);
+
+            acc = vfmaq_f32(acc, vcvtq_f32_s32(ret), vmulq_f32(vcvt_f32_f16(ad), vcvt_f32_f16(bd)));
+            a_ptr++;
+            b_ptr++;
+        }
+        vst1q_f32(s, acc);
+        s += ncols_interleaved;
+    }
+    return;
+
+#endif  // defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
+    ggml_gemv_q8_0_4x4_q8_0_generic(n, s, bs, vx, vy, nr, nc);
+}
+
+void ggml_gemv_q8_0_4x8_q8_0(int                        n,
+                             float * GGML_RESTRICT      s,
+                             size_t                     bs,
+                             const void * GGML_RESTRICT vx,
+                             const void * GGML_RESTRICT vy,
+                             int                        nr,
+                             int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 8;
+
+    assert(n % qk == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    UNUSED(nb);
+    UNUSED(ncols_interleaved);
+    UNUSED(blocklen);
+
+#if defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
+    const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx;
+
+    for (int c = 0; c < nc; c += ncols_interleaved) {
+        const block_q8_0 * a_ptr = (const block_q8_0 *) vy;
+        float32x4_t        acc   = vdupq_n_f32(0);
+
+        for (int b = 0; b < nb; b++) {
+            int8x16x4_t b_low  = vld1q_s8_x4((const int8_t *) b_ptr->qs);
+            int8x16x4_t b_high = vld1q_s8_x4((const int8_t *) b_ptr->qs + 64);
+            float16x4_t bd     = vld1_f16((const __fp16 *) b_ptr->d);
+
+            int8x8x4_t  a_chunks = vld1_s8_x4(a_ptr->qs);
+            int8x16_t   a0       = vcombine_s8(a_chunks.val[0], a_chunks.val[0]);
+            int8x16_t   a1       = vcombine_s8(a_chunks.val[1], a_chunks.val[1]);
+            int8x16_t   a2       = vcombine_s8(a_chunks.val[2], a_chunks.val[2]);
+            int8x16_t   a3       = vcombine_s8(a_chunks.val[3], a_chunks.val[3]);
+            float16x4_t ad       = vld1_dup_f16((const __fp16 *) &a_ptr->d);
+
+            int32x4_t ret0 = vdupq_n_s32(0);
+            int32x4_t ret1 = vdupq_n_s32(0);
+
+            // 0..7
+            ret0 = vdotq_s32(ret0, b_low.val[0], a0);
+            ret1 = vdotq_s32(ret1, b_low.val[1], a0);
+            // 8..15
+            ret0 = vdotq_s32(ret0, b_low.val[2], a1);
+            ret1 = vdotq_s32(ret1, b_low.val[3], a1);
+            // 16..23
+            ret0 = vdotq_s32(ret0, b_high.val[0], a2);
+            ret1 = vdotq_s32(ret1, b_high.val[1], a2);
+            // 24..31
+            ret0 = vdotq_s32(ret0, b_high.val[2], a3);
+            ret1 = vdotq_s32(ret1, b_high.val[3], a3);
+
+            int32x4_t ret = vpaddq_s32(ret0, ret1);
+
+            acc = vfmaq_f32(acc, vcvtq_f32_s32(ret), vmulq_f32(vcvt_f32_f16(ad), vcvt_f32_f16(bd)));
+            a_ptr++;
+            b_ptr++;
+        }
+        vst1q_f32(s, acc);
+        s += ncols_interleaved;
+    }
+    return;
+
+#endif  // defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
+    ggml_gemv_q8_0_4x8_q8_0_generic(n, s, bs, vx, vy, nr, nc);
+}
+
 void ggml_gemm_q4_0_4x4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
     const int qk = QK8_0;
     const int nb = n / qk;
@@ -2610,3 +2737,159 @@ void ggml_gemm_q4_K_8x8_q8_K(int                        n,
 #endif  // defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_MATMUL_INT8)
     ggml_gemm_q4_K_8x8_q8_K_generic(n, s, bs, vx, vy, nr, nc);
 }
+
+
+void ggml_gemm_q8_0_4x4_q8_0(int                        n,
+                             float * GGML_RESTRICT      s,
+                             size_t                     bs,
+                             const void * GGML_RESTRICT vx,
+                             const void * GGML_RESTRICT vy,
+                             int                        nr,
+                             int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 4;
+
+    assert(n % qk == 0);
+    assert(nr % 4 == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    UNUSED(nb);
+    UNUSED(ncols_interleaved);
+    UNUSED(blocklen);
+
+#if defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
+    for (int y = 0; y < nr / 4; y++) {
+        const block_q8_0x4 * a_ptr = (const block_q8_0x4 *) vy + (y * nb);
+        for (int x = 0; x < nc / ncols_interleaved; x++) {
+            const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx + (x * nb);
+
+            float32x4_t sumf[4];
+            for (int m = 0; m < 4; m++) {
+                sumf[m] = vdupq_n_f32(0);
+            }
+
+            for (int l = 0; l < nb; l++) {
+                float32x4_t a_d = vcvt_f32_f16(vld1_f16((const float16_t *) a_ptr[l].d));
+                float32x4_t b_d = vcvt_f32_f16(vld1_f16((const float16_t *) b_ptr[l].d));
+
+                int32x4_t sumi_0 = vdupq_n_s32(0);
+                int32x4_t sumi_1 = vdupq_n_s32(0);
+                int32x4_t sumi_2 = vdupq_n_s32(0);
+                int32x4_t sumi_3 = vdupq_n_s32(0);
+
+                for (int k_group = 0; k_group < 8; k_group += 4) {
+                    int8x16x4_t a = vld1q_s8_x4(a_ptr[l].qs + 16 * k_group);
+                    int8x16x4_t b = vld1q_s8_x4(b_ptr[l].qs + 16 * k_group);
+
+                    for (int k = 0; k < 4; k++) {
+                        sumi_0 = vdotq_laneq_s32(sumi_0, b.val[k], a.val[k], 0);
+                        sumi_1 = vdotq_laneq_s32(sumi_1, b.val[k], a.val[k], 1);
+                        sumi_2 = vdotq_laneq_s32(sumi_2, b.val[k], a.val[k], 2);
+                        sumi_3 = vdotq_laneq_s32(sumi_3, b.val[k], a.val[k], 3);
+                    }
+                }
+
+                sumf[0] = vmlaq_f32(sumf[0], vmulq_laneq_f32(b_d, a_d, 0), vcvtq_f32_s32(sumi_0));
+                sumf[1] = vmlaq_f32(sumf[1], vmulq_laneq_f32(b_d, a_d, 1), vcvtq_f32_s32(sumi_1));
+                sumf[2] = vmlaq_f32(sumf[2], vmulq_laneq_f32(b_d, a_d, 2), vcvtq_f32_s32(sumi_2));
+                sumf[3] = vmlaq_f32(sumf[3], vmulq_laneq_f32(b_d, a_d, 3), vcvtq_f32_s32(sumi_3));
+            }
+
+            for (int m = 0; m < 4; m++) {
+                vst1q_f32(s + (y * 4 + m) * bs + x * 4, sumf[m]);
+            }
+        }
+    }
+    return;
+#endif  // defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_DOTPROD)
+    ggml_gemm_q8_0_4x4_q8_0_generic(n, s, bs, vx, vy, nr, nc);
+}
+
+void ggml_gemm_q8_0_4x8_q8_0(int                        n,
+                             float * GGML_RESTRICT      s,
+                             size_t                     bs,
+                             const void * GGML_RESTRICT vx,
+                             const void * GGML_RESTRICT vy,
+                             int                        nr,
+                             int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 8;
+
+    assert(n % qk == 0);
+    assert(nr % 4 == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    UNUSED(nb);
+    UNUSED(ncols_interleaved);
+    UNUSED(blocklen);
+
+#if defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_MATMUL_INT8)
+    const block_q8_0x4 * b_ptr_base = (const block_q8_0x4 *) vx;
+
+    for (int y = 0; y < nr; y += 4) {
+        const block_q8_0x4 * a_ptr_base = (const block_q8_0x4 *) vy + (y / 4) * nb;
+
+        for (int x = 0; x < nc; x += ncols_interleaved) {
+            const block_q8_0x4 * b_ptr = b_ptr_base + (x / 4) * nb;
+            const block_q8_0x4 * a_ptr = a_ptr_base;
+
+            float32x4_t acc_f32[4];
+            for (int i = 0; i < 4; i++) {
+                acc_f32[i] = vdupq_n_f32(0);
+            }
+
+            for (int b = 0; b < nb; b++) {
+                int32x4_t acc[4];
+                for (int i = 0; i < 4; i++) {
+                    acc[i] = vdupq_n_s32(0);
+                }
+
+                // Process 4 chunks of 8 positions each
+                for (int chunk = 0; chunk < 4; chunk++) {
+                    int8x16_t a01 = vld1q_s8(a_ptr->qs + chunk * 32);
+                    int8x16_t a23 = vld1q_s8(a_ptr->qs + chunk * 32 + 16);
+                    int8x16_t b01 = vld1q_s8(b_ptr->qs + chunk * 32);
+                    int8x16_t b23 = vld1q_s8(b_ptr->qs + chunk * 32 + 16);
+
+                    acc[0] = vmmlaq_s32(acc[0], a01, b01);
+                    acc[1] = vmmlaq_s32(acc[1], a01, b23);
+                    acc[2] = vmmlaq_s32(acc[2], a23, b01);
+                    acc[3] = vmmlaq_s32(acc[3], a23, b23);
+                }
+
+                // Reorder outputs from 2×2 tiles to row-major
+                // acc[0] = [r0c0, r0c1, r1c0, r1c1]
+                // acc[1] = [r0c2, r0c3, r1c2, r1c3]
+                // acc[2] = [r2c0, r2c1, r3c0, r3c1]
+                // acc[3] = [r2c2, r2c3, r3c2, r3c3]
+                int32x4_t row0 = vcombine_s32(vget_low_s32(acc[0]), vget_low_s32(acc[1]));
+                int32x4_t row1 = vcombine_s32(vget_high_s32(acc[0]), vget_high_s32(acc[1]));
+                int32x4_t row2 = vcombine_s32(vget_low_s32(acc[2]), vget_low_s32(acc[3]));
+                int32x4_t row3 = vcombine_s32(vget_high_s32(acc[2]), vget_high_s32(acc[3]));
+
+                // Scales
+                float32x4_t a_d = vcvt_f32_f16(vld1_f16((const __fp16 *) a_ptr->d));
+                float32x4_t b_d = vcvt_f32_f16(vld1_f16((const __fp16 *) b_ptr->d));
+
+                acc_f32[0] = vfmaq_f32(acc_f32[0], vcvtq_f32_s32(row0), vmulq_laneq_f32(b_d, a_d, 0));
+                acc_f32[1] = vfmaq_f32(acc_f32[1], vcvtq_f32_s32(row1), vmulq_laneq_f32(b_d, a_d, 1));
+                acc_f32[2] = vfmaq_f32(acc_f32[2], vcvtq_f32_s32(row2), vmulq_laneq_f32(b_d, a_d, 2));
+                acc_f32[3] = vfmaq_f32(acc_f32[3], vcvtq_f32_s32(row3), vmulq_laneq_f32(b_d, a_d, 3));
+
+                a_ptr++;
+                b_ptr++;
+            }
+
+            for (int row = 0; row < 4; row++) {
+                vst1q_f32(s + (y + row) * bs + x, acc_f32[row]);
+            }
+        }
+    }
+    return;
+#endif  // defined(__aarch64__) && defined(__ARM_NEON) && defined(__ARM_FEATURE_MATMUL_INT8)
+    ggml_gemm_q8_0_4x8_q8_0_generic(n, s, bs, vx, vy, nr, nc);
+}

ggml/src/ggml-cpu/repack.cpp

@@ -692,6 +692,100 @@ void ggml_gemv_iq4_nl_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs
     }
 }
 
+void ggml_gemv_q8_0_4x4_q8_0_generic(int                        n,
+                                     float * GGML_RESTRICT      s,
+                                     size_t                     bs,
+                                     const void * GGML_RESTRICT vx,
+                                     const void * GGML_RESTRICT vy,
+                                     int                        nr,
+                                     int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 4;
+
+    assert(nr == 1);
+    assert(n % qk == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    UNUSED(bs);
+    UNUSED(nr);
+
+    float sumf[4];
+    int   sumi;
+
+    const block_q8_0 * a_ptr = (const block_q8_0 *) vy;
+    for (int x = 0; x < nc / ncols_interleaved; x++) {
+        const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx + (x * nb);
+
+        for (int j = 0; j < ncols_interleaved; j++) {
+            sumf[j] = 0.0;
+        }
+        for (int l = 0; l < nb; l++) {
+            for (int k = 0; k < (qk / blocklen); k++) {
+                for (int j = 0; j < ncols_interleaved; j++) {
+                    sumi = 0;
+                    for (int i = 0; i < blocklen; ++i) {
+                        const int v0 = b_ptr[l].qs[k * ncols_interleaved * blocklen + j * blocklen + i];
+                        sumi += v0 * a_ptr[l].qs[k * blocklen + i];
+                    }
+                    sumf[j] += sumi * GGML_CPU_FP16_TO_FP32(b_ptr[l].d[j]) * GGML_CPU_FP16_TO_FP32(a_ptr[l].d);
+                }
+            }
+        }
+        for (int j = 0; j < ncols_interleaved; j++) {
+            s[x * ncols_interleaved + j] = sumf[j];
+        }
+    }
+}
+
+void ggml_gemv_q8_0_4x8_q8_0_generic(int                        n,
+                                     float * GGML_RESTRICT      s,
+                                     size_t                     bs,
+                                     const void * GGML_RESTRICT vx,
+                                     const void * GGML_RESTRICT vy,
+                                     int                        nr,
+                                     int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 8;
+
+    assert(nr == 1);
+    assert(n % qk == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    UNUSED(bs);
+    UNUSED(nr);
+
+    float sumf[4];
+    int   sumi;
+
+    const block_q8_0 * a_ptr = (const block_q8_0 *) vy;
+    for (int x = 0; x < nc / ncols_interleaved; x++) {
+        const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx + (x * nb);
+
+        for (int j = 0; j < ncols_interleaved; j++) {
+            sumf[j] = 0.0;
+        }
+        for (int l = 0; l < nb; l++) {
+            for (int k = 0; k < (qk / blocklen); k++) {
+                for (int j = 0; j < ncols_interleaved; j++) {
+                    sumi = 0;
+                    for (int i = 0; i < blocklen; ++i) {
+                        const int v0 = b_ptr[l].qs[k * ncols_interleaved * blocklen + j * blocklen + i];
+                        sumi += v0 * a_ptr[l].qs[k * blocklen + i];
+                    }
+                    sumf[j] += sumi * GGML_CPU_FP16_TO_FP32(b_ptr[l].d[j]) * GGML_CPU_FP16_TO_FP32(a_ptr[l].d);
+                }
+            }
+        }
+        for (int j = 0; j < ncols_interleaved; j++) {
+            s[x * ncols_interleaved + j] = sumf[j];
+        }
+    }
+}
+
 void ggml_gemm_q4_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc) {
     const int qk = QK8_0;
     const int nb = n / qk;
@@ -1219,8 +1313,129 @@ void ggml_gemm_iq4_nl_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs
     }
 }
 
+void ggml_gemm_q8_0_4x4_q8_0_generic(int                        n,
+                                     float * GGML_RESTRICT      s,
+                                     size_t                     bs,
+                                     const void * GGML_RESTRICT vx,
+                                     const void * GGML_RESTRICT vy,
+                                     int                        nr,
+                                     int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 4;
+
+    assert(n % qk == 0);
+    assert(nr % 4 == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    float sumf[4][4];
+    int   sumi;
+
+    for (int y = 0; y < nr / 4; y++) {
+        const block_q8_0x4 * a_ptr = (const block_q8_0x4 *) vy + (y * nb);
+        for (int x = 0; x < nc / ncols_interleaved; x++) {
+            const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx + (x * nb);
+            for (int m = 0; m < 4; m++) {
+                for (int j = 0; j < ncols_interleaved; j++) {
+                    sumf[m][j] = 0.0;
+                }
+            }
+            for (int l = 0; l < nb; l++) {
+                for (int k = 0; k < (qk / blocklen); k++) {
+                    for (int m = 0; m < 4; m++) {
+                        for (int j = 0; j < ncols_interleaved; j++) {
+                            sumi = 0;
+                            for (int i = 0; i < blocklen; ++i) {
+                                const int v0 = b_ptr[l].qs[k * ncols_interleaved * blocklen + j * blocklen + i];
+                                sumi += v0 * a_ptr[l].qs[k * 4 * blocklen + m * blocklen + i];
+                            }
+                            sumf[m][j] +=
+                                sumi * GGML_CPU_FP16_TO_FP32(b_ptr[l].d[j]) * GGML_CPU_FP16_TO_FP32(a_ptr[l].d[m]);
+                        }
+                    }
+                }
+            }
+            for (int m = 0; m < 4; m++) {
+                for (int j = 0; j < ncols_interleaved; j++) {
+                    s[(y * 4 + m) * bs + x * ncols_interleaved + j] = sumf[m][j];
+                }
+            }
+        }
+    }
+}
+
+void ggml_gemm_q8_0_4x8_q8_0_generic(int                        n,
+                                     float * GGML_RESTRICT      s,
+                                     size_t                     bs,
+                                     const void * GGML_RESTRICT vx,
+                                     const void * GGML_RESTRICT vy,
+                                     int                        nr,
+                                     int                        nc) {
+    const int qk                = QK8_0;
+    const int nb                = n / qk;
+    const int ncols_interleaved = 4;
+    const int blocklen          = 8;
+
+    assert(n % qk == 0);
+    assert(nr % 4 == 0);
+    assert(nc % ncols_interleaved == 0);
+
+    float sumf[4][4];
+    int   sumi;
+
+    for (int y = 0; y < nr / 4; y++) {
+        const block_q8_0x4 * a_ptr = (const block_q8_0x4 *) vy + (y * nb);
+        for (int x = 0; x < nc / ncols_interleaved; x++) {
+            const block_q8_0x4 * b_ptr = (const block_q8_0x4 *) vx + (x * nb);
+            for (int m = 0; m < 4; m++) {
+                for (int j = 0; j < ncols_interleaved; j++) {
+                    sumf[m][j] = 0.0;
+                }
+            }
+            for (int l = 0; l < nb; l++) {
+                for (int k = 0; k < (qk / blocklen); k++) {
+                    for (int m = 0; m < 4; m++) {
+                        for (int j = 0; j < ncols_interleaved; j++) {
+                            sumi = 0;
+                            for (int i = 0; i < blocklen; ++i) {
+                                const int v0 = b_ptr[l].qs[k * ncols_interleaved * blocklen + j * blocklen + i];
+                                sumi += v0 * a_ptr[l].qs[k * 4 * blocklen + m * blocklen + i];
+                            }
+                            sumf[m][j] +=
+                                sumi * GGML_CPU_FP16_TO_FP32(b_ptr[l].d[j]) * GGML_CPU_FP16_TO_FP32(a_ptr[l].d[m]);
+                        }
+                    }
+                }
+            }
+            for (int m = 0; m < 4; m++) {
+                for (int j = 0; j < ncols_interleaved; j++) {
+                    s[(y * 4 + m) * bs + x * ncols_interleaved + j] = sumf[m][j];
+                }
+            }
+        }
+    }
+}
+
 } // extern "C"
 
+static block_q8_0x4 make_block_q8_0x4(block_q8_0 * in, unsigned int blck_size_interleave) {
+    block_q8_0x4 out;
+
+    for (int i = 0; i < 4; i++) {
+        out.d[i] = in[i].d;
+    }
+
+    const int end = QK8_0 * 4 / blck_size_interleave;
+    for (int i = 0; i < end; ++i) {
+        int src_id     = i % 4;
+        int src_offset = (i / 4) * blck_size_interleave;
+        int dst_offset = i * blck_size_interleave;
+        memcpy(&out.qs[dst_offset], &in[src_id].qs[src_offset], blck_size_interleave);
+    }
+    return out;
+}
+
 static block_q4_0x4 make_block_q4_0x4(block_q4_0 * in, unsigned int blck_size_interleave) {
     block_q4_0x4 out;
 
@@ -1534,6 +1749,38 @@ static int repack_q4_0_to_q4_0_8_bl(struct ggml_tensor * t, int interleave_block
     GGML_UNUSED(data_size);
 }
 
+static int repack_q8_0_to_q8_0_4_bl(struct ggml_tensor *       t,
+                                    int                        interleave_block,
+                                    const void * GGML_RESTRICT data,
+                                    size_t                     data_size) {
+    GGML_ASSERT(t->type == GGML_TYPE_Q8_0);
+    GGML_ASSERT(interleave_block == 4 || interleave_block == 8);
+    constexpr int nrows_interleaved = 4;
+
+    block_q8_0x4 *     dst = (block_q8_0x4 *) t->data;
+    const block_q8_0 * src = (const block_q8_0 *) data;
+    block_q8_0         dst_tmp[4];
+    int                nrow    = ggml_nrows(t);
+    int                nblocks = t->ne[0] / QK8_0;
+
+    GGML_ASSERT(data_size == nrow * nblocks * sizeof(block_q8_0));
+
+    if (t->ne[1] % nrows_interleaved != 0 || t->ne[0] % 8 != 0) {
+        return -1;
+    }
+
+    for (int b = 0; b < nrow; b += nrows_interleaved) {
+        for (int64_t x = 0; x < nblocks; x++) {
+            for (int i = 0; i < nrows_interleaved; i++) {
+                dst_tmp[i] = src[x + i * nblocks];
+            }
+            *dst++ = make_block_q8_0x4(dst_tmp, interleave_block);
+        }
+        src += nrows_interleaved * nblocks;
+    }
+    return 0;
+}
+
 static block_iq4_nlx4 make_block_iq4_nlx4(block_iq4_nl * in, unsigned int blck_size_interleave) {
     block_iq4_nlx4 out;
 
@@ -1702,6 +1949,14 @@ template <> int repack<block_iq4_nl, 8, 8>(struct ggml_tensor * t, const void *
     return repack_iq4_nl_to_iq4_nl_8_bl(t, 8, data, data_size);
 }
 
+template <> int repack<block_q8_0, 4, 4>(struct ggml_tensor * t, const void * data, size_t data_size) {
+    return repack_q8_0_to_q8_0_4_bl(t, 4, data, data_size);
+}
+
+template <> int repack<block_q8_0, 8, 4>(struct ggml_tensor * t, const void * data, size_t data_size) {
+    return repack_q8_0_to_q8_0_4_bl(t, 8, data, data_size);
+}
+
 // gemv
 template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS, ggml_type PARAM_TYPE>
 void gemv(int, float *, size_t, const void *, const void *, int, int);
@@ -1738,6 +1993,14 @@ template <> void gemv<block_iq4_nl, 8, 8, GGML_TYPE_Q8_0>(int n, float * s, size
     ggml_gemv_iq4_nl_8x8_q8_0(n, s, bs, vx, vy, nr, nc);
 }
 
+template <> void gemv<block_q8_0, 4, 4, GGML_TYPE_Q8_0>(int n, float * s, size_t bs, const void * vx, const void * vy, int nr, int nc) {
+    ggml_gemv_q8_0_4x4_q8_0(n, s, bs, vx, vy, nr, nc);
+}
+
+template <> void gemv<block_q8_0, 8, 4, GGML_TYPE_Q8_0>(int n, float * s, size_t bs, const void * vx, const void * vy, int nr, int nc) {
+    ggml_gemv_q8_0_4x8_q8_0(n, s, bs, vx, vy, nr, nc);
+}
+
 // gemm
 template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS, ggml_type PARAM_TYPE>
 void gemm(int, float *, size_t, const void *, const void *, int, int);
@@ -1774,6 +2037,14 @@ template <> void gemm<block_iq4_nl, 8, 8, GGML_TYPE_Q8_0>(int n, float * s, size
     ggml_gemm_iq4_nl_8x8_q8_0(n, s, bs, vx, vy, nr, nc);
 }
 
+template <> void gemm<block_q8_0, 4, 4, GGML_TYPE_Q8_0>(int n, float * s, size_t bs, const void * vx, const void * vy, int nr, int nc) {
+    ggml_gemm_q8_0_4x4_q8_0(n, s, bs, vx, vy, nr, nc);
+}
+
+template <> void gemm<block_q8_0, 8, 4, GGML_TYPE_Q8_0>(int n, float * s, size_t bs, const void * vx, const void * vy, int nr, int nc) {
+    ggml_gemm_q8_0_4x8_q8_0(n, s, bs, vx, vy, nr, nc);
+}
+
 class tensor_traits_base : public ggml::cpu::tensor_traits {
   public:
     virtual int repack(struct ggml_tensor * t, const void * data, size_t data_size) = 0;
@@ -2168,6 +2439,10 @@ static const ggml::cpu::tensor_traits * ggml_repack_get_optimal_repack_type(cons
     static const ggml::cpu::repack::tensor_traits<block_iq4_nl, 4, 4, GGML_TYPE_Q8_0> iq4_nl_4x4_q8_0;
     static const ggml::cpu::repack::tensor_traits<block_iq4_nl, 8, 8, GGML_TYPE_Q8_0> iq4_nl_8x8_q8_0;
 
+    // instance for Q8_0
+    static const ggml::cpu::repack::tensor_traits<block_q8_0, 4, 4, GGML_TYPE_Q8_0> q8_0_4x4_q8_0;
+    static const ggml::cpu::repack::tensor_traits<block_q8_0, 8, 4, GGML_TYPE_Q8_0> q8_0_4x8_q8_0;
+
     if (cur->type == GGML_TYPE_Q4_0) {
         if (ggml_cpu_has_avx2() || (ggml_cpu_has_sve() && ggml_cpu_has_matmul_int8() && ggml_cpu_get_sve_cnt() == QK8_0)
             || (ggml_cpu_has_riscv_v() && (ggml_cpu_get_rvv_vlen() >= QK4_0))) {
@@ -2218,6 +2493,17 @@ static const ggml::cpu::tensor_traits * ggml_repack_get_optimal_repack_type(cons
                 return &iq4_nl_4x4_q8_0;
             }
         }
+    } else if (cur->type == GGML_TYPE_Q8_0) {
+        if (ggml_cpu_has_neon() && ggml_cpu_has_matmul_int8()) {
+            if (cur->ne[1] % 4 == 0) {
+                return &q8_0_4x8_q8_0;
+            }
+        }
+        if (ggml_cpu_has_neon() && ggml_cpu_has_dotprod()) {
+            if (cur->ne[1] % 4 == 0) {
+                return &q8_0_4x4_q8_0;
+            }
+        }
     }
 
     return nullptr;

ggml/src/ggml-cpu/repack.h

@@ -98,6 +98,10 @@ void ggml_gemm_q4_K_8x8_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const vo
 void ggml_gemm_q2_K_8x8_q8_K(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
 void ggml_gemm_iq4_nl_4x4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
 void ggml_gemm_iq4_nl_8x8_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemv_q8_0_4x4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemv_q8_0_4x8_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemm_q8_0_4x4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemm_q8_0_4x8_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
 
 // Native implementations
 void ggml_quantize_mat_q8_0_4x4_generic(const float * GGML_RESTRICT x, void * GGML_RESTRICT vy, int64_t k);
@@ -120,6 +124,10 @@ void ggml_gemm_q4_K_8x8_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs,
 void ggml_gemm_q2_K_8x8_q8_K_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
 void ggml_gemm_iq4_nl_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
 void ggml_gemm_iq4_nl_8x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemv_q8_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemv_q8_0_4x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemm_q8_0_4x4_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
+void ggml_gemm_q8_0_4x8_q8_0_generic(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, const void * GGML_RESTRICT vy, int nr, int nc);
 
 #if defined(__cplusplus)
 } // extern "C"

ggml/src/ggml-cuda/argmax.cu

@@ -21,7 +21,7 @@ static __global__ void argmax_f32(const float * __restrict__ x, int32_t * __rest
     }
 
 #pragma unroll
-    for (int offset = 16; offset > 0; offset >>= 1) {
+    for (int offset = WARP_SIZE/2; offset > 0; offset >>= 1) {
         const float val = __shfl_xor_sync(0xFFFFFFFF, maxval, offset, WARP_SIZE);
         const int   col = __shfl_xor_sync(0xFFFFFFFF, argmax, offset, WARP_SIZE);
         if (val > maxval) {
@@ -50,7 +50,7 @@ static __global__ void argmax_f32(const float * __restrict__ x, int32_t * __rest
                 argmax = shared_argmax[lane_id];
             }
 #pragma unroll
-            for (int offset = 16; offset > 0; offset >>= 1) {
+            for (int offset = WARP_SIZE/2; offset > 0; offset >>= 1) {
                 const float val = __shfl_xor_sync(0xFFFFFFFF, maxval, offset, WARP_SIZE);
                 const int   col = __shfl_xor_sync(0xFFFFFFFF, argmax, offset, WARP_SIZE);
                 if (val > maxval) {

ggml/src/ggml-cuda/mma.cuh

@@ -76,15 +76,31 @@ namespace ggml_cuda_mma {
         // For the A/C matrices this means I major == row major, J major == column major.
         // For the B matrix this means I major == column major, J major == row major.
         // MIRRORED == Each data value is held exactly once per thread subgroup.
-        DATA_LAYOUT_I_MAJOR           =  0, // Always used for Turing, Ampere, Ada Lovelace, consumer Blackwell.
-        DATA_LAYOUT_I_MAJOR_MIRRORED  = 10,
-        DATA_LAYOUT_J_MAJOR_MIRRORED  = 20,
+        DATA_LAYOUT_I_MAJOR           =  0, // Always used for Turing, Ampere, Ada Lovelace, consumer Blackwell, matrix A&B for RDNA4 and CDNA.
+        DATA_LAYOUT_J_MAJOR           = 10, // Matrix C for CDNA and RDNA4, int and float matrix C for RDNA3.
+        DATA_LAYOUT_I_MAJOR_MIRRORED  = 20,
+        DATA_LAYOUT_J_MAJOR_MIRRORED  = 30,
+        DATA_LAYOUT_I_MAJOR_DUAL      = 40, // Matrix A&B for RDNA3.
     };
     // Implemented mma combinations are:
     //   - (I_MAJOR, I_MAJOR)          -> I_MAJOR
     //   - (I_MAJOR, I_MAJOR_MIRRORED) -> I_MAJOR
     //   - (I_MAJOR, J_MAJOR_MIRRORED) -> I_MAJOR
 
+    constexpr bool is_i_major(const data_layout dl) {
+        return dl == DATA_LAYOUT_I_MAJOR ||
+               dl == DATA_LAYOUT_I_MAJOR_MIRRORED ||
+               dl == DATA_LAYOUT_I_MAJOR_DUAL;
+    }
+
+    constexpr data_layout get_input_data_layout() {
+#if defined(RDNA3)
+        return DATA_LAYOUT_I_MAJOR_DUAL;
+#else
+        return DATA_LAYOUT_I_MAJOR;
+#endif // defined(RDNA3)
+    }
+
     template <int I_, int J_, typename T, data_layout ds_=DATA_LAYOUT_I_MAJOR>
     struct tile {};
 
@@ -115,9 +131,9 @@ namespace ggml_cuda_mma {
             } else if constexpr (I == 32 && J == 4) {
                 return threadIdx.x % 32;
             } else if constexpr (I == 16 && J == 16) {
-                return 4 * (threadIdx.x / 16) + l;
+                return threadIdx.x % 16;
             } else if constexpr (I == 32 && J == 32) {
-                return 4 * (threadIdx.x / 32) + 8 * (l / 4) + (l % 4);
+                return threadIdx.x % 32;
             } else {
                 NO_DEVICE_CODE;
                 return -1;
@@ -132,9 +148,9 @@ namespace ggml_cuda_mma {
             } else if constexpr (I == 32 && J == 4) {
                 return 2 * (threadIdx.x / 32) + l;
             } else if constexpr (I == 16 && J == 16) {
-                return threadIdx.x % 16;
+                return 4 * (threadIdx.x / 16) + l;
             } else if constexpr (I == 32 && J == 32) {
-                return threadIdx.x % 32;
+                return 4 * (threadIdx.x / 32) + 8 * (l / 4) + (l % 4);
             } else {
                 NO_DEVICE_CODE;
                 return -1;
@@ -171,28 +187,19 @@ namespace ggml_cuda_mma {
             }
         }
 #elif defined(AMD_WMMA_AVAILABLE)
-#if defined(RDNA4)
         static constexpr int ne = I * J / 32;
-#elif defined(RDNA3)
-        static constexpr int ne = (I == 16 && J == 16) ? I * J / 32 : I * J / 16;
-#endif // defined(RDNA4)
         T x[ne] = {0};
 
         static constexpr __device__ bool supported() {
             if (I == 16 && J == 16) return true;
+            if (I == 16 && J == 8) return true;
+            if (I == 16 && J == 4) return true;
             return false;
         }
 
         static __device__ __forceinline__ int get_i(const int l) {
-            if constexpr (I == 16 && J == 16) {
-#if defined(RDNA4)
-                return 8 * (threadIdx.x / 16) + l;
-#elif defined(RDNA3)
-                return 2 * l + (threadIdx.x / 16);
-#else
-                NO_DEVICE_CODE;
-                return -1;
-#endif // defined(RDNA4)
+            if constexpr (supported()) {
+                return threadIdx.x % 16;
             } else {
                 NO_DEVICE_CODE;
                 return -1;
@@ -201,7 +208,17 @@ namespace ggml_cuda_mma {
 
         static __device__ __forceinline__ int get_j(const int l) {
             if constexpr (I == 16 && J == 16) {
-                return threadIdx.x % 16;
+                // matrix C
+#if defined(RDNA3)
+                return 2 * l + (threadIdx.x / 16);
+#else
+                return ne * (threadIdx.x / 16) + l;
+#endif // defined(RDNA3)
+            } else if constexpr (I == 16 && J == 8) {
+                // mmq input for RDNA4
+                return ne * (threadIdx.x / 16) + l;
+            } else if constexpr (I == 16 && J == 4) {
+                return ne * (threadIdx.x / 16) + l;
             } else {
                 NO_DEVICE_CODE;
                 return -1;
@@ -293,12 +310,7 @@ namespace ggml_cuda_mma {
             }
         }
 #elif defined(AMD_WMMA_AVAILABLE)
-#if defined(RDNA3)
-        // RDNA3 has duplicated data as input.
-        static constexpr int ne = I * J / 32 * 2;
-#else
         static constexpr int ne = I * J / 32;
-#endif // defined(RDNA3)
         half2 x[ne] = {{0.0f, 0.0f}};
 
         static constexpr __device__ bool supported() {
@@ -317,14 +329,7 @@ namespace ggml_cuda_mma {
 
         static __device__ __forceinline__ int get_j(const int l) {
             if constexpr (I == 16 && J == 8) {
-#if defined(RDNA4)
                 return 4 * (threadIdx.x / 16) + l;
-#elif defined(RDNA3)
-                return l;
-#else
-                NO_DEVICE_CODE;
-                return -1;
-#endif // defined(RDNA4)
             } else {
                 NO_DEVICE_CODE;
                 return -1;
@@ -382,42 +387,19 @@ namespace ggml_cuda_mma {
         static constexpr data_layout dl = DATA_LAYOUT_I_MAJOR;
 
 #if defined(AMD_WMMA_AVAILABLE)
-#if defined(RDNA3)
-        // RDNA3 has duplicated data as input.
-        static constexpr int ne = I * J / 32 * 2;
-#else
         static constexpr int ne = I * J / 32;
-#endif // defined(RDNA3)
         nv_bfloat162 x[ne] = {{0.0f, 0.0f}};
 
         static constexpr __device__ bool supported() {
-            if (I == 16 && J == 8) return true;
-            return false;
+            return tile<I_, J_, half2, DATA_LAYOUT_I_MAJOR>::supported();
         }
 
         static __device__ __forceinline__ int get_i(const int l) {
-            if constexpr (I == 16 && J == 8) {
-                return threadIdx.x % 16;
-            } else {
-                NO_DEVICE_CODE;
-                return -1;
-            }
+            return tile<I_, J_, half2, DATA_LAYOUT_I_MAJOR>::get_i(l);
         }
 
         static __device__ __forceinline__ int get_j(const int l) {
-            if constexpr (I == 16 && J == 8) {
-#if defined(RDNA4)
-                return 4 * (threadIdx.x / 16) + l;
-#elif defined(RDNA3)
-                return l;
-#else
-                NO_DEVICE_CODE;
-                return -1;
-#endif // defined(RDNA4)
-            } else {
-                NO_DEVICE_CODE;
-                return -1;
-            }
+            return tile<I_, J_, half2, DATA_LAYOUT_I_MAJOR>::get_j(l);
         }
 #else
         static constexpr int ne = I * J / WARP_SIZE;
@@ -458,6 +440,28 @@ namespace ggml_cuda_mma {
 #endif  // defined(AMD_WMMA_AVAILABLE)
     };
 
+    template <int I_, int J_, typename T>
+    struct tile<I_, J_, T, DATA_LAYOUT_J_MAJOR> {
+        static constexpr int         I  = I_;
+        static constexpr int         J  = J_;
+        static constexpr data_layout dl = DATA_LAYOUT_J_MAJOR;
+
+        static constexpr int ne = tile<I_, J_, T, DATA_LAYOUT_I_MAJOR>::ne;
+        T x[ne] = {0};
+
+        static constexpr __device__ bool supported() {
+            return tile<I_, J_, T, DATA_LAYOUT_I_MAJOR>::supported();
+        }
+
+        static __device__ __forceinline__ int get_i(const int l) {
+            return tile<I_, J_, T, DATA_LAYOUT_I_MAJOR>::get_j(l);
+        }
+
+        static __device__ __forceinline__ int get_j(const int l) {
+            return tile<I_, J_, T, DATA_LAYOUT_I_MAJOR>::get_i(l);
+        }
+    };
+
     template <int I_, int J_>
     struct tile<I_, J_, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> {
         static constexpr int         I  = I_;
@@ -524,6 +528,42 @@ namespace ggml_cuda_mma {
         }
     };
 
+    template <int I_, int J_, typename T>
+    struct tile<I_, J_, T, DATA_LAYOUT_I_MAJOR_DUAL> {
+        static constexpr int         I  = I_;
+        static constexpr int         J  = J_;
+        static constexpr data_layout dl = DATA_LAYOUT_I_MAJOR_DUAL;
+
+        static constexpr int         ne = I * J / 32 * 2;
+
+        T x[ne] = {0};
+
+        static constexpr __device__ bool supported() {
+            if (I == 16 && J == 16) return true;
+            if (I == 16 && J == 8)  return true;
+            if (I == 16 && J == 4)  return true;
+            return false;
+        }
+
+        static __device__ __forceinline__ int get_i(const int l) {
+            if constexpr (supported()) {
+                return threadIdx.x % 16;
+            } else {
+                NO_DEVICE_CODE;
+                return -1;
+            }
+        }
+
+        static __device__ __forceinline__ int get_j(const int l) {
+            if constexpr (supported()) {
+                return l;
+            } else {
+                NO_DEVICE_CODE;
+                return -1;
+            }
+        }
+    };
+
 #if defined(TURING_MMA_AVAILABLE)
     template <int I, int J>
     static __device__ __forceinline__ tile<I, J/2, half2> get_half2(const tile<I, J, float> & tile_float) {
@@ -569,55 +609,28 @@ namespace ggml_cuda_mma {
                 t.x[l] = xs0[t.get_i(l)*stride + t.get_j(l)];
             }
         } else {
-            int64_t * xi = (int64_t *) t.x;
-            const int64_t * xs = (int64_t *) ((const int *) xs0 + (threadIdx.x % t.I) * stride + 2 * (threadIdx.x / t.I));
-            xi[0] = xs[0];
+            ggml_cuda_memcpy_1<sizeof(t.x)>(t.x, xs0 + t.get_i(0) * stride + t.get_j(0));
         }
 #elif defined(AMD_WMMA_AVAILABLE)
-        if constexpr (std::is_same_v<T, half2> || std::is_same_v<T, nv_bfloat162>) {
-#if defined(RDNA4)
-                ggml_cuda_memcpy_1<sizeof(t.x)>(t.x, xs0 + t.get_i(0) * stride + t.get_j(0));
-#elif defined(RDNA3)
-                ggml_cuda_memcpy_1<sizeof(t.x)/2>(t.x, xs0 + t.get_i(0) * stride + t.get_j(0));
-                ggml_cuda_memcpy_1<sizeof(t.x)/2>(t.x + t.ne/2, xs0 + t.get_i(0) * stride + t.get_j(t.ne/2));
-#else
-                NO_DEVICE_CODE;
-#endif // defined(RDNA4)
-        } else if constexpr (std::is_same_v<T, int>) {
-            if constexpr (I == 16 && J == 4) {
-                int64_t * xi = (int64_t *) t.x;
-#if defined(RDNA4)
-                const int64_t * xs = (int64_t *) ((const int *) xs0 + (threadIdx.x % t.I) * stride + 2 * (threadIdx.x / t.I));
-                xi[0] = xs[0];
-#elif defined(RDNA3)
-                static_assert(tile<I,J,T>::ne >= 4, "fragment too small");
-                const int64_t * xs = (int64_t *) ((const int *) xs0 + (threadIdx.x % t.I) * stride);
-                xi[0] = xs[0];
-                xi[1] = xs[1];
-#endif // defined(RDNA4)
-            } else if constexpr (I == 16 && J == 8) {
-                int64_t * xi = (int64_t *) t.x;
-#if defined(RDNA4)
-                const int64_t * xs = (int64_t *) ((const int *) xs0 + (threadIdx.x % t.I) * stride + 4 * (threadIdx.x / t.I));
-                xi[0] = xs[0];
-
-                const int64_t * xs1 = (int64_t *) ((const int *) xs0 + (threadIdx.x % t.I) * stride + 4 * (threadIdx.x / t.I) + 2);
-                xi[1] = xs1[0];
-#elif defined(RDNA3)
-                static_assert(tile<I,J,T>::ne >= 8, "fragment too small");
-                const int64_t * xs = (int64_t *) ((const int *) xs0 + (threadIdx.x % t.I) * stride);
-                // contiguous four 64-bit chunks per lane for the wider RDNA3 fragment
-                xi[0] = xs[0];
-                xi[1] = xs[1];
-                const int64_t * xs1 = xs + 2;
-                xi[2] = xs1[0];
-                xi[3] = xs1[1];
-#endif // defined(RDNA4)
-            } else {
-                NO_DEVICE_CODE;
+        // All wmma layout has contiguous data when i-major.
+        if constexpr (is_i_major(dl)) {
+            // the data must be aligned to 16 bytes when bigger than ggml_cuda_get_max_cpy_bytes()
+            constexpr int aligned_copy_bytes = ggml_cuda_get_max_cpy_bytes();
+            if constexpr (sizeof(t.x) > aligned_copy_bytes) {
+                static_assert(sizeof(t.x) % aligned_copy_bytes == 0, "bad type size");
+                constexpr int aligned_copy_count = sizeof(t.x)/aligned_copy_bytes;
+#pragma unroll
+                for (int i = 0; i < aligned_copy_count; ++i) {
+                    ggml_cuda_memcpy_1<aligned_copy_bytes>(t.x + t.ne/aligned_copy_count*i, xs0 + t.get_i(0) * stride + t.get_j(t.ne/aligned_copy_count*i));
                 }
             } else {
-            NO_DEVICE_CODE;
+                ggml_cuda_memcpy_1<sizeof(t.x)>(t.x, xs0 + t.get_i(0) * stride + t.get_j(0));
+            }
+        } else {
+#pragma unroll
+            for (int l = 0; l < t.ne; ++l) {
+                t.x[l] = xs0[t.get_i(l)*stride + t.get_j(l)];
+            }
         }
 #else
 #pragma unroll
@@ -660,9 +673,9 @@ namespace ggml_cuda_mma {
 #endif // TURING_MMA_AVAILABLE
     }
 
-    template <typename T>
+    template <typename T, data_layout dl>
     static __device__ __forceinline__ void load_ldmatrix(
-            tile<16, 8, T> & t, const T * __restrict__ xs0, const int stride) {
+            tile<16, 8, T, dl> & t, const T * __restrict__ xs0, const int stride) {
 #if defined(TURING_MMA_AVAILABLE)
         int * xi = (int * ) t.x;
         const int * xs = (const int *) xs0 + (threadIdx.x % t.I) * stride + (threadIdx.x / t.I) * (t.J / 2);
@@ -832,8 +845,9 @@ namespace ggml_cuda_mma {
 #endif // TURING_MMA_AVAILABLE
     }
 
+    template <data_layout dl_ab, data_layout dl_d>
     static __device__ __forceinline__ void mma(
-            tile<16, 8, float> & D, const tile<16, 8, float> & A, const tile<8, 8, float> & B) {
+            tile<16, 8, float, dl_d> & D, const tile<16, 8, float, dl_ab> & A, const tile<8, 8, float, dl_ab> & B) {
 #ifdef AMPERE_MMA_AVAILABLE
         const int * Axi = (const int *) A.x;
         const int * Bxi = (const int *) B.x;
@@ -887,8 +901,9 @@ namespace ggml_cuda_mma {
 #endif // AMPERE_MMA_AVAILABLE
     }
 
+    template <data_layout dl_ab, data_layout dl_d>
     static __device__ __forceinline__ void mma(
-            tile<16, 16, float> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
+            tile<16, 16, float, dl_d> & D, const tile<16, 8, half2, dl_ab> & A, const tile<16, 8, half2, dl_ab> & B) {
 #ifdef TURING_MMA_AVAILABLE
         const int * Axi = (const int *) A.x;
         const int * Bxi = (const int *) B.x;
@@ -940,8 +955,9 @@ namespace ggml_cuda_mma {
 #endif // TURING_MMA_AVAILABLE
     }
 
+    template <data_layout dl_ab, data_layout dl_d>
     static __device__ __forceinline__ void mma(
-            tile<16, 16, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<16, 8, nv_bfloat162> & B) {
+            tile<16, 16, float, dl_d> & D, const tile<16, 8, nv_bfloat162, dl_ab> & A, const tile<16, 8, nv_bfloat162, dl_ab> & B) {
 #if defined(AMD_WMMA_AVAILABLE)
 #if defined(RDNA4)
         using bf16x8_t = __attribute__((ext_vector_type(8))) __bf16;
@@ -967,8 +983,9 @@ namespace ggml_cuda_mma {
 #endif // AMPERE_MMA_AVAILABLE
     }
 
+    template <data_layout dl_d, data_layout dl_ab>
     static __device__ __forceinline__ void mma(
-            tile<16, 16, int> & D, const tile<16, 8, int> & A, const tile<16, 8, int> & B) {
+            tile<16, 16, int, dl_d> & D, const tile<16, 8, int, dl_ab> & A, const tile<16, 8, int, dl_ab> & B) {
 #if defined(AMD_MFMA_AVAILABLE)
         using int32x4_t = __attribute__((__vector_size__(4 * sizeof(int)))) int;
         int32x4_t * acc = (int32x4_t *) D.x;
@@ -1122,8 +1139,9 @@ namespace ggml_cuda_mma {
 #endif // __CUDA_ARCH__ >= GGML_CUDA_CC_VOLTA
     }
 
+    template <data_layout dl_d, data_layout dl_ab>
     static __device__ __forceinline__ void mma(
-            tile<16, 16, int> & D, const tile<16, 4, int> & A, const tile<16, 4, int> & B) {
+            tile<16, 16, int, dl_d> & D, const tile<16, 4, int, dl_ab> & A, const tile<16, 4, int, dl_ab> & B) {
 #if defined(AMD_WMMA_AVAILABLE)
         using int32x8_t = __attribute__((__vector_size__(8 * sizeof(int)))) int;
         int32x8_t * acc = (int32x8_t *) D.x;

ggml/src/ggml-cuda/mmf.cuh

@@ -32,11 +32,13 @@ static __global__ void mul_mat_f(
 #if (!defined(GGML_USE_HIP) && !defined(GGML_USE_MUSA)) || defined(AMD_WMMA_AVAILABLE)
 #if defined(AMD_WMMA_AVAILABLE)
     // Special case for tf32, just dummy mma layout as wmma doesn't support it.
-    constexpr int tile_B_I = std::is_same_v<T, float> ? 8 : 16;
-    constexpr int tile_C_J = std::is_same_v<T, float> ? 8 : 16;
-    typedef tile<16,       8, T>     tile_A;
-    typedef tile<tile_B_I, 8, T>     tile_B;
-    typedef tile<16,       tile_C_J, float> tile_C;
+    constexpr bool is_tf32 = std::is_same_v<T, float>;
+    constexpr int tile_B_I = is_tf32 ? 8 : 16;
+    constexpr int tile_C_J = is_tf32 ? 8 : 16;
+    constexpr data_layout ab_layout = is_tf32 ? DATA_LAYOUT_I_MAJOR : get_input_data_layout();
+    typedef tile<16,       8,        T,     ab_layout>           tile_A;
+    typedef tile<tile_B_I, 8,        T,     ab_layout>           tile_B;
+    typedef tile<16,       tile_C_J, float, DATA_LAYOUT_J_MAJOR> tile_C;
 #else
 #ifdef VOLTA_MMA_AVAILABLE
     if constexpr (!std::is_same_v<T, half2>) {NO_DEVICE_CODE;} else {
@@ -272,11 +274,13 @@ static __global__ void mul_mat_f_ids(
 #if (!defined(GGML_USE_HIP) && !defined(GGML_USE_MUSA)) || defined(AMD_WMMA_AVAILABLE)
 #if defined(AMD_WMMA_AVAILABLE)
     // Special case for tf32, just dummy mma layout as wmma doesn't support it.
-    constexpr int tile_B_I = std::is_same_v<T, float> ? 8 : 16;
-    constexpr int tile_C_J = std::is_same_v<T, float> ? 8 : 16;
-    typedef tile<16,       8, T>     tile_A;
-    typedef tile<tile_B_I, 8, T>     tile_B;
-    typedef tile<16,       tile_C_J, float> tile_C;
+    constexpr bool is_tf32 = std::is_same_v<T, float>;
+    constexpr int tile_B_I = is_tf32 ? 8 : 16;
+    constexpr int tile_C_J = is_tf32 ? 8 : 16;
+    constexpr data_layout ab_layout = is_tf32 ? DATA_LAYOUT_I_MAJOR : get_input_data_layout();
+    typedef tile<16,       8,        T,     ab_layout>           tile_A;
+    typedef tile<tile_B_I, 8,        T,     ab_layout>           tile_B;
+    typedef tile<16,       tile_C_J, float, DATA_LAYOUT_J_MAJOR> tile_C;
 #else
 #ifdef VOLTA_MMA_AVAILABLE
     if constexpr (!std::is_same_v<T, half2>) {NO_DEVICE_CODE;} else {

ggml/src/ggml-cuda/mmq.cuh

@@ -797,9 +797,10 @@ template <int mmq_x, int mmq_y, mmq_q8_1_ds_layout ds_layout>
 static __device__ __forceinline__ void vec_dot_q8_0_q8_1_mma(
     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int k00) {
 #if defined(AMD_MFMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
-    typedef tile<16,  8, int> tile_A;
-    typedef tile<16,  8, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  8, int, input_layout>        tile_A;
+    typedef tile<16,  8, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -966,9 +967,10 @@ template <int mmq_x, int mmq_y>
 static __device__ __forceinline__ void vec_dot_q8_1_q8_1_mma(
     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int k00) {
 #if defined(AMD_MFMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
-    typedef tile<16,  8, int> tile_A;
-    typedef tile<16,  8, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  8, int, input_layout>        tile_A;
+    typedef tile<16,  8, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -1130,10 +1132,11 @@ template <int mmq_x, int mmq_y>
 static __device__ __forceinline__ void vec_dot_q8_0_16_q8_1_mma(
     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int k00) {
 #if defined(AMD_MFMA_AVAILABLE)
-    typedef tile<16,  8, int> tile_A;
-    typedef tile<16,  8, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
-    typedef tile<64,  2, int> tile_load;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  8, int, input_layout>        tile_A;
+    typedef tile<16,  8, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
+    typedef tile<64,  2, int, input_layout>        tile_load;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -1179,9 +1182,10 @@ static __device__ __forceinline__ void vec_dot_q8_0_16_q8_1_mma(
         }
     }
 #elif defined(AMD_WMMA_AVAILABLE) //wmma instructions can handle 16x4 tiles, does not require loading 64x2 tiles
-    typedef tile<16,  4, int> tile_A;
-    typedef tile<16,  4, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  4, int, input_layout>        tile_A;
+    typedef tile<16,  4, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -1435,10 +1439,11 @@ template <int mmq_x, int mmq_y>
 static __device__ __forceinline__ void vec_dot_q2_K_q8_1_mma(
     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int k00) {
 #if defined(AMD_MFMA_AVAILABLE)
-    typedef tile<16,  8, int> tile_A;
-    typedef tile<16,  8, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
-    typedef tile<64,  2, int> tile_load;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  8, int, input_layout>        tile_A;
+    typedef tile<16,  8, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
+    typedef tile<64,  2, int, input_layout>        tile_load;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -1501,10 +1506,10 @@ static __device__ __forceinline__ void vec_dot_q2_K_q8_1_mma(
         }
     }
 #elif defined(AMD_WMMA_AVAILABLE) //wmma instructions can handle 16x4 tiles, does not require loading 64x2 tiles
-
-    typedef tile<16,  4, int> tile_A;
-    typedef tile<16,  4, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  4, int, input_layout>        tile_A;
+    typedef tile<16,  4, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -2265,10 +2270,11 @@ template <int mmq_x, int mmq_y>
 static __device__ __forceinline__ void vec_dot_q6_K_q8_1_mma(
     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int k00) {
 #if defined(AMD_MFMA_AVAILABLE)
-    typedef tile<16,  8, int> tile_A;
-    typedef tile<16,  8, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
-    typedef tile<64,  2, int> tile_load;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  8, int, input_layout>        tile_A;
+    typedef tile<16,  8, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
+    typedef tile<64,  2, int, input_layout>        tile_load;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -2316,9 +2322,10 @@ static __device__ __forceinline__ void vec_dot_q6_K_q8_1_mma(
         }
     }
 #elif defined(AMD_WMMA_AVAILABLE) //wmma instructions can handle 16x4 tiles, does not require loading 64x2 tiles
-    typedef tile<16,  4, int> tile_A;
-    typedef tile<16,  4, int> tile_B;
-    typedef tile<16, 16, int> tile_C;
+    constexpr data_layout input_layout = get_input_data_layout();
+    typedef tile<16,  4, int, input_layout>        tile_A;
+    typedef tile<16,  4, int, input_layout>        tile_B;
+    typedef tile<16, 16, int, DATA_LAYOUT_J_MAJOR> tile_C;
 
     constexpr int granularity = mmq_get_granularity_device(mmq_x);
     constexpr int rows_per_warp = granularity;
@@ -3015,7 +3022,7 @@ static __device__ __forceinline__ void mmq_write_back_mma(
 
 #if defined(AMD_MFMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
     constexpr int tileC_IJ = mmq_get_granularity_device(0);
-    typedef tile<tileC_IJ, tileC_IJ, int> tile_C;
+    typedef tile<tileC_IJ, tileC_IJ, int, DATA_LAYOUT_J_MAJOR> tile_C;
     constexpr int rows_per_warp = granularity;
 #else
     typedef tile<16, 8, int> tile_C;

ggml/src/ggml-hexagon/ggml-hexagon.cpp

@@ -1976,9 +1976,6 @@ static bool ggml_hexagon_supported_mul_mat(const struct ggml_hexagon_session * s
             break;
 
         case GGML_TYPE_F16:
-            if (!opt_experimental) {
-                return false;
-            }
             break;
 
         default:
@@ -2164,9 +2161,15 @@ static bool ggml_hexagon_supported_activations(const struct ggml_hexagon_session
     }
 
     // src0, src1 & dst must be mapped to the same session
+    if(src1){
         if (!hex_supported_buffer(sess, src0, src1, dst)) {
             return false;
         }
+    }else{
+        if (!hex_supported_buffer(sess, src0, dst)) {
+            return false;
+        }
+    }
 
     return true;
 }
@@ -2665,6 +2668,10 @@ static void ggml_hexagon_unary(const struct ggml_tensor * op, uint32_t flags) {
                 req.op    = HTP_OP_UNARY_SILU;
                 supported = true;
             }
+            else if (ggml_get_unary_op(dst) == GGML_UNARY_OP_GELU){
+                req.op    = HTP_OP_UNARY_GELU;
+                supported = true;
+            }
             break;
 
         case GGML_OP_GLU:
@@ -2680,6 +2687,7 @@ static void ggml_hexagon_unary(const struct ggml_tensor * op, uint32_t flags) {
         case GGML_OP_SOFT_MAX:
             req.op    = HTP_OP_SOFTMAX;
             supported = true;
+            break;
 
         default:
             break;
@@ -2959,6 +2967,8 @@ static ggml_status ggml_backend_hexagon_graph_compute(ggml_backend_t backend, gg
             case GGML_OP_UNARY:
                 if (ggml_get_unary_op(node) == GGML_UNARY_OP_SILU) {
                     ggml_hexagon_unary(node, flags);
+                } else if (ggml_get_unary_op(node) == GGML_UNARY_OP_GELU) {
+                    ggml_hexagon_unary(node, flags);
                 }
                 break;
             case GGML_OP_GLU:
@@ -3257,7 +3267,6 @@ static bool ggml_backend_hexagon_device_supports_op(ggml_backend_dev_t dev, cons
     auto sess = static_cast<ggml_hexagon_session *>(dev->context);
 
     bool supp = false;
-
     switch (op->op) {
         case GGML_OP_NONE:
         case GGML_OP_RESHAPE:
@@ -3297,10 +3306,13 @@ static bool ggml_backend_hexagon_device_supports_op(ggml_backend_dev_t dev, cons
             if (ggml_get_unary_op(op) == GGML_UNARY_OP_SILU) {
                 supp = ggml_hexagon_supported_activations(sess, op);
             }
+            else if (ggml_get_unary_op(op) == GGML_UNARY_OP_GELU){
+                supp = ggml_hexagon_supported_activations(sess, op);
+            }
             break;
 
         case GGML_OP_GLU:
-            if ((ggml_get_glu_op(op) == GGML_GLU_OP_SWIGLU) /* || (ggml_get_glu_op(op) == GGML_GLU_OP_SWIGLU_OAI) */) {
+            if ((ggml_get_glu_op(op) == GGML_GLU_OP_SWIGLU) || (ggml_get_glu_op(op) == GGML_GLU_OP_SWIGLU_OAI) ) {
                 supp = ggml_hexagon_supported_activations(sess, op);
             }
             break;

ggml/src/ggml-hexagon/htp/act-ops.c

@@ -231,7 +231,7 @@ static void glu_swiglu_oai_fp32_per_thread(const struct htp_tensor * src0,
         // x (src0_spad_data) = std::min(src0_p[k], limit);
         hvx_min_scalar_f32((const uint8_t *) src0, limit, src0_spad_data, nc);
         // y1 (src1_spad_data) = std::clamp(src1_p[k], -limit, limit);
-        hvx_clamp_scalar_f32((const uint8_t *) src1, limit, limit, src1_spad_data, nc);
+        hvx_clamp_scalar_f32((const uint8_t *) src1, -limit, limit, src1_spad_data, nc);
         // y (src1_spad_data)  = y1 + 1.f
         hvx_add_scalar_f32(src1_spad_data, 1.0, src1_spad_data, nc);
         // x1 (dst_spad_data) = alpha * (x)
@@ -255,6 +255,91 @@ static void glu_swiglu_oai_fp32_per_thread(const struct htp_tensor * src0,
          src1->ne[3], dst->ne[0], dst->ne[1], dst->ne[2], dst->ne[3], (unsigned) HAP_perf_qtimer_count_to_us(t2 - t1));
 }
 
+
+static void unary_gelu_fp32_per_thread(const struct htp_tensor * src0,
+                                       struct htp_tensor *       dst,
+                                       const int32_t *           op_params,
+                                       struct htp_spad *         src0_spad,
+                                       struct htp_spad *         dst_spad,
+                                       uint32_t                  nth,
+                                       uint32_t                  ith,
+                                       uint32_t                  src0_nrows_per_thread) {
+    htp_act_preamble2;
+
+    uint64_t t1, t2;
+    t1 = HAP_perf_get_qtimer_count();
+
+    const size_t src0_row_size = nb01;
+    const size_t dst_row_size  = nb1;
+
+    const uint32_t src0_nrows = ne01 * ne02 * ne03;
+
+    const uint32_t src0_start_row = src0_nrows_per_thread * ith;
+    const uint32_t src0_end_row   = MIN(src0_start_row + src0_nrows_per_thread, src0_nrows);
+
+    // no work for this thread
+    if (src0_start_row >= src0_end_row) {
+        return;
+    }
+
+    int is_aligned = 1;
+    int opt_path   = 0;
+    if (!htp_is_aligned((void *) src0->data, VLEN) || !htp_is_aligned((void *) dst->data, VLEN)) {
+        is_aligned = 0;
+        FARF(HIGH, "silu-f32: unaligned addresses in elementwise op, possibly slower execution\n");
+    }
+    if ((1 == is_aligned) && !(nb01 & (VLEN - 1))) {
+        opt_path = 1;
+    }
+
+    const uint8_t * restrict data_src0 = (const uint8_t *) src0->data;
+    uint8_t * restrict data_dst        = (uint8_t *) dst->data;
+
+    uint8_t * restrict src0_spad_data = src0_spad->data + (ith * src0_row_size);
+    uint8_t * restrict dst_spad_data  = dst_spad->data + (ith * dst_row_size);
+
+    const int BLOCK = 8;
+    for (uint32_t ir = src0_start_row; ir < src0_end_row; ir += BLOCK) {
+        const uint32_t block_end = MIN(ir + BLOCK, src0_end_row);
+
+        // Prefetch next block
+        if (block_end < src0_end_row) {
+            const float * restrict prefetch_ptr = (float *) (data_src0 + (block_end * src0_row_size));
+            htp_l2fetch(prefetch_ptr, 1, block_end * src0_row_size, src0_row_size);
+        }
+
+        // Process rows in current block
+        for (uint32_t ib = ir; ib < block_end; ib++) {
+            const float * restrict src0 = (float *) (data_src0 + (ib * src0_row_size));
+            float * restrict dst        = (float *) (data_dst + (ib * dst_row_size));
+
+            // gelu = x * sigmoid(1.702 * x) // current implementation
+            if (1 == opt_path) {
+                hvx_mul_scalar_f32((const uint8_t *) src0, (float) 1.702, (uint8_t *) src0_spad_data, ne0);
+                hvx_fast_sigmoid_f32((const uint8_t *) src0_spad_data, (uint8_t *) src0_spad_data, ne0);
+                hvx_mul_f32_opt((const uint8_t *) src0, src0_spad_data, (uint8_t *) dst, ne0);
+            } else {
+                hvx_mul_scalar_f32( (const uint8_t *) src0, (float)1.702, (uint8_t *) src0_spad_data, ne0);
+                hvx_sigmoid_f32((const uint8_t *) src0_spad_data, (uint8_t *) src0_spad_data, ne0);
+                hvx_mul_f32((const uint8_t *) src0, src0_spad_data, (uint8_t *) dst, ne0);
+            }
+        }
+    }
+
+    t2 = HAP_perf_get_qtimer_count();
+
+    FARF(HIGH, "gelu-f32 %d/%d/%d: %ux%ux%ux%u (%u:%u) -> %ux%ux%ux%u usec %u\n", ith, nth, opt_path, ne00, ne01, ne02,
+         ne03, src0_start_row, src0_end_row, ne0, ne1, ne2, ne3, (unsigned) HAP_perf_qtimer_count_to_us(t2 - t1));
+}
+
+static void unary_gelu_fp32(unsigned int n, unsigned int i, void * data) {
+    struct htp_ops_context * octx = (struct htp_ops_context *) data;
+    unary_gelu_fp32_per_thread(&octx->src0, &octx->dst, octx->op_params, &octx->src0_spad, &octx->dst_spad, n, i,
+                               octx->src0_nrows_per_thread);
+}
+
+
+
 static void unary_silu_fp32_per_thread(const struct htp_tensor * src0,
                                        struct htp_tensor *       dst,
                                        const int32_t *           op_params,
@@ -371,7 +456,10 @@ static int execute_op_activations_fp32(struct htp_ops_context * octx) {
             act_op_func = glu_swiglu_oai_fp32;
             op_type     = "swiglu-oai-f32";
             break;
-
+        case HTP_OP_UNARY_GELU:
+            act_op_func = unary_gelu_fp32;
+            op_type     = "gelu-f32";
+            break;
         default:
             FARF(ERROR, "Unsupported activations Op %u\n", octx->op);
             return HTP_STATUS_NO_SUPPORT;

ggml/src/ggml-hexagon/htp/htp-msg.h

@@ -51,11 +51,12 @@ enum htp_op {
     HTP_OP_MUL_MAT_ID     = 5,
     HTP_OP_RMS_NORM       = 6,
     HTP_OP_UNARY_SILU     = 7,
-    HTP_OP_GLU_SWIGLU     = 8,
-    HTP_OP_GLU_SWIGLU_OAI = 9,
-    HTP_OP_SOFTMAX        = 10,
-    HTP_OP_ADD_ID         = 11,
-    HTP_OP_ROPE           = 12,
+    HTP_OP_UNARY_GELU     = 8,
+    HTP_OP_GLU_SWIGLU     = 9,
+    HTP_OP_GLU_SWIGLU_OAI = 10,
+    HTP_OP_SOFTMAX        = 11,
+    HTP_OP_ADD_ID         = 12,
+    HTP_OP_ROPE           = 13,
     INVALID
 };
 

ggml/src/ggml-hexagon/htp/hvx-utils.c

@@ -49,6 +49,8 @@ void hvx_mul_f32(const uint8_t * restrict src0,
         FARF(HIGH, "hvx_mul_f32: unaligned loop in hvx op, possibly slower execution\n");
     }
 
+
+    bool handled_leftover = false;
     if (0 == unaligned_loop) {
         HVX_Vector * restrict vec_in1 = (HVX_Vector *) src0;
         HVX_Vector * restrict vec_in2 = (HVX_Vector *) src1;
@@ -60,18 +62,59 @@ void hvx_mul_f32(const uint8_t * restrict src0,
             *vec_out++   = Q6_Vsf_equals_Vqf32(v);
         }
     } else {
+        int step_of_1 = num_elems_whole >> 5;  // divby 32, because 32 float = 128 bytes per HVX vector
+        int leftover_size = left_over * sizeof(float);
+
+
+        HVX_Vector * restrict vec_in1 = (HVX_Vector *) src0;
+        HVX_Vector * restrict vec_in2 = (HVX_Vector *) src1;
+        HVX_UVector * restrict vec_out = (HVX_UVector *) dst;
+
+        HVX_Vector slinep;
+        HVX_Vector slinec;
+        HVX_Vector sline;
+        HVX_Vector sline2p;
+        HVX_Vector sline2c;
+        HVX_Vector sline2;
+
+        slinep  = *vec_in1++;
+        sline2p = *vec_in2++;
         #pragma unroll(4)
-        for (int i = 0; i < num_elems_whole; i += VLEN_FP32) {
-            HVX_Vector in1 = *(HVX_UVector *) (src0 + i * SIZEOF_FP32);
-            HVX_Vector in2 = *(HVX_UVector *) (src1 + i * SIZEOF_FP32);
+        for (int i = step_of_1 - 1; i > 0; i--) {
+            slinec  = *vec_in1++;
+            sline2c = *vec_in2++;
+            sline   = Q6_V_valign_VVR(slinec, slinep, (size_t) src0);
+            sline2  = Q6_V_valign_VVR(sline2c, sline2p, (size_t) src1);
 
-            HVX_Vector out = Q6_Vqf32_vmpy_VsfVsf(in1, in2);
-
-            *(HVX_UVector *) (dst + i * SIZEOF_FP32) = Q6_Vsf_equals_Vqf32(out);
-        }
+            *((HVX_UVector *) (vec_out++)) = Q6_Vsf_equals_Vqf32(Q6_Vqf32_vmpy_VsfVsf(sline, sline2));
+            slinep                         = slinec;
+            sline2p                        = sline2c;
         }
+        if (step_of_1 > 1) {
+            slinec  = htp_is_aligned(vec_in1, VLEN) && left_over == 0 ? slinep : *vec_in1++;
+            sline2c = htp_is_aligned(vec_in2, VLEN) && left_over == 0 ? sline2p : *vec_in2++;
 
+            sline                          = Q6_V_valign_VVR(slinec, slinep, (size_t) src0);
+            sline2                         = Q6_V_valign_VVR(sline2c, sline2p, (size_t) src1);
+            *((HVX_UVector *) (vec_out++)) = Q6_Vsf_equals_Vqf32(Q6_Vqf32_vmpy_VsfVsf(sline, sline2));
+            slinep                         = slinec;
+            sline2p                        = sline2c;
+        }
         if (left_over > 0) {
+            slinec = (is_in_one_chunk(vec_in1, leftover_size, VLEN) ? slinep : *vec_in1++);
+
+            sline   = Q6_V_valign_VVR(slinec, slinep, (size_t) src0);
+            sline2c = (is_in_one_chunk(vec_in2, leftover_size, VLEN) ? sline2p : *vec_in2++);
+            sline2  = Q6_V_valign_VVR(sline2c, sline2p, (size_t) src1);
+
+            HVX_Vector out = Q6_Vqf32_vmpy_VsfVsf(sline, sline2);
+            hvx_vec_store_u(vec_out, leftover_size, Q6_Vsf_equals_Vqf32(out));
+            handled_leftover = true;
+        }
+    }
+
+
+    if (left_over > 0 && !handled_leftover) {
         const float * src0f = (const float *) src0 + num_elems_whole;
         const float * src1f = (const float *) src1 + num_elems_whole;
         float *       dstf  = (float *) dst + num_elems_whole;
@@ -464,7 +507,7 @@ void hvx_mul_scalar_f32(const uint8_t * restrict src, const float val, uint8_t *
     }
 
     HVX_Vector val_vec = hvx_vec_splat_fp32(val);
-
+    bool handled_leftover = false;
     if (0 == unaligned_loop) {
         HVX_Vector * restrict vec_in1 = (HVX_Vector *) src;
         HVX_Vector * restrict vec_out = (HVX_Vector *) dst;
@@ -475,17 +518,47 @@ void hvx_mul_scalar_f32(const uint8_t * restrict src, const float val, uint8_t *
             *vec_out++   = Q6_Vsf_equals_Vqf32(v);
         }
     } else {
+        int step_of_1 = num_elems >> 5;  // divby 32, because 32 float = 128 bytes per HVX vector
+        int leftover_size = left_over * sizeof(float);
+
+        HVX_Vector *  input_v_ptr  = (HVX_Vector *) src;
+        HVX_UVector * output_v_ptr = (HVX_UVector *) dst;
+
+        HVX_Vector slinep;
+        HVX_Vector slinec;
+        HVX_Vector sline;
+
+        slinep = *input_v_ptr++;
+
         #pragma unroll(4)
-        for (int i = 0; i < num_elems_whole; i += VLEN_FP32) {
-            HVX_Vector in = *(HVX_UVector *) (src + i * SIZEOF_FP32);
+        for (int i = step_of_1 - 1; i > 0; i--) {
+            slinec                              = *input_v_ptr++;
+            sline                               = Q6_V_valign_VVR(slinec, slinep, (size_t) src);
+            *((HVX_UVector *) (output_v_ptr++)) = Q6_Vsf_equals_Vqf32(Q6_Vqf32_vmpy_VsfVsf(sline, val_vec));
+            /* Prepare slinep for next iteration */
+            slinep                              = slinec;
+        }
 
-            HVX_Vector out = Q6_Vqf32_vmpy_VsfVsf(in, val_vec);
+        if (step_of_1 > 0) {
+            slinec = htp_is_aligned(input_v_ptr, VLEN) && left_over == 0 ? slinep : *input_v_ptr++;
+            sline  = Q6_V_valign_VVR(slinec, slinep, (size_t) src);
+            *((HVX_UVector *) (output_v_ptr++)) = Q6_Vsf_equals_Vqf32(Q6_Vqf32_vmpy_VsfVsf(sline, val_vec));
 
-            *(HVX_UVector *) (dst + i * SIZEOF_FP32) = Q6_Vsf_equals_Vqf32(out);
+            slinep = slinec;
+        }
+
+        if (leftover_size > 0) {
+            slinec = (is_in_one_chunk(input_v_ptr, leftover_size, VLEN) ? slinep : *input_v_ptr++);
+
+            sline = Q6_V_valign_VVR(slinec, slinep, (size_t) src);
+
+            HVX_Vector sout = Q6_Vsf_equals_Vqf32(Q6_Vqf32_vmpy_VsfVsf(sline, val_vec));
+            hvx_vec_store_u(output_v_ptr, leftover_size, sout);
+            handled_leftover = true;
         }
     }
 
-    if (left_over > 0) {
+    if (left_over > 0 && !handled_leftover) {
         const float * srcf = (const float *) src + num_elems_whole;
         float *       dstf = (float *) dst + num_elems_whole;
 
@@ -875,35 +948,45 @@ float hvx_self_max_f32(const uint8_t * restrict src, const int num_elems) {
 void hvx_min_scalar_f32(const uint8_t * restrict src, const float val, uint8_t * restrict dst, const int num_elems) {
     size_t left_over       = num_elems & (VLEN_FP32 - 1);
     size_t num_elems_whole = num_elems - left_over;
-
+    int unalign_address = 0;
     if ((0 == htp_is_aligned((void *) src, VLEN)) || (0 == htp_is_aligned((void *) dst, VLEN))) {
         FARF(HIGH, "hvx_min_scalar_f32: unaligned address in hvx op, possibly slower execution\n");
+        unalign_address = 1;
     }
 
-    assert((1 == htp_is_aligned((void *) src, VLEN)) || (0 == num_elems_whole));
-
     const float * src_f = (const float *) src;
 
-    HVX_Vector vec_min = Q6_V_vsplat_R(val);
+    HVX_Vector vec_min = hvx_vec_splat_fp32(val);
 
+    if(unalign_address == 0){
         HVX_Vector * restrict vec_in  = (HVX_Vector *) src;
         HVX_Vector * restrict vec_out = (HVX_Vector *) dst;
 
         #pragma unroll(4)
         for (int i = 0; i < num_elems_whole; i += VLEN_FP32) {
-        vec_min    = Q6_Vsf_vmin_VsfVsf(vec_min, *vec_in++);
-        *vec_out++ = Q6_Vsf_equals_Vqf32(vec_min);
+            HVX_Vector min_clamp    = Q6_Vsf_vmin_VsfVsf(vec_min, *vec_in++);
+            *vec_out++ = (min_clamp);
+        }
+    }else{
+        HVX_UVector * restrict vec_in  = (HVX_Vector *) src;
+        HVX_UVector * restrict vec_out = (HVX_Vector *) dst;
+
+        #pragma unroll(4)
+        for (int i = 0; i < num_elems_whole; i += VLEN_FP32) {
+            HVX_Vector min_clamp     = Q6_Vsf_vmin_VsfVsf(vec_min, *vec_in++);
+            *vec_out++ = (min_clamp);
+        }
     }
 
     if (left_over > 0 ) {
         const float * srcf = (const float *) src + num_elems_whole;
         float *       dstf = (float *) dst + num_elems_whole;
 
-        HVX_Vector in = *(HVX_UVector *) srcf;
+        HVX_UVector in = *(HVX_UVector *) srcf;
 
-        vec_min = Q6_Vsf_vmin_VsfVsf(vec_min, in);
+        HVX_UVector min_clamp = Q6_Vsf_vmin_VsfVsf(vec_min, in);
 
-        hvx_vec_store_u((void *) dstf, left_over * SIZEOF_FP32, Q6_Vsf_equals_Vqf32(vec_min));
+        hvx_vec_store_u((void *) dstf, left_over * SIZEOF_FP32, (min_clamp));
     }
 }
 
@@ -915,18 +998,21 @@ void hvx_clamp_scalar_f32(const uint8_t * restrict src,
     size_t left_over       = num_elems & (VLEN_FP32 - 1);
     size_t num_elems_whole = num_elems - left_over;
 
+    int unalign_address = 0;
     if ((0 == htp_is_aligned((void *) src, VLEN)) || (0 == htp_is_aligned((void *) dst, VLEN))) {
         FARF(HIGH, "hvx_clamp_scalar_f32: unaligned address in hvx op, possibly slower execution\n");
+        unalign_address = 1;
     }
 
-    assert((1 == htp_is_aligned((void *) src, VLEN)) || (0 == num_elems_whole));
-
-    HVX_Vector * restrict vec_in  = (HVX_Vector *) src;
-    HVX_Vector * restrict vec_out = (HVX_Vector *) dst;
-
     HVX_Vector range_left  = hvx_vec_splat_fp32(limit_left);
     HVX_Vector range_right = hvx_vec_splat_fp32(limit_right);
 
+    if(unalign_address == 0){
+        HVX_Vector * restrict vec_in  = (HVX_Vector *) src;
+        HVX_Vector * restrict vec_out = (HVX_Vector *) dst;
+
+
+
         #pragma unroll(4)
         for (int i = 0; i < num_elems_whole; i += VLEN_FP32) {
             HVX_Vector in_vec = *vec_in++;
@@ -936,25 +1022,46 @@ void hvx_clamp_scalar_f32(const uint8_t * restrict src,
             HVX_VectorPred pred_cap_left  = Q6_Q_vcmp_gt_VsfVsf(range_left, in_vec);
 
             in_vec = Q6_V_vmux_QVV(pred_cap_right, range_right, temp_v);
-        in_vec = Q6_V_vmux_QVV(pred_cap_left, range_left, temp_v);
+            in_vec = Q6_V_vmux_QVV(pred_cap_left, range_left, in_vec);
+
+            *vec_out++ = in_vec;
+        }
+
+    }else{
+
+        HVX_UVector * restrict vec_in  = (HVX_UVector *) src;
+        HVX_UVector * restrict vec_out = (HVX_UVector *) dst;
+
+        #pragma unroll(4)
+        for (int i = 0; i < num_elems_whole; i += VLEN_FP32) {
+            HVX_Vector in_vec = *vec_in++;
+            HVX_Vector temp_v = in_vec;
+
+            HVX_VectorPred pred_cap_right = Q6_Q_vcmp_gt_VsfVsf(in_vec, range_right);
+            HVX_VectorPred pred_cap_left  = Q6_Q_vcmp_gt_VsfVsf(range_left, in_vec);
+
+            in_vec = Q6_V_vmux_QVV(pred_cap_right, range_right, temp_v);
+            in_vec = Q6_V_vmux_QVV(pred_cap_left, range_left, in_vec);
+
+            *vec_out++ = in_vec;
+        }
 
-        *vec_out++ = Q6_Vsf_equals_Vqf32(in_vec);
     }
 
     if (left_over > 0) {
         const float * srcf = (const float *) src + num_elems_whole;
         float *       dstf = (float *) dst + num_elems_whole;
 
-        HVX_Vector in = *(HVX_UVector *) srcf;
+        HVX_Vector in_vec = *(HVX_UVector *) srcf;
 
-        HVX_Vector temp_v = in;
+        HVX_Vector temp_v = in_vec;
 
-        HVX_VectorPred pred_cap_right = Q6_Q_vcmp_gt_VsfVsf(in, range_right);
-        HVX_VectorPred pred_cap_left  = Q6_Q_vcmp_gt_VsfVsf(range_left, in);
+        HVX_VectorPred pred_cap_right = Q6_Q_vcmp_gt_VsfVsf(in_vec, range_right);
+        HVX_VectorPred pred_cap_left  = Q6_Q_vcmp_gt_VsfVsf(range_left, in_vec);
 
-        in = Q6_V_vmux_QVV(pred_cap_right, range_right, temp_v);
-        in = Q6_V_vmux_QVV(pred_cap_left, range_left, temp_v);
+        in_vec = Q6_V_vmux_QVV(pred_cap_right, range_right, temp_v);
+        in_vec = Q6_V_vmux_QVV(pred_cap_left, range_left, in_vec);
 
-        hvx_vec_store_u((void *) dstf, left_over * SIZEOF_FP32, Q6_Vsf_equals_Vqf32(in));
+        hvx_vec_store_u((void *) dstf, left_over * SIZEOF_FP32, in_vec);
     }
 }

ggml/src/ggml-hexagon/htp/hvx-utils.h

@@ -265,12 +265,16 @@ static inline void hvx_bcast_fp32_a(uint8_t * restrict dst, float elem, uint32_t
     }
 }
 
+
+/* Return whether 'n' elements from vector are in the one chunk of 'chunk_size'. */
 static __attribute__((always_inline)) int32_t is_in_one_chunk(void * addr, uint32_t n, uint32_t chunk_size) {
     uint32_t left_off  = (size_t) addr & (chunk_size - 1);
     uint32_t right_off = left_off + n;
     return right_off <= chunk_size;
 }
 
+
+
 static void hvx_vec_dump_fp16_n(char * pref, HVX_Vector v, uint32_t n) {
     HVX_VectorAlias u = { .v = v };
 
@@ -994,6 +998,59 @@ static inline void hvx_fast_sigmoid_f32(const uint8_t * restrict src, uint8_t *
     }
 }
 
+
+static inline void hvx_sigmoid_f32(const uint8_t * restrict src, uint8_t * restrict dst, const int num_elems){
+    int step_of_1 = num_elems >> 5;  // divby 32, because 32 float = 128 bytes per HVX vector
+    int leftover = num_elems - (step_of_1 * VLEN_FP32);
+
+    int32_t leftover_size = leftover * sizeof(float);
+
+    static const float kMinExp = -87.f;  // 0
+    static const float kMaxExp = 87.f;   // 1
+
+    const HVX_Vector one     = hvx_vec_splat_fp32(1.f);
+    const HVX_Vector max_exp = hvx_vec_splat_fp32(kMaxExp);
+    const HVX_Vector min_exp = hvx_vec_splat_fp32(kMinExp);
+
+    const float *input = (float *)src;
+    float *output = (float *)dst;
+
+    HVX_Vector *  input_v_ptr  = (HVX_Vector *) input;
+    HVX_UVector * output_v_ptr = (HVX_UVector *) output;
+
+    HVX_Vector slinep;
+    HVX_Vector slinec;
+    HVX_Vector sline;
+
+    slinep = *input_v_ptr++;
+    #pragma unroll(4)
+    for (int i = step_of_1 - 1; i > 0; i--) {
+        slinec                              = *input_v_ptr++;
+        sline                               = Q6_V_valign_VVR(slinec, slinep, (size_t) input);
+        *((HVX_UVector *) (output_v_ptr++)) = hvx_vec_fast_sigmoid_fp32_guard(sline, one, max_exp, min_exp);
+        /* Prepare slinep for next iteration */
+        slinep                              = slinec;
+    }
+
+    if (step_of_1 > 0) {
+        slinec = htp_is_aligned(input_v_ptr, 128) && leftover == 0 ? slinep : *input_v_ptr++;
+        sline  = Q6_V_valign_VVR(slinec, slinep, (size_t) input);
+        *((HVX_UVector *) (output_v_ptr++)) = hvx_vec_fast_sigmoid_fp32_guard(sline, one, max_exp, min_exp);
+        ;
+
+        slinep = slinec;
+    }
+    if (leftover > 0) {
+        slinec = (is_in_one_chunk(input_v_ptr, leftover_size, 128) ? slinep : *input_v_ptr++);
+
+        sline = Q6_V_valign_VVR(slinec, slinep, (size_t) input);
+
+        HVX_Vector sout = hvx_vec_fast_sigmoid_fp32_guard(sline, one, max_exp, min_exp);
+        hvx_vec_store_u(output_v_ptr, leftover_size, sout);
+    }
+}
+
+
 float hvx_sum_of_squares_f32(const uint8_t * restrict src, const int num_elems);
 void  hvx_mul_f32(const uint8_t * restrict src0,
                   const uint8_t * restrict src1,

ggml/src/ggml-hexagon/htp/main.c

@@ -798,6 +798,7 @@ static void htp_packet_callback(dspqueue_t queue, int error, void * context) {
                 break;
 
             case HTP_OP_UNARY_SILU:
+            case HTP_OP_UNARY_GELU:
                 if (n_bufs != 2) {
                     FARF(ERROR, "Bad act-req buffer list");
                     continue;
@@ -806,6 +807,7 @@ static void htp_packet_callback(dspqueue_t queue, int error, void * context) {
                 break;
 
             case HTP_OP_GLU_SWIGLU:
+            case HTP_OP_GLU_SWIGLU_OAI:
             case HTP_OP_SOFTMAX:
                 if ((n_bufs != 2) && (n_bufs != 3)) {
                     FARF(ERROR, "Bad act-req buffer list");

ggml/src/ggml-hexagon/htp/matmul-ops.c

@@ -903,7 +903,7 @@ static void vec_dot_f16_f32(const int n, float * restrict s, const void * restri
         const float * restrict vy  = (const float * restrict) y;
 
         for (uint32_t i = 0; i < n; i++) {
-            rsum += vx[i] * (__fp16) vy[i];
+            rsum += (float)vx[i] * vy[i];
         }
         *s = rsum;
         return;
@@ -917,7 +917,7 @@ static void vec_dot_f16_f32(const int n, float * restrict s, const void * restri
 
     // for some reason we need volatile here so that the compiler doesn't try anything funky
     volatile HVX_Vector rsum = Q6_V_vsplat_R(0);
-
+    float r_sum_scalar = 0.0f;
     uint32_t i = 0;
 
     for (i = 0; i < nv0; i++) {
@@ -926,31 +926,42 @@ static void vec_dot_f16_f32(const int n, float * restrict s, const void * restri
         HVX_Vector     x  = vx[i];
         HVX_VectorPair xp = Q6_Wqf32_vmpy_VhfVhf(Q6_Vh_vshuff_Vh(x), Q6_Vh_vsplat_R(0x3C00));  // mul by 1.0
 
-        HVX_Vector hi = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_hi_W(xp)), Q6_V_hi_W(yp));
-        HVX_Vector lo = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_lo_W(xp)), Q6_V_lo_W(yp));
+        //NOTE: need volatile here to prevent compiler optimization
+        // Seem compiler cannot guarantee read-after-write??
+        volatile HVX_Vector hi = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_hi_W(xp)), Q6_V_hi_W(yp));
+        volatile HVX_Vector lo = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_lo_W(xp)), Q6_V_lo_W(yp));
 
         HVX_Vector sum = Q6_Vqf32_vadd_Vqf32Vqf32(hi, lo);
         rsum           = Q6_Vqf32_vadd_Vqf32Vqf32(rsum, sum);
     }
 
     if (nv1) {
-        HVX_VectorPair yp = vy[i];
+        // HVX_VectorPair yp = vy[i];
 
-        HVX_Vector     x  = vx[i];
-        HVX_VectorPair xp = Q6_Wqf32_vmpy_VhfVhf(Q6_Vh_vshuff_Vh(x), Q6_Vh_vsplat_R(0x3C00));  // mul by 1.0
+        // HVX_Vector     x  = vx[i];
+        // HVX_VectorPair xp = Q6_Wqf32_vmpy_VhfVhf(Q6_Vh_vshuff_Vh(x), Q6_Vh_vsplat_R(0x3C00));  // mul by 1.0
 
-        if (nv1 >= 32) {
-            HVX_Vector hi = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_hi_W(xp)), Q6_V_hi_W(yp));
-            rsum          = Q6_Vqf32_vadd_Vqf32Vqf32(rsum, hi);
-            nv1 -= 32;
-        }
+        // if (nv1 >= 32) {
+        //     volatile HVX_Vector hi = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_hi_W(xp)), Q6_V_hi_W(yp));
+        //     rsum          = Q6_Vqf32_vadd_Vqf32Vqf32(rsum, hi);
+        //     nv1 -= 32;
+        // }
 
+        // rsum = hvx_vec_qf32_reduce_sum(rsum);
+
+        // if (nv1) {
+        //     volatile HVX_Vector lo  = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_lo_W(xp)), Q6_V_lo_W(yp));
+        //     HVX_Vector sum = hvx_vec_qf32_reduce_sum_n(lo, nv1);
+        //     rsum           = Q6_Vqf32_vadd_Vqf32Vqf32(rsum, sum);
+        // }
+
+        //process the remainder using scalar loop
         rsum = hvx_vec_qf32_reduce_sum(rsum);
+        const __fp16 * restrict sx = (const __fp16 * restrict) x;
+        const float * restrict sy  = (const float * restrict) y;
 
-        if (nv1) {
-            HVX_Vector lo  = Q6_Vqf32_vmpy_VsfVsf(Q6_Vsf_equals_Vqf32(Q6_V_lo_W(xp)), Q6_V_lo_W(yp));
-            HVX_Vector sum = hvx_vec_qf32_reduce_sum_n(lo, nv1);
-            rsum           = Q6_Vqf32_vadd_Vqf32Vqf32(rsum, sum);
+        for (uint32_t i = nv0 * 64; i < n; i++) {
+            r_sum_scalar += (float) sx[i] * sy[i];
         }
 
         // hvx_vec_dump_fp16("X", x);
@@ -961,7 +972,7 @@ static void vec_dot_f16_f32(const int n, float * restrict s, const void * restri
         rsum = hvx_vec_qf32_reduce_sum(rsum);
     }
 
-    *s = hvx_vec_get_fp32(Q6_Vsf_equals_Vqf32(rsum));
+    *s = hvx_vec_get_fp32(Q6_Vsf_equals_Vqf32(rsum)) + r_sum_scalar;
 
 #    ifdef HTP_DEBUG
     {
@@ -1498,9 +1509,6 @@ static void matmul_f16_f32(struct htp_tensor * restrict src0,
     uint64_t t1, t2;
     t1 = HAP_perf_get_qtimer_count();
 
-    const size_t src0_row_size = sizeof(__fp16) * ne00;
-    const size_t src1_row_size = sizeof(float) * ne10;
-
     assert(ne12 % ne02 == 0);
     assert(ne13 % ne03 == 0);
 
@@ -1510,8 +1518,6 @@ static void matmul_f16_f32(struct htp_tensor * restrict src0,
     // This is the size of the rest of the dimensions of the result
     const uint32_t nr1 = ne1 * ne2 * ne3;
 
-    uint32_t chunk_size = 64;
-
     // distribute the thread work across the inner or outer loop based on which one is larger
     uint32_t nchunk0 = nr0 > nr1 ? nth : 1;  // parallelize by src0 rows
     uint32_t nchunk1 = nr0 > nr1 ? 1 : nth;  // parallelize by src1 rows
@@ -1544,11 +1550,11 @@ static void matmul_f16_f32(struct htp_tensor * restrict src0,
     const uint32_t blck_0 = 64;
     const uint32_t blck_1 = 64;
 
-    float tmp[32];
+    __attribute__((aligned(128))) float tmp[64];
 
     for (uint32_t iir1 = ir1_start; iir1 < ir1_end; iir1 += blck_1) {
         for (uint32_t iir0 = ir0_start; iir0 < ir0_end; iir0 += blck_0) {
-            for (uint32_t ir1 = iir1; ir1 < iir1 + blck_1 && ir1 < ir1_end; ir1++) {
+            for (uint32_t ir1 = iir1; ir1 < MIN(iir1 + blck_1, ir1_end); ir1++) {
                 const uint32_t i13 = (ir1 / (ne12 * ne1));
                 const uint32_t i12 = (ir1 - i13 * ne12 * ne1) / ne1;
                 const uint32_t i11 = (ir1 - i13 * ne12 * ne1 - i12 * ne1);
@@ -1561,13 +1567,16 @@ static void matmul_f16_f32(struct htp_tensor * restrict src0,
                 const uint32_t i2 = i12;
                 const uint32_t i3 = i13;
 
-                const uint8_t * restrict src0_row = (const uint8_t *) src0->data + (0 + i02 * nb02 + i03 * nb03);
+                const uint8_t * restrict src0_base = (const uint8_t *) src0->data + (0 + i02 * nb02 + i03 * nb03);
                 const uint8_t * restrict src1_col =
-                    (const uint8_t *) src1->data + (i11 + i12 * ne11 + i13 * ne12 * ne11) * src1_row_size;
+                    (const uint8_t *) src1->data + (i11 * nb11 + i12 * nb12 + i13 * nb13);
                 float * dst_col = (float *) ((uint8_t * restrict) dst->data + (i1 * nb1 + i2 * nb2 + i3 * nb3));
 
-                for (uint32_t ir0 = iir0; ir0 < iir0 + blck_0 && ir0 < ir0_end; ir0++) {
-                    vec_dot_f16_f32(ne00, &tmp[ir0 - iir0], src0_row + ir0 * src0_row_size, src1_col);
+                const uint32_t ir0_block_end = MIN(iir0 + blck_0, ir0_end);
+                for (uint32_t ir0 = iir0; ir0 < ir0_block_end; ir0++) {
+                    // Use nb01 stride for non-contiguous src0 support
+                    const uint8_t * restrict src0_row = src0_base + ir0 * nb01;
+                    vec_dot_f16_f32(ne00, &tmp[ir0 - iir0], src0_row, src1_col);
                 }
 
                 hvx_copy_fp32_ua((uint8_t *) &dst_col[iir0], (uint8_t *) tmp, MIN(iir0 + blck_0, ir0_end) - iir0);

gguf-py/gguf/constants.py

@@ -413,6 +413,7 @@ class MODEL_ARCH(IntEnum):
     JAIS             = auto()
     NEMOTRON         = auto()
     NEMOTRON_H       = auto()
+    NEMOTRON_H_MOE   = auto()
     EXAONE           = auto()
     EXAONE4          = auto()
     GRANITE          = auto()
@@ -642,6 +643,7 @@ class MODEL_TENSOR(IntEnum):
     V_MMPROJ_PEG         = auto()
     V_ENC_EMBD_CLS       = auto()
     V_ENC_EMBD_PATCH     = auto()
+    V_ENC_EMBD_NORM      = auto()
     V_ENC_EMBD_POS       = auto()
     V_ENC_INPUT_NORM     = auto()
     V_ENC_ATTN_QKV       = auto()
@@ -660,6 +662,7 @@ class MODEL_TENSOR(IntEnum):
     V_LAYER_SCALE_2      = auto()
     V_PRE_NORM           = auto()
     V_POST_NORM          = auto()
+    V_MM_POST_NORM       = auto()
     V_MM_INP_NORM        = auto()
     V_MM_INP_PROJ        = auto() # gemma3
     V_MM_SOFT_EMB_NORM   = auto() # gemma3
@@ -786,6 +789,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
     MODEL_ARCH.JAIS:             "jais",
     MODEL_ARCH.NEMOTRON:         "nemotron",
     MODEL_ARCH.NEMOTRON_H:       "nemotron_h",
+    MODEL_ARCH.NEMOTRON_H_MOE:   "nemotron_h_moe",
     MODEL_ARCH.EXAONE:           "exaone",
     MODEL_ARCH.EXAONE4:          "exaone4",
     MODEL_ARCH.GRANITE:          "granite",
@@ -1014,6 +1018,7 @@ TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
     MODEL_TENSOR.V_MMPROJ_PEG:              "mm.model.peg.{bid}",
     MODEL_TENSOR.V_ENC_EMBD_CLS:            "v.class_embd",
     MODEL_TENSOR.V_ENC_EMBD_PATCH:          "v.patch_embd",
+    MODEL_TENSOR.V_ENC_EMBD_NORM:           "v.norm_embd",
     MODEL_TENSOR.V_ENC_EMBD_POS:            "v.position_embd",
     MODEL_TENSOR.V_ENC_ATTN_QKV:            "v.blk.{bid}.attn_qkv",
     MODEL_TENSOR.V_ENC_ATTN_Q:              "v.blk.{bid}.attn_q",
@@ -1032,6 +1037,7 @@ TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
     MODEL_TENSOR.V_LAYER_SCALE_2:           "v.blk.{bid}.ls2",
     MODEL_TENSOR.V_PRE_NORM:                "v.pre_ln",
     MODEL_TENSOR.V_POST_NORM:               "v.post_ln",
+    MODEL_TENSOR.V_MM_POST_NORM:            "mm.post_norm",
     MODEL_TENSOR.V_MM_INP_PROJ:             "mm.input_projection",
     MODEL_TENSOR.V_MM_INP_NORM:             "mm.input_norm",
     MODEL_TENSOR.V_MM_SOFT_EMB_NORM:        "mm.soft_emb_norm",
@@ -1092,6 +1098,7 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
         MODEL_TENSOR.V_MMPROJ_PEG,
         MODEL_TENSOR.V_ENC_EMBD_CLS,
         MODEL_TENSOR.V_ENC_EMBD_PATCH,
+        MODEL_TENSOR.V_ENC_EMBD_NORM,
         MODEL_TENSOR.V_ENC_EMBD_POS,
         MODEL_TENSOR.V_ENC_INPUT_NORM,
         MODEL_TENSOR.V_ENC_ATTN_QKV,
@@ -1110,6 +1117,7 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
         MODEL_TENSOR.V_LAYER_SCALE_2,
         MODEL_TENSOR.V_PRE_NORM,
         MODEL_TENSOR.V_POST_NORM,
+        MODEL_TENSOR.V_MM_POST_NORM,
         MODEL_TENSOR.V_MM_INP_PROJ,
         MODEL_TENSOR.V_MM_INP_NORM,
         MODEL_TENSOR.V_MM_SOFT_EMB_NORM,
@@ -2529,6 +2537,33 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
         MODEL_TENSOR.FFN_DOWN,
         MODEL_TENSOR.FFN_UP,
     ],
+    MODEL_ARCH.NEMOTRON_H_MOE: [
+        MODEL_TENSOR.TOKEN_EMBD,
+        MODEL_TENSOR.OUTPUT_NORM,
+        MODEL_TENSOR.OUTPUT,
+        MODEL_TENSOR.ATTN_NORM,
+        MODEL_TENSOR.SSM_IN,
+        MODEL_TENSOR.SSM_CONV1D,
+        MODEL_TENSOR.SSM_DT,
+        MODEL_TENSOR.SSM_A,
+        MODEL_TENSOR.SSM_D,
+        MODEL_TENSOR.SSM_NORM,
+        MODEL_TENSOR.SSM_OUT,
+        MODEL_TENSOR.ATTN_Q,
+        MODEL_TENSOR.ATTN_K,
+        MODEL_TENSOR.ATTN_V,
+        MODEL_TENSOR.ATTN_OUT,
+        MODEL_TENSOR.FFN_DOWN,
+        MODEL_TENSOR.FFN_UP,
+        # experts
+        MODEL_TENSOR.FFN_GATE_INP,
+        MODEL_TENSOR.FFN_UP_EXP,
+        MODEL_TENSOR.FFN_DOWN_EXP,
+        # shared expert
+        MODEL_TENSOR.FFN_DOWN_SHEXP,
+        MODEL_TENSOR.FFN_UP_SHEXP,
+        MODEL_TENSOR.FFN_EXP_PROBS_B,
+    ],
     MODEL_ARCH.EXAONE: [
         MODEL_TENSOR.TOKEN_EMBD,
         MODEL_TENSOR.OUTPUT_NORM,
@@ -3328,6 +3363,7 @@ class VisionProjectorType:
     LIGHTONOCR = "lightonocr"
     COGVLM = "cogvlm"
     JANUS_PRO = "janus_pro"
+    GLM4V = "glm4v"
 
 
 # Items here are (block size, type size)

gguf-py/gguf/tensor_mapping.py

@@ -379,6 +379,7 @@ class TensorNameMap:
             "model.layers.{bid}.feed_forward.gate",               # lfm2moe
             "model.layers.{bid}.mlp.router.gate",               # afmoe
             "layers.{bid}.gate",                                # mistral-large
+            "backbone.layers.{bid}.mixer.gate",                 # nemotron-h-moe
         ),
 
         MODEL_TENSOR.FFN_GATE_INP_SHEXP: (
@@ -392,6 +393,7 @@ class TensorNameMap:
             "model.layers.{bid}.mlp.expert_bias",                           # afmoe
             "model.layers.{bid}.feed_forward.expert_bias",                  # lfm2moe
             "model.layers.{bid}.block_sparse_moe.e_score_correction",       # minimax-m2
+            "backbone.layers.{bid}.mixer.gate.e_score_correction"           # nemotron-h-moe
         ),
 
         # Feed-forward up
@@ -440,7 +442,7 @@ class TensorNameMap:
             "layers.{bid}.feed_forward.experts.w3",                 # mixtral (merged)
             "transformer.decoder_layer.{bid}.moe.linear_v",         # Grok (merged)
             "transformer.blocks.{bid}.ffn.experts.mlp.v1",          # dbrx
-            "model.layers.{bid}.mlp.experts.up_proj",               # qwen2moe olmoe (merged) ernie4.5-moe
+            "model.layers.{bid}.mlp.experts.up_proj",               # qwen2moe olmoe (merged) ernie4.5-moe, nemotron-h-moe (merged)
             "model.layers.{bid}.block_sparse_moe.experts.w3",       # phimoe (merged)
             "model.layers.{bid}.feed_forward.experts.up_proj",      # llama4
             "encoder.layers.{bid}.mlp.experts.mlp.w1",              # nomic-bert-moe
@@ -454,6 +456,7 @@ class TensorNameMap:
             "model.layers.{bid}.feed_forward.down_proj",
             "model.layers.{bid}.mlp.shared_mlp.up_proj",             # hunyuan
             "layers.{bid}.shared_experts.w3",                        # mistral-large
+            "backbone.layers.{bid}.mixer.shared_experts.up_proj",    # nemotron-h-moe
         ),
 
         MODEL_TENSOR.FFN_UP_CHEXP: (
@@ -548,7 +551,7 @@ class TensorNameMap:
             "layers.{bid}.feed_forward.experts.w2",                 # mixtral (merged)
             "transformer.decoder_layer.{bid}.moe.linear_1",         # Grok (merged)
             "transformer.blocks.{bid}.ffn.experts.mlp.w2",          # dbrx
-            "model.layers.{bid}.mlp.experts.down_proj",             # qwen2moe olmoe (merged) ernie4.5-moe
+            "model.layers.{bid}.mlp.experts.down_proj",             # qwen2moe olmoe (merged) ernie4.5-moe nemotron-h-moe (merged)
             "model.layers.{bid}.block_sparse_moe.output_linear",    # granitemoe
             "model.layers.{bid}.block_sparse_moe.experts.w2",       # phimoe (merged)
             "model.layers.{bid}.feed_forward.experts.down_proj",    # llama4
@@ -563,6 +566,7 @@ class TensorNameMap:
             "model.layers.{bid}.shared_mlp.output_linear",             # granitemoe
             "model.layers.{bid}.mlp.shared_mlp.down_proj",             # hunyuan
             "layers.{bid}.shared_experts.w2",                          # mistral-large
+            "backbone.layers.{bid}.mixer.shared_experts.down_proj",    # nemotron-h-moe
         ),
 
         MODEL_TENSOR.FFN_DOWN_CHEXP: (
@@ -706,6 +710,7 @@ class TensorNameMap:
             "model.layers.{bid}.mamba.dt_proj",         # jamba falcon-h1 granite-hybrid
             "model.layers.layers.{bid}.mixer.dt_proj",  # plamo2
             "model.layers.{bid}.linear_attn.dt_proj",   # qwen3next
+            "backbone.layers.{bid}.mixer.dt",           # nemotron-h-moe
         ),
 
         MODEL_TENSOR.SSM_DT_NORM: (
@@ -1207,6 +1212,7 @@ class TensorNameMap:
         MODEL_TENSOR.V_MMPROJ_FC: (
             "model.connector.modality_projection.proj", # SmolVLM
             "model.vision.linear_proj.linear_proj", # cogvlm
+            "visual.merger.proj", # glm4v
         ),
 
         MODEL_TENSOR.V_MMPROJ_MLP: (
@@ -1240,6 +1246,10 @@ class TensorNameMap:
             "model.vision.patch_embedding.proj", # cogvlm
         ),
 
+        MODEL_TENSOR.V_ENC_EMBD_NORM: (
+            "visual.post_conv_layernorm", # glm4v
+        ),
+
         MODEL_TENSOR.V_ENC_EMBD_POS: (
             "vision_tower.vision_model.embeddings.position_embedding",
             "model.vision_tower.embeddings.position_embeddings", # Intern-S1
@@ -1249,6 +1259,7 @@ class TensorNameMap:
             "vision_tower.patch_embed.pos_emb", # kimi-vl
             "visual.pos_embed", # qwen3vl
             "model.vision.patch_embedding.position_embedding", # cogvlm
+            "visual.embeddings.position_embedding", # glm4v
         ),
 
         MODEL_TENSOR.V_ENC_ATTN_QKV: (
@@ -1404,6 +1415,11 @@ class TensorNameMap:
             "vision_model.layernorm_post", # llama4
             "visual.merger.ln_q", # qwen2vl
             "vision_tower.encoder.final_layernorm", # kimi-vl
+            "visual.post_layernorm", # glm4v
+        ),
+
+        MODEL_TENSOR.V_MM_POST_NORM: (
+            "visual.merger.post_projection_norm", # glm4v
         ),
 
         MODEL_TENSOR.V_MM_INP_PROJ: (
@@ -1473,6 +1489,7 @@ class TensorNameMap:
         MODEL_TENSOR.V_MM_PATCH_MERGER: (
             "multi_modal_projector.patch_merger.merging_layer", # mistral small 3.1 - hf
             "patch_merger.merging_layer", # mistral
+            "visual.downsample", # glm4v
         ),
 
         MODEL_TENSOR.V_DS_NORM: (
@@ -1493,14 +1510,17 @@ class TensorNameMap:
 
         MODEL_TENSOR.V_MM_UP: (
             "model.vision.linear_proj.dense_h_to_4h", # cogvlm
+            "visual.merger.up_proj", # glm4v
         ),
 
         MODEL_TENSOR.V_MM_DOWN: (
             "model.vision.linear_proj.dense_4h_to_h", # cogvlm
+            "visual.merger.down_proj", # glm4v
         ),
 
         MODEL_TENSOR.V_MM_GATE: (
             "model.vision.linear_proj.gate_proj", # cogvlm
+            "visual.merger.gate_proj", # glm4v
         ),
 
         MODEL_TENSOR.V_TOK_BOI: (

gguf-py/gguf/utility.py

@@ -288,7 +288,7 @@ class LocalTensor:
     data_range: LocalTensorRange
 
     def mmap_bytes(self) -> np.ndarray:
-        return np.memmap(self.data_range.filename, offset=self.data_range.offset, shape=self.data_range.size)
+        return np.memmap(self.data_range.filename, mode='r', offset=self.data_range.offset, shape=self.data_range.size)
 
 
 class SafetensorsLocal:

models/templates/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16.jinja

@@ -0,0 +1,204 @@
+{% macro render_extra_keys(json_dict, handled_keys) %}
+    {%- if json_dict is mapping %}
+        {%- for json_key in json_dict if json_key not in handled_keys %}
+            {%- if json_dict[json_key] is mapping or (json_dict[json_key] is sequence and json_dict[json_key] is not string) %}
+                {{- '\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | tojson | safe) ~ '</' ~ json_key ~ '>' }}
+            {%- else %}
+                {{-'\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | string) ~ '</' ~ json_key ~ '>' }}
+            {%- endif %}
+        {%- endfor %}
+    {%- endif %}
+{% endmacro %}
+{%- set enable_thinking = enable_thinking if enable_thinking is defined else True %}
+{%- set truncate_history_thinking = truncate_history_thinking if truncate_history_thinking is defined else True %}
+
+{%- set ns = namespace(last_user_idx = -1) %}
+{%- set loop_messages = messages %}
+{%- for m in loop_messages %}
+  {%- if m["role"] == "user" %}
+    {%- set ns.last_user_idx = loop.index0 %}
+  {%- endif %}
+{%- endfor %}
+
+{%- if messages[0]["role"] == "system" %}
+    {%- set system_message = messages[0]["content"] %}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set system_message = "" %}
+    {%- set loop_messages = messages %}
+{%- endif %}
+{%- if not tools is defined %}
+    {%- set tools = [] %}
+{%- endif %}
+{# Recompute last_user_idx relative to loop_messages after handling system #}
+{%- set ns = namespace(last_user_idx = -1) %}
+{%- for m in loop_messages %}
+  {%- if m["role"] == "user" %}
+    {%- set ns.last_user_idx = loop.index0 %}
+  {%- endif %}
+{%- endfor %}
+{%- if system_message is defined %}
+    {{- "<|im_start|>system\n" + system_message }}
+{%- else %}
+    {%- if tools is iterable and tools | length > 0 %}
+        {{- "<|im_start|>system\n" }}
+    {%- endif %}
+{%- endif %}
+{%- if tools is iterable and tools | length > 0 %}
+    {%- if system_message is defined and system_message | length > 0 %}
+        {{- "\n\n" }}
+    {%- endif %}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n" }}
+    {{- "<tools>" }}
+    {%- for tool in tools %}
+        {%- if tool.function is defined %}
+            {%- set tool = tool.function %}
+        {%- endif %}
+        {{- "\n<function>\n<name>" ~ tool.name ~ "</name>" }}
+        {%- if tool.description is defined %}
+            {{- '\n<description>' ~ (tool.description | trim) ~ '</description>' }}
+        {%- endif %}
+        {{- '\n<parameters>' }}
+        {%- if tool.parameters is defined and tool.parameters is mapping and tool.parameters.properties is defined and tool.parameters.properties is mapping %}
+            {%- for param_name, param_fields in tool.parameters.properties|items %}
+                {{- '\n<parameter>' }}
+                {{- '\n<name>' ~ param_name ~ '</name>' }}
+                {%- if param_fields.type is defined %}
+                    {{- '\n<type>' ~ (param_fields.type | string) ~ '</type>' }}
+                {%- endif %}
+                {%- if param_fields.description is defined %}
+                    {{- '\n<description>' ~ (param_fields.description | trim) ~ '</description>' }}
+                {%- endif %}
+                {%- if param_fields.enum is defined %}
+                    {{- '\n<enum>' ~ (param_fields.enum | tojson | safe) ~ '</enum>' }}
+                {%- endif %}
+                {%- set handled_keys = ['name', 'type', 'description', 'enum'] %}
+                {{- render_extra_keys(param_fields, handled_keys) }}
+                {{- '\n</parameter>' }}
+            {%- endfor %}
+        {%- endif %}
+        {% set handled_keys = ['type', 'properties', 'required'] %}
+        {{- render_extra_keys(tool.parameters, handled_keys) }}
+        {%- if tool.parameters is defined and tool.parameters.required is defined %}
+            {{- '\n<required>' ~ (tool.parameters.required | tojson | safe) ~ '</required>' }}
+        {%- endif %}
+        {{- '\n</parameters>' }}
+        {%- set handled_keys = ['type', 'name', 'description', 'parameters'] %}
+        {{- render_extra_keys(tool, handled_keys) }}
+        {{- '\n</function>' }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+{%- endif %}
+
+
+{%- if system_message is defined %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if tools is iterable and tools | length > 0 %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+
+{%- for message in loop_messages %}
+    {%- if message.role == "assistant" %}
+        {# Add reasoning content in to content field for unified processing below. #}
+        {%- if message.reasoning_content is defined and message.reasoning_content is string and message.reasoning_content | trim | length > 0 %}
+            {%- set content = "<think>\n" ~ message.reasoning_content ~ "\n</think>\n" ~ (message.content | default('', true)) %}
+        {%- else %}
+            {%- set content = message.content | default('', true) %}
+            {%- if content is string -%}
+                {# Allow downstream logic to to take care of broken thought, only handle coherent reasoning here. #}
+                {%- if '<think>' not in content and '</think>' not in content -%}
+                    {%- set content = "<think></think>" ~ content -%}
+                {%- endif -%}
+            {%- else -%}
+                {%- set content = content -%}
+            {%- endif -%}
+        {%- endif %}
+        {%- if message.tool_calls is defined and message.tool_calls is iterable and message.tool_calls | length > 0 %}
+            {# Assistant message has tool calls. #}
+            {{- '<|im_start|>assistant\n' }}
+                {%- set include_content = not (truncate_history_thinking and loop.index0 < ns.last_user_idx) %}
+                {%- if content is string and content | trim | length > 0 %}
+                    {%- if include_content %}
+                        {{- (content | trim) ~ '\n' -}}
+                    {%- else %}
+                        {%- set c = (content | string) %}
+                        {%- if '</think>' in c %}
+                            {# Keep only content after the last closing think. Also generation prompt causes this. #}
+                            {%- set c = c.split('</think>')[-1] %}
+                        {%- elif '<think>' in c %}
+                            {# If <think> was opened but never closed, drop the trailing think segment #}
+                            {%- set c = c.split('<think>')[0] %}
+                        {%- endif %}
+                        {%- set c = "<think></think>" ~ c | trim %}
+                        {%- if c | length > 0 %}
+                            {{- c ~ '\n' -}}
+                        {%- endif %}
+                    {%- endif %}
+                {%- else %}
+                    {{- "<think></think>" -}}
+                {%- endif %}
+                {%- for tool_call in message.tool_calls %}
+                    {%- if tool_call.function is defined %}
+                        {%- set tool_call = tool_call.function %}
+                    {%- endif %}
+                    {{- '<tool_call>\n<function=' ~ tool_call.name ~ '>\n' -}}
+                        {%- if tool_call.arguments is defined %}
+                            {%- for args_name, args_value in tool_call.arguments|items %}
+                                {{- '<parameter=' ~ args_name ~ '>\n' -}}
+                                    {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                                {{- args_value ~ '\n</parameter>\n' -}}
+                            {%- endfor %}
+                        {%- endif %}
+                    {{- '</function>\n</tool_call>\n' -}}
+                {%- endfor %}
+                {{- '<|im_end|>\n' }}
+        {%- else %}
+            {# Assistant message doesn't have tool calls. #}
+            {%- if not (truncate_history_thinking and loop.index0 < ns.last_user_idx) %}
+                {{- '<|im_start|>assistant\n' ~ (content | default('', true) | string | trim) ~ '<|im_end|>\n' }}
+            {%- else %}
+                {%- set c = (content | default('', true) | string) %}
+                {%- if '<think>' in c and '</think>' in c %}
+                    {%- set c = "<think></think>" ~ c.split('</think>')[-1] %}
+                {%- endif %}
+                {%- set c = c | trim %}
+                {%- if c | length > 0 %}
+                    {{- '<|im_start|>assistant\n' ~ c ~ '<|im_end|>\n' }}
+                {%- else %}
+                    {{- '<|im_start|>assistant\n<|im_end|>\n' }}
+                {%- endif %}
+            {%- endif %}
+        {%- endif %}
+    {%- elif message.role == "user" or message.role == "system" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- set content = message.content | string %}
+        {{- content }}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user\n' }}
+        {%- endif %}
+        {{- '<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>\n' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endfor %}
+
+{%- if add_generation_prompt %}
+    {%- if enable_thinking %}
+        {{- '<|im_start|>assistant\n<think>\n' }}
+    {%- else %}
+        {{- '<|im_start|>assistant\n<think></think>' }}
+    {%- endif %}
+{%- endif %}

scripts/snapdragon/adb/run-mtmd.sh

@@ -0,0 +1,65 @@
+#!/bin/sh
+#
+
+# Basedir on device
+basedir=/data/local/tmp/llama.cpp
+
+cli_opts=
+
+branch=.
+[ "$B" != "" ] && branch=$B
+
+adbserial=
+[ "$S" != "" ] && adbserial="-s $S"
+
+model="gemma-3-4b-it-Q4_0.gguf"
+[ "$M" != "" ] && model="$M"
+
+mmproj="mmproj-F16.gguf"
+[ "$MMPROJ" != "" ] && mmproj="$MMPROJ"
+
+image=
+[ "$IMG" != "" ] && image="$IMG"
+
+device="HTP0"
+[ "$D" != "" ] && device="$D"
+
+verbose=
+[ "$V" != "" ] && verbose="GGML_HEXAGON_VERBOSE=$V"
+
+experimental="GGML_HEXAGON_EXPERIMENTAL=1"
+[ "$E" != "" ] && experimental="GGML_HEXAGON_EXPERIMENTAL=$E"
+
+sched=
+[ "$SCHED" != "" ] && sched="GGML_SCHED_DEBUG=2" cli_opts="$cli_opts -v"
+
+profile=
+[ "$PROF" != "" ] && profile="GGML_HEXAGON_PROFILE=$PROF GGML_HEXAGON_OPSYNC=1"
+
+opmask=
+[ "$OPMASK" != "" ] && opmask="GGML_HEXAGON_OPMASK=$OPMASK"
+
+nhvx=
+[ "$NHVX" != "" ] && nhvx="GGML_HEXAGON_NHVX=$NHVX"
+
+ndev=
+[ "$NDEV" != "" ] && ndev="GGML_HEXAGON_NDEV=$NDEV"
+
+# MTMD backend device for vision model (defaults to CPU if not set)
+mtmd_backend=
+[ "$MTMD_DEVICE" != "" ] && mtmd_backend="MTMD_BACKEND_DEVICE=$MTMD_DEVICE"
+
+set -x
+
+adb $adbserial shell " \
+  cd $basedir; ulimit -c unlimited;        \
+    LD_LIBRARY_PATH=$basedir/$branch/lib   \
+    ADSP_LIBRARY_PATH=$basedir/$branch/lib \
+    $verbose $experimental $sched $opmask $profile $nhvx $ndev $mtmd_backend       \
+      ./$branch/bin/llama-mtmd-cli --no-mmap -m $basedir/../gguf/$model   \
+         --mmproj $basedir/../gguf/$mmproj \
+         --image $basedir/../gguf/$image \
+         --poll 1000 -t 6 --cpu-mask 0xfc --cpu-strict 1             \
+         --ctx-size 8192 --batch-size 128 -ctk q8_0 -ctv q8_0 -fa on \
+         -ngl 99 --device $device -v $cli_opts $@ \
+"