llama: sync llama.cpp to commit 8962422

runner.go: Support GGUF LoRAs
The current CGo bindings for loading LoRAs only supports the older GGLA file format, which is no longer supported. This switches to use functions that load the newer GGUF LoRAs.
2024-09-03 21:23:46 -04:00 · 2024-09-03 21:15:14 -04:00 · 2024-09-03 21:15:14 -04:00 · 2024-09-03 21:15:14 -04:00 · 2024-09-03 21:15:14 -04:00 · 2024-09-03 21:15:14 -04:00
349 changed files with 189409 additions and 3607 deletions
--- a/.dockerignore
+++ b/.dockerignore
@@ -7,5 +7,3 @@ llm/llama.cpp
 .env
 .cache
 test_data
-llm/build
-llama/build
--- a/.gitattributes
+++ b/.gitattributes
@@ -1,3 +1,5 @@
 llm/ext_server/* linguist-vendored
+llama/**/*.{cpp,hpp,h,c,cu,cuh,m} linguist-vendored
+
 * text=auto
 *.go text eol=lf
--- a/.github/workflows/release.yaml
+++ b/.github/workflows/release.yaml
@@ -102,8 +102,7 @@ jobs:
        with:
          name: generate-windows-cpu
          path: |
-            build/**/*
-            build/**/*.a
+            llm/build/**/bin/*
            llm/build/**/*.a
            dist/windows-amd64/**

@@ -177,7 +176,7 @@ jobs:
        with:
          name: generate-windows-rocm
          path: |
-            build/**/*
+            llm/build/**/bin/*
            dist/windows-amd64/**
      - uses: actions/upload-artifact@v4
        with:
@@ -266,7 +265,7 @@ jobs:
        with:
          name: generate-windows-cuda-${{ matrix.cuda.version }}
          path: |
-            build/**/*
+            llm/build/**/bin/*
            dist/windows-amd64/**
      - uses: actions/upload-artifact@v4
        with:
@@ -274,134 +273,7 @@ jobs:
          path: dist/deps/*


-  # windows arm64 generate, go build, and zip file (no installer)
-  # Output of this build is aggregated into the final x86 build
-  # for a unified windows installer
-  windows-arm64:
-    runs-on: windows-arm64
-    environment: release
-    env:
-      KEY_CONTAINER: ${{ vars.KEY_CONTAINER }}
-    steps:
-      # The current Windows arm64 beta image has effectively zero dev tools installed...
-      - name: Install git and gzip
-        run: |
-          Set-ExecutionPolicy Bypass -Scope Process -Force
-          [System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072
-          iex ((New-Object System.Net.WebClient).DownloadString('https://community.chocolatey.org/install.ps1'))
-          choco install -y --no-progress git gzip
-          echo "C:\Program Files\Git\cmd" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
-          echo "C:\ProgramData\chocolatey\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
-      - name: Install Visual Studio 2022
-        run: |
-          $components = @(
-            "Microsoft.VisualStudio.Component.CoreEditor",
-            "Microsoft.VisualStudio.Workload.CoreEditor",
-            "Microsoft.VisualStudio.Component.Roslyn.Compiler",
-            "Microsoft.Component.MSBuild",
-            "Microsoft.VisualStudio.Component.TextTemplating",
-            "Microsoft.VisualStudio.Component.Debugger.JustInTime",
-            "Microsoft.VisualStudio.Component.VC.CoreIde",
-            "Microsoft.VisualStudio.Component.VC.Tools.x86.x64",
-            "Microsoft.VisualStudio.Component.Windows11SDK.22621",
-            "Microsoft.VisualStudio.Component.VC.Tools.ARM64EC",
-            "Microsoft.VisualStudio.Component.VC.Tools.ARM64",
-            "Microsoft.VisualStudio.Component.VC.ATL",
-            "Microsoft.VisualStudio.Component.VC.ATL.ARM64",
-            "Microsoft.VisualStudio.Component.Graphics",
-            "Microsoft.VisualStudio.Component.VC.Redist.14.Latest",
-            "Microsoft.VisualStudio.ComponentGroup.NativeDesktop.Core",
-            "Microsoft.VisualStudio.Component.Windows11Sdk.WindowsPerformanceToolkit",
-            "Microsoft.VisualStudio.Component.CppBuildInsights",
-            "Microsoft.VisualStudio.Component.VC.DiagnosticTools",
-            "Microsoft.VisualStudio.ComponentGroup.WebToolsExtensions.CMake",
-            "Microsoft.VisualStudio.Component.VC.CMake.Project",
-            "Microsoft.VisualStudio.Component.VC.ASAN",
-            "Microsoft.VisualStudio.Component.Vcpkg",
-            "Microsoft.VisualStudio.Workload.NativeDesktop"
-          )
-          $config = @{
-                "version" = "1.0"
-                "components"  = $components
-                "extensions"  = @()
-            }
-          $configPath = "${env:RUNNER_TEMP}\vsconfig"
-          $config | ConvertTo-Json | Out-File -FilePath $configPath
-          $bootstrapperFilePath = "${env:RUNNER_TEMP}\vs_community.exe"
-          write-host "Downloading Visual Studio 2022"
-          Invoke-WebRequest -Uri "https://aka.ms/vs/17/release/vs_community.exe" -outfile $bootstrapperFilePath
-          $bootstrapperArgumentList = ('/c', $bootstrapperFilePath, '--config', $configPath, '--quiet', '--wait' )
-          write-host "Installing Visual Studio 2022"
-          $process = Start-Process -FilePath cmd.exe -ArgumentList $bootstrapperArgumentList -Wait -PassThru
-          $exitCode = $process.ExitCode
-          write-host $exitCode
-      # pacman in mingw/msys2 is ~broken on windows arm right now - hangs consistently during attempts to install
-      # so we'll use this alternative GCC binary
-      - name: Install llvm-mingw GCC
-        run: |
-          $gcc_url="https://github.com/mstorsjo/llvm-mingw/releases/download/20240619/llvm-mingw-20240619-ucrt-aarch64.zip"
-          write-host "Downloading llvm-mingw"
-          Invoke-WebRequest -Uri "${gcc_url}" -OutFile "${env:RUNNER_TEMP}\gcc.zip"
-          write-host "Unpacking llvm-mingw"
-          expand-archive -path "${env:RUNNER_TEMP}\gcc.zip" -destinationpath "c:\"
-          mv c:\llvm-mingw-* c:\llvm-mingw
-          echo "c:\llvm-mingw\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
-      - name: Verify GCC
-        run: |
-          echo $env:PATH
-          gcc --version
-      - uses: actions/checkout@v4
-      - name: Set Version
-        run: |
-          $ver=${env:GITHUB_REF_NAME}.trim("v")
-          write-host VERSION=$ver | Out-File -FilePath ${env:GITHUB_ENV} -Encoding utf8 -Append
-      - uses: 'google-github-actions/auth@v2'
-        with:
-          project_id: 'ollama'
-          credentials_json: '${{ secrets.GOOGLE_SIGNING_CREDENTIALS }}'
-      - run: echo "${{ vars.OLLAMA_CERT }}" | Out-File -FilePath ollama_inc.crt -Encoding utf8
-      - name: install Windows SDK 8.1 to get signtool
-        run: |
-          $ErrorActionPreference = "Stop"
-          write-host "downloading SDK"
-          Invoke-WebRequest -Uri "https://go.microsoft.com/fwlink/p/?LinkId=323507" -OutFile "${env:RUNNER_TEMP}\sdksetup.exe"
-          Start-Process "${env:RUNNER_TEMP}\sdksetup.exe" -ArgumentList @("/q") -NoNewWindow -Wait
-          write-host "Win SDK 8.1 installed"
-          gci -path 'C:\Program Files (x86)\Windows Kits\' -r -fi 'signtool.exe'
-      - name: install signing plugin
-        run: |
-          $ErrorActionPreference = "Stop"
-          write-host "downloading plugin"
-          Invoke-WebRequest -Uri "https://github.com/GoogleCloudPlatform/kms-integrations/releases/download/cng-v1.0/kmscng-1.0-windows-amd64.zip" -OutFile "${env:RUNNER_TEMP}\plugin.zip"
-          Expand-Archive -Path "${env:RUNNER_TEMP}\plugin.zip" -DestinationPath ${env:RUNNER_TEMP}\plugin\
-          write-host "Installing plugin"
-          & "${env:RUNNER_TEMP}\plugin\*\kmscng.msi" /quiet
-          write-host "plugin installed"
-      - uses: actions/setup-go@v5
-        with:
-          go-version-file: go.mod
-          cache: true
-      - run: go get ./...
-      - run: |
-          $gopath=(get-command go).source | split-path -parent
-          $gccpath=(get-command gcc).source | split-path -parent
-          & "C:\Program Files\Microsoft Visual Studio\2022\Community\Common7\Tools\Launch-VsDevShell.ps1"
-          cd $env:GITHUB_WORKSPACE
-          $env:CMAKE_SYSTEM_VERSION="10.0.22621.0"
-          $env:PATH="$gopath;$gccpath;$env:PATH;C:\Program Files\Microsoft Visual Studio\2022\Community\Common7\IDE\CommonExtensions\Microsoft\CMake\CMake\bin"
-          echo $env:PATH
-          $env:ARCH="arm64"
-          .\scripts\build_windows.ps1 buildOllama buildApp gatherDependencies distZip
-        name: 'Windows Build'
-      - uses: actions/upload-artifact@v4
-        with:
-          name: windows-arm64
-          path: |
-            dist/windows-arm64/**
-            dist/windows-arm64-app.exe
-            dist/ollama-windows-arm64.zip
-
-  # Import the prior generation steps plus the full arm64 build, and build the final windows assets
+  # Import the prior generation steps and build the final windows assets
  build-windows:
    environment: release
    runs-on: windows
@@ -409,7 +281,6 @@ jobs:
      - generate-windows-cuda
      - generate-windows-rocm
      - generate-windows-cpu
-      - windows-arm64
    env:
      KEY_CONTAINER: ${{ vars.KEY_CONTAINER }}
    steps:
@@ -467,11 +338,7 @@ jobs:
      - uses: actions/download-artifact@v4
        with:
          name: generate-windows-rocm
-      - uses: actions/download-artifact@v4
-        with:
-          name: windows-arm64
-          path: dist
-      - run: dir build
+      - run: dir llm/build
      - run: |
          $gopath=(get-command go).source | split-path -parent
          & "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\Common7\Tools\Launch-VsDevShell.ps1"
@@ -492,7 +359,9 @@ jobs:
    environment: release
    runs-on: linux
    env:
-      PLATFORM: linux/amd64
+      OLLAMA_SKIP_MANIFEST_CREATE: '1'
+      BUILD_ARCH: amd64
+      PUSH: '1'
    steps:
      - uses: actions/checkout@v4
        with:
@@ -500,8 +369,14 @@ jobs:
      - name: Set Version
        shell: bash
        run: echo "VERSION=${GITHUB_REF_NAME#v}" >> $GITHUB_ENV
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ vars.DOCKER_USER }}
+          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
      - run: |
          ./scripts/build_linux.sh
+          ./scripts/build_docker.sh
      - uses: actions/upload-artifact@v4
        with:
          name: dist-linux-amd64
@@ -515,7 +390,9 @@ jobs:
    environment: release
    runs-on: linux-arm64
    env:
-      PLATFORM: linux/arm64
+      OLLAMA_SKIP_MANIFEST_CREATE: '1'
+      BUILD_ARCH: arm64
+      PUSH: '1'
    steps:
      - uses: actions/checkout@v4
        with:
@@ -544,8 +421,14 @@ jobs:
          sudo usermod -aG docker $USER
          sudo apt-get install acl
          sudo setfacl --modify user:$USER:rw /var/run/docker.sock
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ vars.DOCKER_USER }}
+          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
      - run: |
          ./scripts/build_linux.sh
+          ./scripts/build_docker.sh
      - uses: actions/upload-artifact@v4
        with:
          name: dist-linux-arm64
@@ -553,178 +436,6 @@ jobs:
            dist/*linux*
            !dist/*-cov

-  # Container image build
-  build-container-image:
-    environment: release
-    strategy:
-      matrix:
-        runner:
-          - linux
-          - linux-arm64
-    runs-on: ${{ matrix.runner }}
-    env:
-      FINAL_IMAGE_REPO: ollama/ollama
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - name: 'Install Docker'
-        if: ${{ startsWith(matrix.runner, 'linux-arm64') }}
-        run: |
-          sudo apt-get update
-          sudo apt-get install -y ca-certificates curl
-          sudo install -m 0755 -d /etc/apt/keyrings
-          sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc
-          sudo chmod a+r /etc/apt/keyrings/docker.asc
-          echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \
-            $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
-            sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
-          sudo apt-get update
-          sudo apt-get install -y docker-ce docker-ce-cli containerd.io
-          sudo usermod -aG docker $USER
-          sudo apt-get install acl
-          sudo setfacl --modify user:$USER:rw /var/run/docker.sock
-      - name: Docker meta
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.FINAL_IMAGE_REPO }}
-          flavor: |
-            latest=false
-          tags: |
-            type=ref,enable=true,priority=600,prefix=0.0.0-pr,suffix=,event=pr
-            type=semver,pattern={{version}}
-      - name: Set Version
-        shell: bash
-        run: |
-          machine=$(uname -m)
-          case ${machine} in
-            x86_64) echo ARCH=amd64; echo PLATFORM_PAIR=linux-amd64 ;;
-            aarch64) echo ARCH=arm64; echo PLATFORM_PAIR=linux-arm64 ;;
-          esac >>$GITHUB_ENV
-          echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ env.DOCKER_METADATA_OUTPUT_VERSION }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_ENV
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
-        with:
-          username: ${{ vars.DOCKER_USER }}
-          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
-      - name: Build and push by digest
-        id: build
-        uses: docker/build-push-action@v6
-        with:
-          context: "."
-          platforms: linux/${{ env.ARCH }}
-          build-args: |
-            GOFLAGS
-          outputs: type=image,name=${{ env.FINAL_IMAGE_REPO }},push-by-digest=true,name-canonical=true,push=true
-      - name: Export digest
-        run: |
-          mkdir -p /tmp/digests
-          digest="${{ steps.build.outputs.digest }}"
-          touch "/tmp/digests/${digest#sha256:}"
-      - name: Upload digest
-        uses: actions/upload-artifact@v4
-        with:
-          name: digests-${{ env.PLATFORM_PAIR }}
-          path: /tmp/digests/*
-          if-no-files-found: error
-          retention-days: 1
-  merge:
-    environment: release
-    runs-on: linux
-    needs:
-      - build-container-image
-    env:
-      FINAL_IMAGE_REPO: ollama/ollama
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - name: Download digests
-        uses: actions/download-artifact@v4
-        with:
-          path: /tmp/digests
-          pattern: digests-*
-          merge-multiple: true
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Docker meta
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.FINAL_IMAGE_REPO }}
-          flavor: |
-            latest=false
-          tags: |
-            type=ref,enable=true,priority=600,prefix=0.0.0-pr,suffix=,event=pr
-            type=semver,pattern={{version}}
-      - name: Set Version
-        shell: bash
-        run: |
-          machine=$(uname -m)
-          case ${machine} in
-            x86_64) echo ARCH=amd64; echo PLATFORM_PAIR=linux-amd64 ;;
-            aarch64) echo ARCH=arm64; echo PLATFORM_PAIR=linux-arm64 ;;
-          esac >>$GITHUB_ENV
-          echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ env.DOCKER_METADATA_OUTPUT_VERSION }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_ENV
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
-        with:
-          username: ${{ vars.DOCKER_USER }}
-          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
-      - name: Create manifest list and push
-        working-directory: /tmp/digests
-        run: |
-          docker buildx imagetools create $(jq -cr '.tags | map("-t " + .) | join(" ")' <<< "$DOCKER_METADATA_OUTPUT_JSON") \
-            $(printf '${{ env.FINAL_IMAGE_REPO }}@sha256:%s ' *)
-      - name: Inspect image
-        run: |
-          docker buildx imagetools inspect ${{ env.FINAL_IMAGE_REPO }}:${{ steps.meta.outputs.version }}          
-  build-container-image-rocm:
-    environment: release
-    runs-on: linux
-    env:
-      FINAL_IMAGE_REPO: ollama/ollama
-      ARCH: amd64
-      PLATFORM_PAIR: linux-amd64
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - name: Docker meta
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.FINAL_IMAGE_REPO }}
-          flavor: |
-            latest=false
-          tags: |
-            type=ref,enable=true,priority=600,prefix=0.0.0-pr,suffix=,event=pr
-            type=semver,pattern={{version}}
-      - name: Set Version
-        shell: bash
-        run: |
-          echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ env.DOCKER_METADATA_OUTPUT_VERSION }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_ENV
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
-        with:
-          username: ${{ vars.DOCKER_USER }}
-          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
-      - name: Build and push by digest
-        id: build
-        uses: docker/build-push-action@v6
-        with:
-          context: "."
-          target: runtime-rocm
-          build-args: |
-            GOFLAGS
-          tags: ${{ env.FINAL_IMAGE_REPO }}:${{ env.DOCKER_METADATA_OUTPUT_VERSION}}-rocm
-          push: true
-
  # Aggregate all the assets and ship a release
  release:
    needs:
@@ -737,6 +448,8 @@ jobs:
    permissions:
      contents: write
    env:
+      OLLAMA_SKIP_IMAGE_BUILD: '1'
+      PUSH: '1'
      GH_TOKEN: ${{ github.token }}
    steps:
      - uses: actions/checkout@v4
@@ -745,6 +458,12 @@ jobs:
        run: |
          echo "VERSION=${GITHUB_REF_NAME#v}" >> $GITHUB_ENV
          echo "RELEASE_VERSION=$(echo ${GITHUB_REF_NAME} | cut -f1 -d-)" >> $GITHUB_ENV
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ vars.DOCKER_USER }}
+          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
+      - run: ./scripts/build_docker.sh
      - name: Retrieve built artifact
        uses: actions/download-artifact@v4
        with:
@@ -755,6 +474,8 @@ jobs:
          ls -lh dist/
          (cd dist; find . -type f | xargs sha256sum > ../sha256sum.txt)
          mv sha256sum.txt dist/
+          mv dist/linux-???64 .
+          mv dist/linux-amd64-rocm .
          cat dist/sha256sum.txt
      - name: Create or update Release
        run: |
--- a/.github/workflows/test.yaml
+++ b/.github/workflows/test.yaml
@@ -81,6 +81,12 @@ jobs:
        if: ${{ ! startsWith(matrix.os, 'windows-') }}
        name: 'Unix Go Generate'
      - run: go build .
+      - uses: actions/upload-artifact@v4
+        with:
+          name: ${{ matrix.os }}-${{ matrix.arch }}-libraries
+          path: |
+            llm/build/**/bin/*
+            llm/build/**/*.a
  generate-cuda:
    needs: [changes]
    if: ${{ needs.changes.outputs.GENERATE_CUDA == 'True' }}
@@ -108,6 +114,12 @@ jobs:
          go generate -x ./...
        env:
          OLLAMA_SKIP_CPU_GENERATE: '1'
+      - uses: actions/upload-artifact@v4
+        with:
+          name: cuda-${{ matrix.cuda-version }}-libraries
+          path: |
+            llm/build/**/bin/*
+            dist/windows-amd64/**
  generate-rocm:
    needs: [changes]
    if: ${{ needs.changes.outputs.GENERATE_ROCM == 'True' }}
@@ -135,6 +147,12 @@ jobs:
          go generate -x ./...
        env:
          OLLAMA_SKIP_CPU_GENERATE: '1'
+      - uses: actions/upload-artifact@v4
+        with:
+          name: rocm-${{ matrix.rocm-version }}-libraries
+          path: |
+            llm/build/**/bin/*
+            dist/windows-amd64/**

  # ROCm generation step
  generate-windows-rocm:
@@ -171,6 +189,7 @@ jobs:
        name: go generate
        env:
          OLLAMA_SKIP_CPU_GENERATE: '1'
+      # TODO - do we need any artifacts?

  # CUDA generation step
  generate-windows-cuda:
@@ -212,6 +231,7 @@ jobs:
          go generate -x ./...
        env:
          OLLAMA_SKIP_CPU_GENERATE: '1'
+      # TODO - do we need any artifacts?

  lint:
    strategy:
@@ -243,6 +263,14 @@ jobs:
            arm64) echo ARCH=arm64 ;;
          esac >>$GITHUB_ENV
        shell: bash
+      - run: |
+          mkdir -p llm/build/linux/$ARCH/stub/bin
+          touch llm/build/linux/$ARCH/stub/bin/ollama_llama_server
+        if: ${{ startsWith(matrix.os, 'ubuntu-') }}
+      - run: |
+          mkdir -p llm/build/darwin/$ARCH/stub/bin
+          touch llm/build/darwin/$ARCH/stub/bin/ollama_llama_server
+        if: ${{ startsWith(matrix.os, 'macos-') }}
      - uses: golangci/golangci-lint-action@v6
        with:
          args: --timeout 8m0s -v
@@ -273,10 +301,23 @@ jobs:
          cache: true
      - run: |
          case ${{ matrix.arch }} in
-            amd64) echo ARCH=amd64 ;;
+            amd64) echo ARCH=x86_64 ;;
            arm64) echo ARCH=arm64 ;;
          esac >>$GITHUB_ENV
        shell: bash
+      - run: |
+          mkdir -p llm/build/linux/$ARCH/stub/bin
+          touch llm/build/linux/$ARCH/stub/bin/ollama_llama_server
+        if: ${{ startsWith(matrix.os, 'ubuntu-') }}
+      - run: |
+          mkdir -p llm/build/darwin/$ARCH/stub/bin
+          touch llm/build/darwin/$ARCH/stub/bin/ollama_llama_server
+        if: ${{ startsWith(matrix.os, 'macos-') }}
+        shell: bash
      - run: go generate ./...
      - run: go build
      - run: go test -v ./...
+      - uses: actions/upload-artifact@v4
+        with:
+          name: ${{ matrix.os }}-binaries
+          path: ollama
--- a/.gitignore
+++ b/.gitignore
@@ -5,14 +5,10 @@
 .swp
 dist
 ollama
-ggml-metal.metal
 .cache
 *.exe
 .idea
 test_data
 *.crt
 llm/build
-build/*/*/*
-!build/**/placeholder
-llama/build
 __debug_bin*
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -18,7 +18,7 @@ See the [development documentation](./docs/development.md) for instructions on h

 * New features: new features (e.g. API fields, environment variables) add surface area to Ollama and make it harder to maintain in the long run as they cannot be removed without potentially breaking users in the future.
 * Refactoring: large code improvements are important, but can be harder or take longer to review and merge.
-* Documentation: small updates to fill in or correct missing documentation is helpful, however large documentation additions can be hard to maintain over time.
+* Documentation: small updates to fill in or dorrect missing documentation is helpful, however large documentation additions can be hard to maintain over time.

 ### Issues that may not be accepted

--- a/145
+++ b/145
@@ -16,12 +16,12 @@ FROM --platform=linux/amd64 nvidia/cuda:$CUDA_VERSION_11-devel-centos7 AS cuda-1
 ARG CMAKE_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/devtoolset-10/root/usr/bin:$PATH
+ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ARG CUDA_V11_ARCHITECTURES
-ENV GOARCH=amd64
+ENV GOARCH amd64
 RUN --mount=type=cache,target=/root/.ccache \
    OLLAMA_SKIP_STATIC_GENERATE=1 \
    OLLAMA_SKIP_CPU_GENERATE=1 \
@@ -33,12 +33,12 @@ FROM --platform=linux/amd64 nvidia/cuda:$CUDA_VERSION_12-devel-centos7 AS cuda-1
 ARG CMAKE_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/devtoolset-10/root/usr/bin:$PATH
+ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ARG CUDA_V12_ARCHITECTURES
-ENV GOARCH=amd64
+ENV GOARCH amd64
 RUN --mount=type=cache,target=/root/.ccache \
    OLLAMA_SKIP_STATIC_GENERATE=1 \
    OLLAMA_SKIP_CPU_GENERATE=1 \
@@ -47,32 +47,32 @@ RUN --mount=type=cache,target=/root/.ccache \
    OLLAMA_CUSTOM_CUDA_DEFS="-DGGML_CUDA_USE_GRAPHS=on" \
    bash gen_linux.sh

-FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_11-devel-rockylinux8 AS cuda-11-build-runner-arm64
+FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_11-devel-rockylinux8 AS cuda-11-build-server-arm64
 ARG CMAKE_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/gcc-toolset-10/root/usr/bin:$PATH
+ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ARG CUDA_V11_ARCHITECTURES
-ENV GOARCH=arm64
+ENV GOARCH arm64
 RUN OLLAMA_SKIP_STATIC_GENERATE=1 \
    OLLAMA_SKIP_CPU_GENERATE=1 \
    CMAKE_CUDA_ARCHITECTURES="${CUDA_V11_ARCHITECTURES}" \
    CUDA_VARIANT="_v11" \
    bash gen_linux.sh

-FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_12-devel-rockylinux8 AS cuda-12-build-runner-arm64
+FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_12-devel-rockylinux8 AS cuda-12-build-server-arm64
 ARG CMAKE_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/gcc-toolset-10/root/usr/bin:$PATH
+ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ARG CUDA_V12_ARCHITECTURES
-ENV GOARCH=arm64
+ENV GOARCH arm64
 RUN --mount=type=cache,target=/root/.ccache \
    OLLAMA_SKIP_STATIC_GENERATE=1 \
    OLLAMA_SKIP_CPU_GENERATE=1 \
@@ -86,13 +86,13 @@ FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete AS rocm-b
 ARG CMAKE_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/devtoolset-10/root/usr/bin:$PATH
-ENV LIBRARY_PATH=/opt/amdgpu/lib64
+ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
+ENV LIBRARY_PATH /opt/amdgpu/lib64
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ARG AMDGPU_TARGETS
-ENV GOARCH=amd64
+ENV GOARCH amd64
 RUN --mount=type=cache,target=/root/.ccache \
    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_SKIP_CPU_GENERATE=1 bash gen_linux.sh
 RUN mkdir -p ../../dist/linux-amd64-rocm/lib/ollama && \
@@ -103,11 +103,11 @@ ARG CMAKE_VERSION
 ARG GOLANG_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/devtoolset-10/root/usr/bin:$PATH
+ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 ARG OLLAMA_CUSTOM_CPU_DEFS
 ARG CGO_CFLAGS
-ENV GOARCH=amd64
+ENV GOARCH amd64
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate

 FROM --platform=linux/amd64 cpu-builder-amd64 AS static-build-amd64
@@ -128,11 +128,11 @@ ARG CMAKE_VERSION
 ARG GOLANG_VERSION
 COPY ./scripts/rh_linux_deps.sh /
 RUN CMAKE_VERSION=${CMAKE_VERSION} GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh
-ENV PATH=/opt/rh/gcc-toolset-10/root/usr/bin:$PATH
+ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
 COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 ARG OLLAMA_CUSTOM_CPU_DEFS
 ARG CGO_CFLAGS
-ENV GOARCH=arm64
+ENV GOARCH arm64
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate

 FROM --platform=linux/arm64 cpu-builder-arm64 AS static-build-arm64
@@ -143,112 +143,73 @@ RUN --mount=type=cache,target=/root/.ccache \
    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" bash gen_linux.sh


-# Intermediate stages used for ./scripts/build_linux.sh
+# Intermediate stage used for ./scripts/build_linux.sh
 FROM --platform=linux/amd64 cpu-build-amd64 AS build-amd64
-ENV CGO_ENABLED=1
+ENV CGO_ENABLED 1
 WORKDIR /go/src/github.com/ollama/ollama
 COPY . .
-COPY --from=static-build-amd64 /go/src/github.com/ollama/ollama/llm/build/ llm/build/
-COPY --from=cpu_avx-build-amd64 /go/src/github.com/ollama/ollama/build/ build/
-COPY --from=cpu_avx2-build-amd64 /go/src/github.com/ollama/ollama/build/ build/
+COPY --from=static-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
+COPY --from=cpu_avx-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
+COPY --from=cpu_avx2-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 COPY --from=cuda-11-build-amd64 /go/src/github.com/ollama/ollama/dist/ dist/
-COPY --from=cuda-11-build-amd64 /go/src/github.com/ollama/ollama/build/ build/
+COPY --from=cuda-11-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 COPY --from=cuda-12-build-amd64 /go/src/github.com/ollama/ollama/dist/ dist/
-COPY --from=cuda-12-build-amd64 /go/src/github.com/ollama/ollama/build/ build/
+COPY --from=cuda-12-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/dist/ dist/
-COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/build/ build/
+COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 ARG GOFLAGS
 ARG CGO_CFLAGS
 RUN --mount=type=cache,target=/root/.ccache \
    go build -trimpath -o dist/linux-amd64/bin/ollama .
-RUN cd dist/linux-$GOARCH && \
-    tar --exclude runners -cf - . | pigz --best > ../ollama-linux-$GOARCH.tgz
-RUN cd dist/linux-$GOARCH-rocm && \
-    tar -cf - . | pigz --best > ../ollama-linux-$GOARCH-rocm.tgz

+# Intermediate stage used for ./scripts/build_linux.sh
 FROM --platform=linux/arm64 cpu-build-arm64 AS build-arm64
-ENV CGO_ENABLED=1
+ENV CGO_ENABLED 1
 ARG GOLANG_VERSION
 WORKDIR /go/src/github.com/ollama/ollama
 COPY . .
-COPY --from=static-build-arm64 /go/src/github.com/ollama/ollama/llm/build/ llm/build/
-COPY --from=cuda-11-build-runner-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
-COPY --from=cuda-11-build-runner-arm64 /go/src/github.com/ollama/ollama/build/ build/
-COPY --from=cuda-12-build-runner-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
-COPY --from=cuda-12-build-runner-arm64 /go/src/github.com/ollama/ollama/build/ build/
-ARG GOFLAGS
-ARG CGO_CFLAGS
-RUN --mount=type=cache,target=/root/.ccache \
-    go build -trimpath -o dist/linux-arm64/bin/ollama .
-RUN cd dist/linux-$GOARCH && \
-    tar --exclude runners -cf - . | pigz --best > ../ollama-linux-$GOARCH.tgz
-
-FROM --platform=linux/amd64 scratch AS dist-amd64
-COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/ollama-linux-*.tgz /
-FROM --platform=linux/arm64 scratch AS dist-arm64
-COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/ollama-linux-*.tgz /
-FROM dist-$TARGETARCH as dist
-
-
-# Optimized container images do not cary nested payloads
-FROM --platform=linux/amd64 static-build-amd64 AS container-build-amd64
-WORKDIR /go/src/github.com/ollama/ollama
-COPY . .
-ARG GOFLAGS
-ARG CGO_CFLAGS
-RUN --mount=type=cache,target=/root/.ccache \
-    go build -trimpath -o dist/linux-amd64/bin/ollama .
-
-FROM --platform=linux/arm64 static-build-arm64 AS container-build-arm64
-WORKDIR /go/src/github.com/ollama/ollama
-COPY . .
+COPY --from=static-build-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
+COPY --from=cuda-11-build-server-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
+COPY --from=cuda-11-build-server-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
+COPY --from=cuda-12-build-server-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
+COPY --from=cuda-12-build-server-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 ARG GOFLAGS
 ARG CGO_CFLAGS
 RUN --mount=type=cache,target=/root/.ccache \
    go build -trimpath -o dist/linux-arm64/bin/ollama .

-FROM --platform=linux/amd64 ubuntu:22.04 AS runtime-amd64
-RUN apt-get update && \
-    apt-get install -y ca-certificates && \
-    apt-get clean && rm -rf /var/lib/apt/lists/*
-COPY --from=container-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/
-COPY --from=cpu-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=cpu_avx-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=cpu_avx2-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=cuda-11-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=cuda-12-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
+# Strip out ROCm dependencies to keep the primary image lean
+FROM --platform=linux/amd64 ubuntu:22.04 as amd64-libs-without-rocm
+COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /scratch/
+RUN cd /scratch/ollama/ && rm -rf rocblas libamd* libdrm* libroc* libhip* libhsa*

-FROM --platform=linux/arm64 ubuntu:22.04 AS runtime-arm64
-RUN apt-get update && \
-    apt-get install -y ca-certificates && \
+# Runtime stages
+FROM --platform=linux/amd64 ubuntu:22.04 as runtime-amd64
+COPY --from=amd64-libs-without-rocm /scratch/ /lib/
+RUN apt-get update && apt-get install -y ca-certificates && \
    apt-get clean && rm -rf /var/lib/apt/lists/*
-COPY --from=container-build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/bin/ /bin/
-COPY --from=cpu-build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/
-COPY --from=cuda-11-build-runner-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/
-COPY --from=cuda-12-build-runner-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/
+COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/

-# ROCm libraries larger so we keep it distinct from the CPU/CUDA image
-FROM --platform=linux/amd64 ubuntu:22.04 AS runtime-rocm
-# Frontload the rocm libraries which are large, and rarely change to increase chance of a common layer
-# across releases
-COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64-rocm/lib/ /lib/
-RUN apt-get update && \
-    apt-get install -y ca-certificates && \
+FROM --platform=linux/arm64 ubuntu:22.04 as runtime-arm64
+COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/
+RUN apt-get update && apt-get install -y ca-certificates && \
    apt-get clean && rm -rf /var/lib/apt/lists/*
-COPY --from=container-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/
-COPY --from=cpu-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=cpu_avx-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=cpu_avx2-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
+COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/bin/ /bin/
+
+# Radeon images are much larger so we keep it distinct from the CPU/CUDA image
+FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete as runtime-rocm
+RUN update-pciids
+COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/
+RUN ln -s /opt/rocm/lib /lib/ollama
 EXPOSE 11434
-ENV OLLAMA_HOST=0.0.0.0
+ENV OLLAMA_HOST 0.0.0.0

 ENTRYPOINT ["/bin/ollama"]
 CMD ["serve"]

 FROM runtime-$TARGETARCH
 EXPOSE 11434
-ENV OLLAMA_HOST=0.0.0.0
+ENV OLLAMA_HOST 0.0.0.0
 ENV PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
 ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
 ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
--- a/README.md
+++ b/README.md
@@ -197,18 +197,6 @@ ollama show llama3.1
 ollama list
 ```

-### List which models are currently loaded
-
-```
-ollama ps
-```
-
-### Stop a model which is currently running
-
-```
-ollama stop llama3.1
-```
-
 ### Start Ollama

 `ollama serve` is used when you want to start ollama without running the desktop application.
@@ -307,7 +295,6 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [Olpaka](https://github.com/Otacon/olpaka) (User-friendly Flutter Web App for Ollama)
 - [OllamaSpring](https://github.com/CrazyNeil/OllamaSpring) (Ollama Client for macOS)
 - [LLocal.in](https://github.com/kartikm7/llocal) (Easy to use Electron Desktop Client for Ollama)
- [AiLama](https://github.com/zeyoyt/ailama) (A Discord User App that allows you to interact with Ollama anywhere in discord )
 - [Ollama with Google Mesop](https://github.com/rapidarchitect/ollama_mesop/) (Mesop Chat Client implementation with Ollama)
 - [Painting Droid](https://github.com/mateuszmigas/painting-droid) (Painting app with AI integrations)
 - [Kerlig AI](https://www.kerlig.com/) (AI writing assistant for macOS)
@@ -319,12 +306,6 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [Go-CREW](https://www.jonathanhecl.com/go-crew/) (Powerful Offline RAG in Golang)
 - [PartCAD](https://github.com/openvmp/partcad/) (CAD model generation with OpenSCAD and CadQuery)
 - [Ollama4j Web UI](https://github.com/ollama4j/ollama4j-web-ui) - Java-based Web UI for Ollama built with Vaadin, Spring Boot and Ollama4j
- [PyOllaMx](https://github.com/kspviswa/pyOllaMx) - macOS application capable of chatting with both Ollama and Apple MLX models.
- [Claude Dev](https://github.com/saoudrizwan/claude-dev) - VSCode extension for multi-file/whole-repo coding
- [Cherry Studio](https://github.com/kangfenmao/cherry-studio) (Desktop client with Ollama support)
- [ConfiChat](https://github.com/1runeberg/confichat) (Lightweight, standalone, multi-platform, and privacy focused LLM chat interface with optional encryption)
- [Archyve](https://github.com/nickthecook/archyve) (RAG-enabling document library)
- [crewAI with Mesop](https://github.com/rapidarchitect/ollama-crew-mesop) (Mesop Web Interface to run crewAI with Ollama)

 ### Terminal

@@ -349,11 +330,6 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [podman-ollama](https://github.com/ericcurtin/podman-ollama)
 - [gollama](https://github.com/sammcj/gollama)
 - [Ollama eBook Summary](https://github.com/cognitivetech/ollama-ebook-summary/)
- [Ollama Mixture of Experts (MOE) in 50 lines of code](https://github.com/rapidarchitect/ollama_moe)
- [vim-intelligence-bridge](https://github.com/pepo-ec/vim-intelligence-bridge) Simple interaction of "Ollama" with the Vim editor
-
-### Apple Vision Pro
- [Enchanted](https://github.com/AugustDev/enchanted)

 ### Database

@@ -363,7 +339,6 @@ See the [API documentation](./docs/api.md) for all endpoints.
 ### Package managers

 - [Pacman](https://archlinux.org/packages/extra/x86_64/ollama/)
- [Gentoo](https://github.com/gentoo/guru/tree/master/app-misc/ollama)
 - [Helm Chart](https://artifacthub.io/packages/helm/ollama-helm/ollama)
 - [Guix channel](https://codeberg.org/tusharhero/ollama-guix)
 - [Nix package](https://search.nixos.org/packages?channel=24.05&show=ollama&from=0&size=50&sort=relevance&type=packages&query=ollama)
@@ -373,7 +348,6 @@ See the [API documentation](./docs/api.md) for all endpoints.

 - [LangChain](https://python.langchain.com/docs/integrations/llms/ollama) and [LangChain.js](https://js.langchain.com/docs/modules/model_io/models/llms/integrations/ollama) with [example](https://js.langchain.com/docs/use_cases/question_answering/local_retrieval_qa)
 - [Firebase Genkit](https://firebase.google.com/docs/genkit/plugins/ollama)
- [crewAI](https://github.com/crewAIInc/crewAI)
 - [LangChainGo](https://github.com/tmc/langchaingo/) with [example](https://github.com/tmc/langchaingo/tree/main/examples/ollama-completion-example)
 - [LangChain4j](https://github.com/langchain4j/langchain4j) with [example](https://github.com/langchain4j/langchain4j-examples/tree/main/ollama-examples/src/main/java)
 - [LangChainRust](https://github.com/Abraxas-365/langchain-rust) with [example](https://github.com/Abraxas-365/langchain-rust/blob/main/examples/llm_ollama.rs)
@@ -401,17 +375,12 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [Portkey](https://portkey.ai/docs/welcome/integration-guides/ollama)
 - [PromptingTools.jl](https://github.com/svilupp/PromptingTools.jl) with an [example](https://svilupp.github.io/PromptingTools.jl/dev/examples/working_with_ollama)
 - [LlamaScript](https://github.com/Project-Llama/llamascript)
- [Gollm](https://docs.gollm.co/examples/ollama-example)
 - [Ollamaclient for Golang](https://github.com/xyproto/ollamaclient)
- [High-level function abstraction in Go](https://gitlab.com/tozd/go/fun)
- [Ollama PHP](https://github.com/ArdaGnsrn/ollama-php)
- [Agents-Flex for Java](https://github.com/agents-flex/agents-flex) with [example](https://github.com/agents-flex/agents-flex/tree/main/agents-flex-llm/agents-flex-llm-ollama/src/test/java/com/agentsflex/llm/ollama)

 ### Mobile

 - [Enchanted](https://github.com/AugustDev/enchanted)
 - [Maid](https://github.com/Mobile-Artificial-Intelligence/maid)
- [ConfiChat](https://github.com/1runeberg/confichat) (Lightweight, standalone, multi-platform, and privacy focused LLM chat interface with optional encryption)

 ### Extensions & Plugins

@@ -436,16 +405,11 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [twinny](https://github.com/rjmacarthy/twinny) (Copilot and Copilot chat alternative using Ollama)
 - [Wingman-AI](https://github.com/RussellCanfield/wingman-ai) (Copilot code and chat alternative using Ollama and Hugging Face)
 - [Page Assist](https://github.com/n4ze3m/page-assist) (Chrome Extension)
- [Plasmoid Ollama Control](https://github.com/imoize/plasmoid-ollamacontrol) (KDE Plasma extension that allows you to quickly manage/control Ollama model)
 - [AI Telegram Bot](https://github.com/tusharhero/aitelegrambot) (Telegram bot using Ollama in backend)
 - [AI ST Completion](https://github.com/yaroslavyaroslav/OpenAI-sublime-text) (Sublime Text 4 AI assistant plugin with Ollama support)
 - [Discord-Ollama Chat Bot](https://github.com/kevinthedang/discord-ollama) (Generalized TypeScript Discord Bot w/ Tuning Documentation)
 - [Discord AI chat/moderation bot](https://github.com/rapmd73/Companion) Chat/moderation bot written in python. Uses Ollama to create personalities.
 - [Headless Ollama](https://github.com/nischalj10/headless-ollama) (Scripts to automatically install ollama client & models on any OS for apps that depends on ollama server)
- [vnc-lm](https://github.com/jk011ru/vnc-lm) (A containerized Discord bot with support for attachments and web links)
- [LSP-AI](https://github.com/SilasMarvin/lsp-ai) (Open-source language server for AI-powered functionality)
- [QodeAssist](https://github.com/Palm1r/QodeAssist) (AI-powered coding assistant plugin for Qt Creator)
- [Obsidian Quiz Generator plugin](https://github.com/ECuiDev/obsidian-quiz-generator)

 ### Supported backends

--- a/app/ollama.iss
+++ b/app/ollama.iss
@@ -28,8 +28,8 @@ AppPublisher={#MyAppPublisher}
 AppPublisherURL={#MyAppURL}
 AppSupportURL={#MyAppURL}
 AppUpdatesURL={#MyAppURL}
-ArchitecturesAllowed=x64compatible arm64
-ArchitecturesInstallIn64BitMode=x64compatible arm64
+ArchitecturesAllowed=x64 arm64
+ArchitecturesInstallIn64BitMode=x64 arm64
 DefaultDirName={localappdata}\Programs\{#MyAppName}
 DefaultGroupName={#MyAppName}
 DisableProgramGroupPage=yes
@@ -48,7 +48,6 @@ OutputDir=..\dist\
 SetupLogging=yes
 CloseApplications=yes
 RestartApplications=no
-RestartIfNeededByRun=no

 ; https://jrsoftware.org/ishelp/index.php?topic=setup_wizardimagefile
 WizardSmallImageFile=.\assets\setup.bmp
@@ -87,21 +86,12 @@ Name: "english"; MessagesFile: "compiler:Default.isl"
 DialogFontSize=12

 [Files]
-#if DirExists("..\dist\windows-amd64")
-Source: "..\dist\windows-amd64-app.exe"; DestDir: "{app}"; DestName: "{#MyAppExeName}" ;Check: not IsArm64();  Flags: ignoreversion 64bit
-Source: "..\dist\windows-amd64\ollama.exe"; DestDir: "{app}"; Check: not IsArm64(); Flags: ignoreversion 64bit
-Source: "..\dist\windows-amd64\lib\ollama\*"; DestDir: "{app}\lib\ollama\"; Check: not IsArm64(); Flags: ignoreversion 64bit recursesubdirs
-#endif
-
-#if DirExists("..\dist\windows-arm64")
-Source: "..\dist\windows-arm64\vc_redist.arm64.exe"; DestDir: "{tmp}"; Check: IsArm64() and vc_redist_needed(); Flags: deleteafterinstall
-Source: "..\dist\windows-arm64-app.exe"; DestDir: "{app}"; DestName: "{#MyAppExeName}" ;Check: IsArm64();  Flags: ignoreversion 64bit
-Source: "..\dist\windows-arm64\ollama.exe"; DestDir: "{app}"; Check: IsArm64(); Flags: ignoreversion 64bit
-Source: "..\dist\windows-arm64\lib\ollama\*"; DestDir: "{app}\lib\ollama\"; Check: IsArm64(); Flags: ignoreversion 64bit recursesubdirs
-#endif
-
+Source: ".\app.exe"; DestDir: "{app}"; DestName: "{#MyAppExeName}" ; Flags: ignoreversion 64bit
+Source: "..\ollama.exe"; DestDir: "{app}"; Flags: ignoreversion 64bit
+Source: "..\dist\windows-{#ARCH}\lib\ollama\runners\*"; DestDir: "{app}\lib\ollama\runners"; Flags: ignoreversion 64bit recursesubdirs
 Source: "..\dist\ollama_welcome.ps1"; DestDir: "{app}"; Flags: ignoreversion
 Source: ".\assets\app.ico"; DestDir: "{app}"; Flags: ignoreversion
+Source: "..\dist\windows-amd64\lib\ollama\*"; DestDir: "{app}\lib\ollama\"; Flags: ignoreversion recursesubdirs

 [Icons]
 Name: "{group}\{#MyAppName}"; Filename: "{app}\{#MyAppExeName}"; IconFilename: "{app}\app.ico"
@@ -109,9 +99,6 @@ Name: "{userstartup}\{#MyAppName}"; Filename: "{app}\{#MyAppExeName}"; IconFilen
 Name: "{userprograms}\{#MyAppName}"; Filename: "{app}\{#MyAppExeName}"; IconFilename: "{app}\app.ico"

 [Run]
-#if DirExists("..\dist\windows-arm64")
-Filename: "{tmp}\vc_redist.arm64.exe"; Parameters: "/install /passive /norestart"; Check: IsArm64() and vc_redist_needed(); StatusMsg: "Installing VC++ Redistributables..."; Flags: waituntilterminated
-#endif
 Filename: "{cmd}"; Parameters: "/C set PATH={app};%PATH% & ""{app}\{#MyAppExeName}"""; Flags: postinstall nowait runhidden

 [UninstallRun]
@@ -167,39 +154,3 @@ begin
  { Pos() returns 0 if not found }
  Result := Pos(';' + ExpandConstant(Param) + ';', ';' + OrigPath + ';') = 0;
 end;
-
-{ --- VC Runtime libraries discovery code - Only install vc_redist if it isn't already installed ----- }
-const VCRTL_MIN_V1 = 14;
-const VCRTL_MIN_V2 = 40;
-const VCRTL_MIN_V3 = 33807;
-const VCRTL_MIN_V4 = 0;
-
- // check if the minimum required vc redist is installed (by looking the registry)
-function vc_redist_needed (): Boolean;
-var
-  sRegKey: string;
-  v1: Cardinal;
-  v2: Cardinal;
-  v3: Cardinal;
-  v4: Cardinal;
-begin
-  sRegKey := 'SOFTWARE\WOW6432Node\Microsoft\VisualStudio\14.0\VC\Runtimes\arm64';
-  if (RegQueryDWordValue (HKEY_LOCAL_MACHINE, sRegKey, 'Major', v1)  and
-      RegQueryDWordValue (HKEY_LOCAL_MACHINE, sRegKey, 'Minor', v2) and
-      RegQueryDWordValue (HKEY_LOCAL_MACHINE, sRegKey, 'Bld', v3) and
-      RegQueryDWordValue (HKEY_LOCAL_MACHINE, sRegKey, 'RBld', v4)) then
-  begin
-    Log ('VC Redist version: ' + IntToStr (v1) +
-        '.' + IntToStr (v2) + '.' + IntToStr (v3) +
-        '.' + IntToStr (v4));
-    { Version info was found. Return true if later or equal to our
-       minimal required version RTL_MIN_Vx }
-    Result := not (
-        (v1 > VCRTL_MIN_V1) or ((v1 = VCRTL_MIN_V1) and
-         ((v2 > VCRTL_MIN_V2) or ((v2 = VCRTL_MIN_V2) and
-          ((v3 > VCRTL_MIN_V3) or ((v3 = VCRTL_MIN_V3) and
-           (v4 >= VCRTL_MIN_V4)))))));
-  end
-  else
-    Result := TRUE;
-end;
--- a/build/darwin/amd64/placeholder
+++ b/build/darwin/amd64/placeholder
@@ -1 +0,0 @@
-This is here to make sure the build/ directory exists for the go:embed command
--- a/build/darwin/arm64/placeholder
+++ b/build/darwin/arm64/placeholder
@@ -1 +0,0 @@
-This is here to make sure the build/ directory exists for the go:embed command
--- a/build/embed_darwin_amd64.go
+++ b/build/embed_darwin_amd64.go
@@ -1,8 +0,0 @@
-package build
-
-import "embed"
-
-// Darwin payloads separated by architecture to avoid duplicate payloads when cross compiling
-
-//go:embed darwin/amd64/*
-var EmbedFS embed.FS
--- a/build/embed_darwin_arm64.go
+++ b/build/embed_darwin_arm64.go
@@ -1,8 +0,0 @@
-package build
-
-import "embed"
-
-// Darwin payloads separated by architecture to avoid duplicate payloads when cross compiling
-
-//go:embed darwin/arm64/*
-var EmbedFS embed.FS
--- a/build/embed_linux.go
+++ b/build/embed_linux.go
@@ -1,6 +0,0 @@
-package build
-
-import "embed"
-
-//go:embed linux/*
-var EmbedFS embed.FS
--- a/build/embed_unused.go
+++ b/build/embed_unused.go
@@ -1,8 +0,0 @@
-//go:build !linux && !darwin
-
-package build
-
-import "embed"
-
-// unused on windows
-var EmbedFS embed.FS
--- a/build/linux/amd64/placeholder
+++ b/build/linux/amd64/placeholder
@@ -1 +0,0 @@
-This is here to make sure the build/ directory exists for the go:embed command
--- a/build/linux/arm64/placeholder
+++ b/build/linux/arm64/placeholder
@@ -1 +0,0 @@
-This is here to make sure the build/ directory exists for the go:embed command
--- a/cmd/cmd.go
+++ b/cmd/cmd.go
@@ -2,7 +2,6 @@ package cmd

 import (
 	"archive/zip"
-	"bufio"
 	"bytes"
 	"context"
 	"crypto/ed25519"
@@ -22,7 +21,6 @@ import (
 	"regexp"
 	"runtime"
 	"slices"
-	"strconv"
 	"strings"
 	"sync/atomic"
 	"syscall"
@@ -346,39 +344,6 @@ func (w *progressWriter) Write(p []byte) (n int, err error) {
 	return len(p), nil
 }

-func loadOrUnloadModel(cmd *cobra.Command, opts *runOptions) error {
-	p := progress.NewProgress(os.Stderr)
-	defer p.StopAndClear()
-
-	spinner := progress.NewSpinner("")
-	p.Add("", spinner)
-
-	client, err := api.ClientFromEnvironment()
-	if err != nil {
-		return err
-	}
-
-	req := &api.GenerateRequest{
-		Model:     opts.Model,
-		KeepAlive: opts.KeepAlive,
-	}
-
-	return client.Generate(cmd.Context(), req, func(api.GenerateResponse) error { return nil })
-}
-
-func StopHandler(cmd *cobra.Command, args []string) error {
-	opts := &runOptions{
-		Model:     args[0],
-		KeepAlive: &api.Duration{Duration: 0},
-	}
-	if err := loadOrUnloadModel(cmd, opts); err != nil {
-		if strings.Contains(err.Error(), "not found") {
-			return fmt.Errorf("couldn't find model \"%s\" to stop", args[0])
-		}
-	}
-	return nil
-}
-
 func RunHandler(cmd *cobra.Command, args []string) error {
 	interactive := true

@@ -457,7 +422,7 @@ func RunHandler(cmd *cobra.Command, args []string) error {
 	opts.ParentModel = info.Details.ParentModel

 	if interactive {
-		if err := loadOrUnloadModel(cmd, &opts); err != nil {
+		if err := loadModel(cmd, &opts); err != nil {
 			return err
 		}

@@ -613,7 +578,7 @@ func ListHandler(cmd *cobra.Command, args []string) error {
 	table.SetHeaderLine(false)
 	table.SetBorder(false)
 	table.SetNoWhiteSpace(true)
-	table.SetTablePadding("    ")
+	table.SetTablePadding("\t")
 	table.AppendBulk(data)
 	table.Render()

@@ -648,15 +613,7 @@ func ListRunningHandler(cmd *cobra.Command, args []string) error {
 				cpuPercent := math.Round(float64(sizeCPU) / float64(m.Size) * 100)
 				procStr = fmt.Sprintf("%d%%/%d%% CPU/GPU", int(cpuPercent), int(100-cpuPercent))
 			}
-
-			var until string
-			delta := time.Since(m.ExpiresAt)
-			if delta > 0 {
-				until = "Stopping..."
-			} else {
-				until = format.HumanTime(m.ExpiresAt, "Never")
-			}
-			data = append(data, []string{m.Name, m.Digest[:12], format.HumanBytes(m.Size), procStr, until})
+			data = append(data, []string{m.Name, m.Digest[:12], format.HumanBytes(m.Size), procStr, format.HumanTime(m.ExpiresAt, "Never")})
 		}
 	}

@@ -667,7 +624,7 @@ func ListRunningHandler(cmd *cobra.Command, args []string) error {
 	table.SetHeaderLine(false)
 	table.SetBorder(false)
 	table.SetNoWhiteSpace(true)
-	table.SetTablePadding("    ")
+	table.SetTablePadding("\t")
 	table.AppendBulk(data)
 	table.Render()

@@ -763,89 +720,125 @@ func ShowHandler(cmd *cobra.Command, args []string) error {
 		return nil
 	}

-	return showInfo(resp, os.Stdout)
+	showInfo(resp)
+
+	return nil
 }

-func showInfo(resp *api.ShowResponse, w io.Writer) error {
-	tableRender := func(header string, rows func() [][]string) {
-		fmt.Fprintln(w, " ", header)
-		table := tablewriter.NewWriter(w)
-		table.SetAlignment(tablewriter.ALIGN_LEFT)
-		table.SetBorder(false)
-		table.SetNoWhiteSpace(true)
-		table.SetTablePadding("    ")
-
-		switch header {
-		case "Template", "System", "License":
-			table.SetColWidth(100)
-		}
-
-		table.AppendBulk(rows())
-		table.Render()
-		fmt.Fprintln(w)
+func showInfo(resp *api.ShowResponse) {
+	modelData := [][]string{
+		{"parameters", resp.Details.ParameterSize},
+		{"quantization", resp.Details.QuantizationLevel},
+	}
+	if resp.ModelInfo != nil {
+		arch := resp.ModelInfo["general.architecture"].(string)
+		modelData = append(modelData,
+			[]string{"arch", arch},
+			[]string{"context length", fmt.Sprintf("%v", resp.ModelInfo[fmt.Sprintf("%s.context_length", arch)].(float64))},
+			[]string{"embedding length", fmt.Sprintf("%v", resp.ModelInfo[fmt.Sprintf("%s.embedding_length", arch)].(float64))},
+		)
 	}

-	tableRender("Model", func() (rows [][]string) {
-		if resp.ModelInfo != nil {
-			arch := resp.ModelInfo["general.architecture"].(string)
-			rows = append(rows, []string{"", "architecture", arch})
-			rows = append(rows, []string{"", "parameters", format.HumanNumber(uint64(resp.ModelInfo["general.parameter_count"].(float64)))})
-			rows = append(rows, []string{"", "context length", strconv.FormatFloat(resp.ModelInfo[fmt.Sprintf("%s.context_length", arch)].(float64), 'f', -1, 64)})
-			rows = append(rows, []string{"", "embedding length", strconv.FormatFloat(resp.ModelInfo[fmt.Sprintf("%s.embedding_length", arch)].(float64), 'f', -1, 64)})
-		} else {
-			rows = append(rows, []string{"", "architecture", resp.Details.Family})
-			rows = append(rows, []string{"", "parameters", resp.Details.ParameterSize})
-		}
-		rows = append(rows, []string{"", "quantization", resp.Details.QuantizationLevel})
-		return
-	})
+	mainTableData := [][]string{
+		{"Model"},
+		{renderSubTable(modelData, false)},
+	}

 	if resp.ProjectorInfo != nil {
-		tableRender("Projector", func() (rows [][]string) {
-			arch := resp.ProjectorInfo["general.architecture"].(string)
-			rows = append(rows, []string{"", "architecture", arch})
-			rows = append(rows, []string{"", "parameters", format.HumanNumber(uint64(resp.ProjectorInfo["general.parameter_count"].(float64)))})
-			rows = append(rows, []string{"", "embedding length", strconv.FormatFloat(resp.ProjectorInfo[fmt.Sprintf("%s.vision.embedding_length", arch)].(float64), 'f', -1, 64)})
-			rows = append(rows, []string{"", "dimensions", strconv.FormatFloat(resp.ProjectorInfo[fmt.Sprintf("%s.vision.projection_dim", arch)].(float64), 'f', -1, 64)})
-			return
-		})
+		projectorData := [][]string{
+			{"arch", "clip"},
+			{"parameters", format.HumanNumber(uint64(resp.ProjectorInfo["general.parameter_count"].(float64)))},
+		}
+
+		if projectorType, ok := resp.ProjectorInfo["clip.projector_type"]; ok {
+			projectorData = append(projectorData, []string{"projector type", projectorType.(string)})
+		}
+
+		projectorData = append(projectorData,
+			[]string{"embedding length", fmt.Sprintf("%v", resp.ProjectorInfo["clip.vision.embedding_length"].(float64))},
+			[]string{"projection dimensionality", fmt.Sprintf("%v", resp.ProjectorInfo["clip.vision.projection_dim"].(float64))},
+		)
+
+		mainTableData = append(mainTableData,
+			[]string{"Projector"},
+			[]string{renderSubTable(projectorData, false)},
+		)
 	}

 	if resp.Parameters != "" {
-		tableRender("Parameters", func() (rows [][]string) {
-			scanner := bufio.NewScanner(strings.NewReader(resp.Parameters))
-			for scanner.Scan() {
-				if text := scanner.Text(); text != "" {
-					rows = append(rows, append([]string{""}, strings.Fields(text)...))
-				}
-			}
-			return
-		})
-	}
-
-	head := func(s string, n int) (rows [][]string) {
-		scanner := bufio.NewScanner(strings.NewReader(s))
-		for scanner.Scan() && (len(rows) < n || n < 0) {
-			if text := scanner.Text(); text != "" {
-				rows = append(rows, []string{"", strings.TrimSpace(text)})
-			}
-		}
-		return
+		mainTableData = append(mainTableData, []string{"Parameters"}, []string{formatParams(resp.Parameters)})
 	}

 	if resp.System != "" {
-		tableRender("System", func() [][]string {
-			return head(resp.System, 2)
-		})
+		mainTableData = append(mainTableData, []string{"System"}, []string{renderSubTable(twoLines(resp.System), true)})
 	}

 	if resp.License != "" {
-		tableRender("License", func() [][]string {
-			return head(resp.License, 2)
-		})
+		mainTableData = append(mainTableData, []string{"License"}, []string{renderSubTable(twoLines(resp.License), true)})
 	}

-	return nil
+	table := tablewriter.NewWriter(os.Stdout)
+	table.SetAutoWrapText(false)
+	table.SetBorder(false)
+	table.SetAlignment(tablewriter.ALIGN_LEFT)
+
+	for _, v := range mainTableData {
+		table.Append(v)
+	}
+
+	table.Render()
+}
+
+func renderSubTable(data [][]string, file bool) string {
+	var buf bytes.Buffer
+	table := tablewriter.NewWriter(&buf)
+	table.SetAutoWrapText(!file)
+	table.SetBorder(false)
+	table.SetNoWhiteSpace(true)
+	table.SetTablePadding("\t")
+	table.SetAlignment(tablewriter.ALIGN_LEFT)
+
+	for _, v := range data {
+		table.Append(v)
+	}
+
+	table.Render()
+
+	renderedTable := buf.String()
+	lines := strings.Split(renderedTable, "\n")
+	for i, line := range lines {
+		lines[i] = "\t" + line
+	}
+
+	return strings.Join(lines, "\n")
+}
+
+func twoLines(s string) [][]string {
+	lines := strings.Split(s, "\n")
+	res := [][]string{}
+
+	count := 0
+	for _, line := range lines {
+		line = strings.TrimSpace(line)
+		if line != "" {
+			count++
+			res = append(res, []string{line})
+			if count == 2 {
+				return res
+			}
+		}
+	}
+	return res
+}
+
+func formatParams(s string) string {
+	lines := strings.Split(s, "\n")
+	table := [][]string{}
+
+	for _, line := range lines {
+		table = append(table, strings.Fields(line))
+	}
+	return renderSubTable(table, false)
 }

 func CopyHandler(cmd *cobra.Command, args []string) error {
@@ -1335,15 +1328,6 @@ func NewCLI() *cobra.Command {
 	runCmd.Flags().Bool("insecure", false, "Use an insecure registry")
 	runCmd.Flags().Bool("nowordwrap", false, "Don't wrap words to the next line automatically")
 	runCmd.Flags().String("format", "", "Response format (e.g. json)")
-
-	stopCmd := &cobra.Command{
-		Use:     "stop MODEL",
-		Short:   "Stop a running model",
-		Args:    cobra.ExactArgs(1),
-		PreRunE: checkServerHeartbeat,
-		RunE:    StopHandler,
-	}
-
 	serveCmd := &cobra.Command{
 		Use:     "serve",
 		Aliases: []string{"start"},
@@ -1411,7 +1395,6 @@ func NewCLI() *cobra.Command {
 		createCmd,
 		showCmd,
 		runCmd,
-		stopCmd,
 		pullCmd,
 		pushCmd,
 		listCmd,
@@ -1438,8 +1421,6 @@ func NewCLI() *cobra.Command {
 				envVars["OLLAMA_TMPDIR"],
 				envVars["OLLAMA_FLASH_ATTENTION"],
 				envVars["OLLAMA_LLM_LIBRARY"],
-				envVars["OLLAMA_GPU_OVERHEAD"],
-				envVars["OLLAMA_LOAD_TIMEOUT"],
 			})
 		default:
 			appendEnvDocs(cmd, envs)
@@ -1451,7 +1432,6 @@ func NewCLI() *cobra.Command {
 		createCmd,
 		showCmd,
 		runCmd,
-		stopCmd,
 		pullCmd,
 		pushCmd,
 		listCmd,
--- a/cmd/cmd_test.go
+++ b/cmd/cmd_test.go
@@ -1,206 +0,0 @@
-package cmd
-
-import (
-	"bytes"
-	"os"
-	"path/filepath"
-	"testing"
-
-	"github.com/google/go-cmp/cmp"
-
-	"github.com/ollama/ollama/api"
-)
-
-func TestShowInfo(t *testing.T) {
-	t.Run("bare details", func(t *testing.T) {
-		var b bytes.Buffer
-		if err := showInfo(&api.ShowResponse{
-			Details: api.ModelDetails{
-				Family:            "test",
-				ParameterSize:     "7B",
-				QuantizationLevel: "FP16",
-			},
-		}, &b); err != nil {
-			t.Fatal(err)
-		}
-
-		expect := `  Model
-    architecture    test    
-    parameters      7B      
-    quantization    FP16    
-
-`
-
-		if diff := cmp.Diff(expect, b.String()); diff != "" {
-			t.Errorf("unexpected output (-want +got):\n%s", diff)
-		}
-	})
-
-	t.Run("bare model info", func(t *testing.T) {
-		var b bytes.Buffer
-		if err := showInfo(&api.ShowResponse{
-			ModelInfo: map[string]any{
-				"general.architecture":    "test",
-				"general.parameter_count": float64(7_000_000_000),
-				"test.context_length":     float64(0),
-				"test.embedding_length":   float64(0),
-			},
-			Details: api.ModelDetails{
-				Family:            "test",
-				ParameterSize:     "7B",
-				QuantizationLevel: "FP16",
-			},
-		}, &b); err != nil {
-			t.Fatal(err)
-		}
-
-		expect := `  Model
-    architecture        test    
-    parameters          7B      
-    context length      0       
-    embedding length    0       
-    quantization        FP16    
-
-`
-		if diff := cmp.Diff(expect, b.String()); diff != "" {
-			t.Errorf("unexpected output (-want +got):\n%s", diff)
-		}
-	})
-
-	t.Run("parameters", func(t *testing.T) {
-		var b bytes.Buffer
-		if err := showInfo(&api.ShowResponse{
-			Details: api.ModelDetails{
-				Family:            "test",
-				ParameterSize:     "7B",
-				QuantizationLevel: "FP16",
-			},
-			Parameters: `
-			stop never
-			stop gonna
-			stop give
-			stop you
-			stop up
-			temperature 99`,
-		}, &b); err != nil {
-			t.Fatal(err)
-		}
-
-		expect := `  Model
-    architecture    test    
-    parameters      7B      
-    quantization    FP16    
-
-  Parameters
-    stop           never    
-    stop           gonna    
-    stop           give     
-    stop           you      
-    stop           up       
-    temperature    99       
-
-`
-		if diff := cmp.Diff(expect, b.String()); diff != "" {
-			t.Errorf("unexpected output (-want +got):\n%s", diff)
-		}
-	})
-
-	t.Run("project info", func(t *testing.T) {
-		var b bytes.Buffer
-		if err := showInfo(&api.ShowResponse{
-			Details: api.ModelDetails{
-				Family:            "test",
-				ParameterSize:     "7B",
-				QuantizationLevel: "FP16",
-			},
-			ProjectorInfo: map[string]any{
-				"general.architecture":         "clip",
-				"general.parameter_count":      float64(133_700_000),
-				"clip.vision.embedding_length": float64(0),
-				"clip.vision.projection_dim":   float64(0),
-			},
-		}, &b); err != nil {
-			t.Fatal(err)
-		}
-
-		expect := `  Model
-    architecture    test    
-    parameters      7B      
-    quantization    FP16    
-
-  Projector
-    architecture        clip       
-    parameters          133.70M    
-    embedding length    0          
-    dimensions          0          
-
-`
-		if diff := cmp.Diff(expect, b.String()); diff != "" {
-			t.Errorf("unexpected output (-want +got):\n%s", diff)
-		}
-	})
-
-	t.Run("system", func(t *testing.T) {
-		var b bytes.Buffer
-		if err := showInfo(&api.ShowResponse{
-			Details: api.ModelDetails{
-				Family:            "test",
-				ParameterSize:     "7B",
-				QuantizationLevel: "FP16",
-			},
-			System: `You are a pirate!
-Ahoy, matey!
-Weigh anchor!
-			`,
-		}, &b); err != nil {
-			t.Fatal(err)
-		}
-
-		expect := `  Model
-    architecture    test    
-    parameters      7B      
-    quantization    FP16    
-
-  System
-    You are a pirate!    
-    Ahoy, matey!         
-
-`
-		if diff := cmp.Diff(expect, b.String()); diff != "" {
-			t.Errorf("unexpected output (-want +got):\n%s", diff)
-		}
-	})
-
-	t.Run("license", func(t *testing.T) {
-		var b bytes.Buffer
-		license, err := os.ReadFile(filepath.Join("..", "LICENSE"))
-		if err != nil {
-			t.Fatal(err)
-		}
-
-		if err := showInfo(&api.ShowResponse{
-			Details: api.ModelDetails{
-				Family:            "test",
-				ParameterSize:     "7B",
-				QuantizationLevel: "FP16",
-			},
-			License: string(license),
-		}, &b); err != nil {
-			t.Fatal(err)
-		}
-
-		expect := `  Model
-    architecture    test    
-    parameters      7B      
-    quantization    FP16    
-
-  License
-    MIT License             
-    Copyright (c) Ollama    
-
-`
-		if diff := cmp.Diff(expect, b.String()); diff != "" {
-			t.Errorf("unexpected output (-want +got):\n%s", diff)
-		}
-	})
-}
--- a/cmd/interactive.go
+++ b/cmd/interactive.go
@@ -18,6 +18,7 @@ import (
 	"github.com/ollama/ollama/api"
 	"github.com/ollama/ollama/envconfig"
 	"github.com/ollama/ollama/parser"
+	"github.com/ollama/ollama/progress"
 	"github.com/ollama/ollama/readline"
 	"github.com/ollama/ollama/types/errtypes"
 )
@@ -30,6 +31,26 @@ const (
 	MultilineSystem
 )

+func loadModel(cmd *cobra.Command, opts *runOptions) error {
+	p := progress.NewProgress(os.Stderr)
+	defer p.StopAndClear()
+
+	spinner := progress.NewSpinner("")
+	p.Add("", spinner)
+
+	client, err := api.ClientFromEnvironment()
+	if err != nil {
+		return err
+	}
+
+	chatReq := &api.ChatRequest{
+		Model:     opts.Model,
+		KeepAlive: opts.KeepAlive,
+	}
+
+	return client.Chat(cmd.Context(), chatReq, func(api.ChatResponse) error { return nil })
+}
+
 func generateInteractive(cmd *cobra.Command, opts runOptions) error {
 	usage := func() {
 		fmt.Fprintln(os.Stderr, "Available Commands:")
@@ -196,7 +217,7 @@ func generateInteractive(cmd *cobra.Command, opts runOptions) error {
 			opts.Model = args[1]
 			opts.Messages = []api.Message{}
 			fmt.Printf("Loading model '%s'\n", opts.Model)
-			if err := loadOrUnloadModel(cmd, &opts); err != nil {
+			if err := loadModel(cmd, &opts); err != nil {
 				return err
 			}
 			continue
@@ -350,7 +371,7 @@ func generateInteractive(cmd *cobra.Command, opts runOptions) error {

 				switch args[1] {
 				case "info":
-					_ = showInfo(resp, os.Stderr)
+					showInfo(resp)
 				case "license":
 					if resp.License == "" {
 						fmt.Println("No license was specified for this model.")
--- a/convert/convert.go
+++ b/convert/convert.go
@@ -208,18 +208,14 @@ func ConvertModel(fsys fs.FS, ws io.WriteSeeker) error {
 		return err
 	}

-	vocabSize := int(p.VocabSize)
-	switch {
-	case vocabSize > len(t.Vocabulary.Tokens):
-		slog.Warn("vocabulary is smaller than expected, padding with dummy tokens", "expect", vocabSize, "actual", len(t.Vocabulary.Tokens))
+	if vocabSize := int(p.VocabSize); vocabSize > len(t.Vocabulary.Tokens) {
+		slog.Warn("vocabulary is smaller than expected, padding with dummy tokens", "expect", p.VocabSize, "actual", len(t.Vocabulary.Tokens))
 		for i := range vocabSize - len(t.Vocabulary.Tokens) {
 			t.Vocabulary.Tokens = append(t.Vocabulary.Tokens, fmt.Sprintf("[PAD%d]", i))
 			t.Vocabulary.Scores = append(t.Vocabulary.Scores, -1)
 			t.Vocabulary.Types = append(t.Vocabulary.Types, tokenTypeUserDefined)
 		}
-	case vocabSize < len(t.Vocabulary.Tokens):
-		return fmt.Errorf("vocabulary is larger than expected '%d' instead of '%d'", len(t.Vocabulary.Tokens), vocabSize)
-	default:
+	} else {
 		slog.Debug("vocabulary", "size", len(t.Vocabulary.Tokens))
 	}

--- a/convert/convert_gemma2.go
+++ b/convert/convert_gemma2.go
@@ -34,20 +34,10 @@ func (p *gemma2Model) KV(t *Tokenizer) llm.KV {
 }

 func (p *gemma2Model) Replacements() []string {
-	return []string{
-		"model.embed_tokens", "token_embd",
-		"model.norm", "output_norm",
-		"model.layers", "blk",
-		"input_layernorm", "attn_norm",
-		"self_attn.q_proj", "attn_q",
-		"self_attn.k_proj", "attn_k",
-		"self_attn.v_proj", "attn_v",
-		"self_attn.o_proj", "attn_output",
-		"mlp.gate_proj", "ffn_gate",
-		"mlp.down_proj", "ffn_down",
-		"mlp.up_proj", "ffn_up",
+	return append(
+		p.gemmaModel.Replacements(),
 		"post_attention_layernorm", "post_attention_norm",
 		"pre_feedforward_layernorm", "ffn_norm",
 		"post_feedforward_layernorm", "post_ffw_norm",
-	}
+	)
 }
--- a/convert/convert_test.go
+++ b/convert/convert_test.go
@@ -15,7 +15,6 @@ import (
 	"os"
 	"path/filepath"
 	"slices"
-	"strings"
 	"testing"

 	"golang.org/x/exp/maps"
@@ -23,12 +22,6 @@ import (
 	"github.com/ollama/ollama/llm"
 )

-type tensorData struct {
-	Offsets []int  `json:"data_offsets"`
-	Type    string `json:"dtype"`
-	Shape   []int  `json:"shape"`
-}
-
 func convertFull(t *testing.T, fsys fs.FS) (*os.File, llm.KV, llm.Tensors) {
 	t.Helper()

@@ -103,7 +96,6 @@ func TestConvertModel(t *testing.T) {
 		"Mistral-7B-Instruct-v0.2",
 		"Mixtral-8x7B-Instruct-v0.1",
 		"gemma-2b-it",
-		"gemma-2-2b-it",
 		// microsoft/Phi-3-mini-128-instruct@d548c233192db00165d842bf8edff054bb3212f8
 		"Phi-3-mini-128k-instruct",
 		"all-MiniLM-L6-v2",
@@ -148,36 +140,6 @@ func TestConvertModel(t *testing.T) {
 	}
 }

-func TestConvertInvalidTensorNames(t *testing.T) {
-	f, err := os.CreateTemp(t.TempDir(), "testmodel")
-	if err != nil {
-		t.Fatal(err)
-	}
-	defer f.Close()
-
-	tempDir := t.TempDir()
-
-	td := map[string]*tensorData{}
-	offset := 4096
-
-	td["model.layers.0.self_attn.q_proj.weight"] = &tensorData{
-		Offsets: []int{0, offset},
-		Type:    "F32",
-		Shape:   []int{4096, 4096},
-	}
-	td["blk.0.attn_q.weight"] = &tensorData{
-		Offsets: []int{offset, offset * 2},
-		Type:    "F32",
-		Shape:   []int{4096, 4096},
-	}
-	generateSafetensorTestData(t, tempDir, td)
-
-	err = ConvertModel(os.DirFS(tempDir), f)
-	if err == nil || !strings.HasPrefix(err.Error(), "duplicate tensor name") {
-		t.Errorf("expected error but didn't get one")
-	}
-}
-
 func TestConvertInvalidDatatype(t *testing.T) {
 	f, err := os.CreateTemp(t.TempDir(), "testmodel")
 	if err != nil {
@@ -186,10 +148,23 @@ func TestConvertInvalidDatatype(t *testing.T) {
 	defer f.Close()

 	tempDir := t.TempDir()
+	generateSafetensorTestData(t, tempDir)

-	td := map[string]*tensorData{}
+	err = ConvertModel(os.DirFS(tempDir), f)
+	if err == nil || err.Error() != "unsupported safetensors model" {
+		t.Errorf("expected error but didn't get one")
+	}
+}
+
+func generateSafetensorTestData(t *testing.T, tempDir string) {
+	type tensorData struct {
+		Offsets []int  `json:"data_offsets"`
+		Type    string `json:"dtype"`
+		Shape   []int  `json:"shape"`
+	}
 	offset := 4096 * 14336

+	td := map[string]*tensorData{}
 	td["model.layers.0.mlp.down_proj.weight"] = &tensorData{
 		Offsets: []int{0, offset},
 		Type:    "I8",
@@ -200,16 +175,8 @@ func TestConvertInvalidDatatype(t *testing.T) {
 		Type:    "U8",
 		Shape:   []int{},
 	}
-	generateSafetensorTestData(t, tempDir, td)

-	err = ConvertModel(os.DirFS(tempDir), f)
-	if err == nil || err.Error() != "unsupported safetensors model" {
-		t.Errorf("expected error but didn't get one")
-	}
-}
-
-func generateSafetensorTestData(t *testing.T, tempDir string, tensorData map[string]*tensorData) {
-	data, err := json.Marshal(tensorData)
+	data, err := json.Marshal(td)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -355,6 +322,11 @@ func TestConvertAdapter(t *testing.T) {
 }

 func generateLoraTestData(t *testing.T, tempDir string) {
+	type tensorData struct {
+		Offsets []int  `json:"data_offsets"`
+		Type    string `json:"dtype"`
+		Shape   []int  `json:"shape"`
+	}
 	offset := 4096 * 8 * 4

 	td := map[string]*tensorData{"__metadata__": nil}
--- a/convert/reader_safetensors.go
+++ b/convert/reader_safetensors.go
@@ -49,19 +49,12 @@ func parseSafetensors(fsys fs.FS, replacer *strings.Replacer, ps ...string) ([]T
 		keys := maps.Keys(headers)
 		slices.Sort(keys)

-		names := make(map[string]struct{}, len(keys))
-
 		for _, key := range keys {
 			if value := headers[key]; value.Type != "" {
 				// bitsandbytes quantized models are unsupported
 				if len(value.Shape) == 0 {
 					return nil, errors.New("unsupported safetensors model")
 				}
-				ggufName := replacer.Replace(key)
-				if _, ok := names[ggufName]; ok {
-					return nil, fmt.Errorf("duplicate tensor name '%s' was found for this model", ggufName)
-				}
-				names[ggufName] = struct{}{}
 				ts = append(ts, safetensor{
 					fs:     fsys,
 					path:   p,
@@ -69,7 +62,7 @@ func parseSafetensors(fsys fs.FS, replacer *strings.Replacer, ps ...string) ([]T
 					offset: safetensorsPad(n, value.Offsets[0]),
 					size:   safetensorsPad(n, value.Offsets[1]) - safetensorsPad(n, value.Offsets[0]),
 					tensorBase: &tensorBase{
-						name:  ggufName,
+						name:  replacer.Replace(key),
 						shape: value.Shape,
 					},
 				})
--- a/convert/testdata/gemma-2-2b-it.json
+++ b/convert/testdata/gemma-2-2b-it.json
@@ -1,312 +0,0 @@
-{
-  "general.architecture": "gemma2",
-  "general.file_type": "1",
-  "general.quantization_version": "2",
-  "gemma2.block_count": "26",
-  "gemma2.context_length": "8192",
-  "gemma2.embedding_length": "2304",
-  "gemma2.feed_forward_length": "9216",
-  "gemma2.attention.head_count": "8",
-  "gemma2.attention.head_count_kv": "4",
-  "gemma2.attention.key_length": "256",
-  "gemma2.attention.value_length": "256",
-  "gemma2.attention.layer_norm_rms_epsilon": "1e-06",
-  "tokenizer.ggml.model": "llama",
-  "tokenizer.ggml.add_bos_token": "true",
-  "tokenizer.ggml.add_eos_token": "false",
-  "tokenizer.ggml.bos_token_id": "2",
-  "tokenizer.ggml.eos_token_id": "1",
-  "tokenizer.ggml.padding_token_id": "0",
-  "tokenizer.ggml.unknown_token_id": "3",
-  "tokenizer.ggml.scores": "0872465d173867d755d3ee728f882b9dc2057a0bfd596fe1e3d131522f1250d8",
-  "tokenizer.ggml.token_type": "8d40143b3477df77beea4139420335ede458bf5e14102f01b0170197b55da8d8",
-  "tokenizer.ggml.tokens": "c6e66de1841f04de8b8d236d461ab720a4c9b9b5414dc293a09c6e10eab45fda",
-  "token_embd.weight": "64a9d30707e659e2e673656d71f5aef7a9fb9fd83bb9a77558dfc5abbe218a05",
-  "blk.0.attn_k.weight": "d8b4437c5edb3cddf6af9987038e1bb2b191c4f0fce0e160d2abace717f5d5d7",
-  "blk.0.attn_norm.weight": "1eb73e3f7aa8e502f6ca31cd19efbb8e4fd9a89692e13e48ac8205545a7fa7e8",
-  "blk.0.attn_output.weight": "39e7b78e57d356a22dd89ce1c4d7163b970712ba756545e1703f97866cd2192e",
-  "blk.0.attn_q.weight": "795058e23b6109febd9d55c89e1eebe6af0714ec8c56fd86a160876a6135ffe8",
-  "blk.0.attn_v.weight": "0cd6e583d1887c020472e961bbb113fe5a0d23ae2f1c2c876fc366cdb7692b52",
-  "blk.0.ffn_down.weight": "51eb4d962189e945a84e94e0dc1aad3f8f90cc1a11e18029670afcd0ea0acb1b",
-  "blk.0.ffn_gate.weight": "9811a29b8ad48432925897ab21dfcb13c5cbd372aeccbbefca9b7866883b4ce3",
-  "blk.0.ffn_norm.weight": "92cbf4652ef503c1de5b10f2be00b3fcf00100980cb3baa8f3013a8d8bf3d851",
-  "blk.0.ffn_up.weight": "af87de21746879483ed1b374cdd76b19ba11ca2b6dbb1beba98efdf3be3e8077",
-  "blk.0.post_attention_norm.weight": "32e135f1f258ffe407018899e39af1725d59d66d60022b9a21575ba160e0357a",
-  "blk.0.post_ffw_norm.weight": "ba286f5ac11b07fbc986173708c66f1920427be5a6d108af38fa0a837c1c8eb6",
-  "blk.1.attn_k.weight": "51584435552051f7fade76beca582b3f7190cf7fc07adcf527c2774d4b1c3901",
-  "blk.1.attn_norm.weight": "6833104c7fbf35a7e799ae56c262b97fffa14789642aee14381b25acd21ed80a",
-  "blk.1.attn_output.weight": "14c39481369087bf292ac9a3ab2ef166f9fe376a9f90c246653213ef264febdc",
-  "blk.1.attn_q.weight": "443f64ae2229f857c69d6bebb7800b685786cb77884c3ae19d4286aeed081325",
-  "blk.1.attn_v.weight": "0df482de2038f1e4c8a7733ac0ddb69ad90759dab5968b942af0155588de4c4a",
-  "blk.1.ffn_down.weight": "66f30763a8bbbcaea609a0087ed75fadb5e771c06378dd2cea94cf17e492e8cf",
-  "blk.1.ffn_gate.weight": "a7151bff00a545fa18b2c92dcd2a14572ccf9beb957a6c494f1374e8ebe174c9",
-  "blk.1.ffn_norm.weight": "e197d71ea11b5276bc0167d2663b88089b3ff42b47ba91e85f6c5d95f6306435",
-  "blk.1.ffn_up.weight": "57c182e0b14cccd1350d388f0c616991702e74281db54637451b70f4ccc24f9b",
-  "blk.1.post_attention_norm.weight": "3c56f837168d784c2d8bac247c130bdca6610c095c8da4558c536ccad7605609",
-  "blk.1.post_ffw_norm.weight": "d2a51d320fd01069dd7ccaa7082f16a7faeb671885607d7900b10a89c354d0fa",
-  "blk.2.attn_k.weight": "bc103c818192de7ce36caaf89dc117be4df13fb902e0bd9a23c64edace5df9b6",
-  "blk.2.attn_norm.weight": "0f2503aa126083a5d6ac72481be1ef66c6014705b573682b35bd864e4749a3d5",
-  "blk.2.attn_output.weight": "05fcd4a1226e482f91803a266f72caca887a93e63c2d2ba5611ab3c68d38743a",
-  "blk.2.attn_q.weight": "6a10b5c2fd423d1e4c4fd60fa8c154a0159b6b2501ea79cae2ef19f45a674e5e",
-  "blk.2.attn_v.weight": "3cf891945a1f8ae7cc908a5c6b729ff5b70f4436c5ffdbf245cc0ed4cc19cd1b",
-  "blk.2.ffn_down.weight": "ea204fd04e0d2fc728a9861a459216bbfec629c152004ba625f52cd8837bd51e",
-  "blk.2.ffn_gate.weight": "3a3518729f1b8b64a82b8792f33987db5418fdb094be0263c68f146a5c38de54",
-  "blk.2.ffn_norm.weight": "754ede678b725de41a34b82f0edf7688b5c065be7c0d46df6f7ad9430d986884",
-  "blk.2.ffn_up.weight": "ffdcb88439f5828ffbd9fc844b03ff91637b790b9838097258cc3ae75935720c",
-  "blk.2.post_attention_norm.weight": "4b3f53b7ba26e8c36b2dfda3b7e5fc4b1065257cefdea235fc7df9af130ac2fd",
-  "blk.2.post_ffw_norm.weight": "e550369e26b8485e2b54ad34b34bc98af5494287dcc513c2c39cf1eaa5b89d07",
-  "blk.3.attn_k.weight": "89f24ea450e37d9e95757651a83205c085d81b354ee9489dd6310a391d8409f3",
-  "blk.3.attn_norm.weight": "24e2ea662b7cb822b4ca5cd61bc17f2709f406d990ec3b4a0dac1cc112db45cf",
-  "blk.3.attn_output.weight": "ac4dad69473c6e3fac56669212cadd8c34ecc5973d945972e974d94805334967",
-  "blk.3.attn_q.weight": "b6a9c9a7d4722b9096631c65de62228dfddca6e26edfe6af7fce01e116ef0f4c",
-  "blk.3.attn_v.weight": "f272a960a40093942309bc342a379984cbacec2d7bc64428db3f64e6b1887ed4",
-  "blk.3.ffn_down.weight": "c0188ba50d8228805982029c277fc0e87aa57473b8363037c648f6d006ff828a",
-  "blk.3.ffn_gate.weight": "a04aec1561ee6c0fbb18c3db49dc62fb533619cf697fd548cbf2279761aaec3b",
-  "blk.3.ffn_norm.weight": "bc053837d44087ec05eb5d9458357b2a5be787789b19cdbbdc694b57697f99a6",
-  "blk.3.ffn_up.weight": "b3ce8b274f20796d3b1a7c08ba27a919066f9de89a782faa544c4a8d6bea1382",
-  "blk.3.post_attention_norm.weight": "9c922dee7a7df5667289e2788e60170238239cee2dfdbbd9e435763f9f416718",
-  "blk.3.post_ffw_norm.weight": "b682544ac953ad2e0b49027ed8916f2e9d1aba5d1587bb4127ac703570c7a03a",
-  "blk.4.attn_k.weight": "143b0cbb4b787b95c2b6212374410e32173ccef2adb914908a2f89a7916de512",
-  "blk.4.attn_norm.weight": "5668f60491b780273745192662d02c9a92a4f692b29d16aa0bbc7413fec4f85b",
-  "blk.4.attn_output.weight": "b9f2bdb68be1e0cf66dd19f8fa2afb105910ad2ef394864cb32cea8f8944e0d5",
-  "blk.4.attn_q.weight": "ddcf1343dafbc2dfcd0b8741225af22fe4b54b2becce29240bd01c34265d126c",
-  "blk.4.attn_v.weight": "6dc7074366e7ed52d9f48c594dcc85bef738e096276cb99d28228c89eecc5b9c",
-  "blk.4.ffn_down.weight": "30334ffc59ce343cf2a1b973174acb7722823463adc07e19a99bd0f404bc9906",
-  "blk.4.ffn_gate.weight": "890f7c8af208d63b28db52c4b8c16c2288a382d87ff5a6a6d6b0a5b3bf27e6cd",
-  "blk.4.ffn_norm.weight": "ff0316cc7847221eb86a90c1ab441d4ee61553d410c66414a7755021b3b12448",
-  "blk.4.ffn_up.weight": "6af97d113f91564c636734f215e25ee602d48eb045458f300b3ec7582be0f41d",
-  "blk.4.post_attention_norm.weight": "69438f231e105e68216b078bdeb35a7cdc8b12c4e2845e18ecf4c8d361d6a321",
-  "blk.4.post_ffw_norm.weight": "0fd535da78bcf2b32c95b05b2b83dc49817393765be90d8cc1ed3d56f47b68ec",
-  "blk.5.attn_k.weight": "0166eb3c6d20dcf3d3c169e94caa8dee057535bb525e29f698fb6f8844f18a6c",
-  "blk.5.attn_norm.weight": "a7808f27f164023d5cde2be00fc23cac6c71aa0ddeb60bc23e12411b80087672",
-  "blk.5.attn_output.weight": "8b65b2027a0842b68c5308f91d6a31de9599d794157d77df8418b19f9e0d9334",
-  "blk.5.attn_q.weight": "966bc626ef2c2394d872087a41c126bb1b67d1d5f6de920204ef5e5b16c34003",
-  "blk.5.attn_v.weight": "9a362aef3f4437fbf0ef6e1ba785f3329c3db2960f93fe36547d2795e9c254ea",
-  "blk.5.ffn_down.weight": "63e53541d34197720c06f297aa8142ac6b6eec002c7987b296f26e8b1400f931",
-  "blk.5.ffn_gate.weight": "d9591fdd32f783e0fc26e20d5d587ee8971ac8ae2e4c818c6eac1c125c7c7f37",
-  "blk.5.ffn_norm.weight": "677334cc60ecce3a7f4ab3acda15d359353d7358872f614ad8914e3780e9fc6e",
-  "blk.5.ffn_up.weight": "a63764110e1c655ffbd55af0669b2dfe4cc29d0e198d33a8e5426461b08a85f7",
-  "blk.5.post_attention_norm.weight": "c55499f859b2c0a7f5cabceaae47309a5ad38bc29d0f4a8db81f1357023162a9",
-  "blk.5.post_ffw_norm.weight": "82752754665f842418f3e302cb5f43d1e0504dcd124c4b8ddb77018b2c793837",
-  "blk.6.attn_k.weight": "e20a5f0d6c807273c8d491439566b428497ac02097cf0aa55e33748c28e14be6",
-  "blk.6.attn_norm.weight": "2c6ba42fd3c73d72073ced03a32dd28d70a89ed9bbbc8fea1ba03a7ade951e6c",
-  "blk.6.attn_output.weight": "4de7c5c2f4a133a266e17ed8c14c52959466b54cc7ab9e19f789a33b4850f284",
-  "blk.6.attn_q.weight": "56462d921800e6b8cd2213fef04c4ff16d728905cb2f4c58e966d0a053a3b0ae",
-  "blk.6.attn_v.weight": "b758dcbff769d6240c2245ede1dbc62c4170a67c77458e866312589220fe29af",
-  "blk.6.ffn_down.weight": "582247fb3c2bf687cbe9413fe18d18ad47bef4b65df7d78905e10335c6134764",
-  "blk.6.ffn_gate.weight": "3035444d5286aefb7a6d04e55bc27e1fac7cf895cd5be02319a431b8e047b4ae",
-  "blk.6.ffn_norm.weight": "e582d24c66e01b96faa20ce6adfda3d8583b11e809bff89969927398175e369a",
-  "blk.6.ffn_up.weight": "6f4b7bbfedeacf61a4866ae0616c4ba6c9e856662e8f00ae6aaec7f52c53e7b4",
-  "blk.6.post_attention_norm.weight": "8fe51b50bd677d21586aecab0b565c4bf9fa68ad50bfe366f45e8fea3c657ca8",
-  "blk.6.post_ffw_norm.weight": "81ba3cb4c2bf5c546b86855b7a885d3fafededc67eb3a35cd3598b03c9e26e65",
-  "blk.7.attn_k.weight": "2e044179cdcae0946708c86bfea7aa0391e1f7e2a09b33fca035d384cc3ca758",
-  "blk.7.attn_norm.weight": "94b48c546b046803c60e75a3acb17a356b710735989938021b565f68df9b4985",
-  "blk.7.attn_output.weight": "65709b4ad7a581f4d75793d39d4032a359f6bcc0c3835205242a0b99e5b66824",
-  "blk.7.attn_q.weight": "8ded993c95d1f7caf201ceb6fa035cd6ed6d351b50b999fa9355dfee9486cb5b",
-  "blk.7.attn_v.weight": "c92d5e2d2d48397542bc03bea25bf39154075e66c5bb1ead85188505aa04ae91",
-  "blk.7.ffn_down.weight": "e8ba8fb57208805ef1dc23cd7c86e9a2d1fb7c52c3940d292cd5bb2eb24b3fac",
-  "blk.7.ffn_gate.weight": "f0f06d6a2e06c5ac252083bc61d05c814e6289d3f4e4a87d2f06918254c02c36",
-  "blk.7.ffn_norm.weight": "ebf8ef775f72624148e09d68a4332187a7a5020c521fe0623da1cd3485ad33e0",
-  "blk.7.ffn_up.weight": "a554adc4fc7122c247c77670e169916ba1794c787b5be30a2b36705138f1f746",
-  "blk.7.post_attention_norm.weight": "3aa6bc21d85c3a0c12b964e82b12feaedfdd13130c3cd2229228e24e0967ebdf",
-  "blk.7.post_ffw_norm.weight": "508bc7b19ee8ff08f0007c890133a462fc57c7e72b16ee8f6dd64def264ef876",
-  "blk.8.attn_k.weight": "363c8e74056642fe9e7c2f3f9769d57319cd3fa0a6022810189ab8d894322885",
-  "blk.8.attn_norm.weight": "685b49a1f1acb169f4df0bdd8e3de6943f3033cebad14b898a72000595610d92",
-  "blk.8.attn_output.weight": "7bde571e4efef1c6a6143f0526721dfb59e0a0ea0e1a3616a322b2eb937efa48",
-  "blk.8.attn_q.weight": "fc993dbc1074c28a0e1d85e5ab2f4ea6a9c6c1affe7ee56027000a275daed9b6",
-  "blk.8.attn_v.weight": "281e8791d3aef9b3864f1cb054da0ae0c2fef4ce0a58b1bad8bc136b2fa0f62b",
-  "blk.8.ffn_down.weight": "b1164a2578a7f87ed99c2bbc76c5dfbbbc6a1a803605391acc3f320fc989ffd7",
-  "blk.8.ffn_gate.weight": "6b39a3b3aaaa79aee61416b54d62160b9258042650e61c6b47bc77c2dd17daf3",
-  "blk.8.ffn_norm.weight": "17ea1362c72da27f12bc936500492035bdef3fd8f940cb12b57f37d42ba8ecb1",
-  "blk.8.ffn_up.weight": "bc3a7c47afc440d2bdf8fbe9ddf2c9220467472c60c8b4ded8c0f181470ec96c",
-  "blk.8.post_attention_norm.weight": "5c506204e00411ef9c8b4134d40eedcc19fffe68dd0af7d7cc49dcabf2dfac7e",
-  "blk.8.post_ffw_norm.weight": "002faec235c3678864e2901eed275ce4e9dc229164a91c9cd4c965142ba62305",
-  "blk.9.attn_k.weight": "0bab39d8c237f1b6d0010db40467142625a9e6f2e0e4c49a56c12b41e4e0b1fa",
-  "blk.9.attn_norm.weight": "de5f38e873b17f07aa7598831b89cc1cae2c9bc3eb2e042ee9af059d2563e84e",
-  "blk.9.attn_output.weight": "8a8184702c25a62df9ff309c0c7badc8587208523b2be3e8fa90ce7080573e6f",
-  "blk.9.attn_q.weight": "7c961b2431b09ddf95377acd07201cb91bf13d9cd3ae0f2c25c7d6a0358d9f50",
-  "blk.9.attn_v.weight": "e22d240cb4743067033e659cbf210ebe2ebbab3e1dea6ccbe5eaa982382ca038",
-  "blk.9.ffn_down.weight": "a426f81210f03d6ad53277416e1fdcdf37d8065e4817613edaf6c67a343426be",
-  "blk.9.ffn_gate.weight": "a82eba825cb77b8e64f85ff99ede2fc71bc9b01751eeb17e9e6c246ee12ea62e",
-  "blk.9.ffn_norm.weight": "1a97f9b1302a3a326d534c5c3fed2db6db0ae45fd0edd381a3e4fc1c75d81030",
-  "blk.9.ffn_up.weight": "5f20bac2bbf03bb42adb92fbf99561651e1edda57e0b61935ac7f6c08c0ed7cb",
-  "blk.9.post_attention_norm.weight": "9f9866d13988e1946b1e1c80d9374a92a6e3be33748f8eaed3e126d1e1a4c796",
-  "blk.9.post_ffw_norm.weight": "a6896dbf698db4dbbe5dbf12417d4fd80e9cad0c539c858892ec0aa5b046bb58",
-  "blk.10.attn_k.weight": "ca8446e5d21ecd4e6a70dca8d321be480be4fba94d70cba065205436feb44270",
-  "blk.10.attn_norm.weight": "4f41fe290e8f21f63b82151b6cce94bf7318d121468816b0c58af0ff7c1658ab",
-  "blk.10.attn_output.weight": "c626d2e9681c5c941bbde43dddfae1a8d4986bf2be4470857bc8e8bd7f869044",
-  "blk.10.attn_q.weight": "1e61b210a13a429977325cf15d781ab77d604cfa862f4270329cbd94237d5835",
-  "blk.10.attn_v.weight": "8ff8d3e3f058ec3b35ada1057f2ed59c06494d0e0be6a8dc3ff9edf9f0e1a115",
-  "blk.10.ffn_down.weight": "bcebc04219f8081a5f483e58103c0ddbbbc631a0a54fd6dd9d55778e041f70ee",
-  "blk.10.ffn_gate.weight": "7a23a1e620ef871384ddf9611ccdcfb893fbf013cc203ac8e72f745420f1eea0",
-  "blk.10.ffn_norm.weight": "e3a375e43c349a1c6c66c22328e513cc1af3137fe839e43dc8e9be2f65914fd7",
-  "blk.10.ffn_up.weight": "5d182e7c94369194fca5f19cbbe668a999911e57f3d363bc7fb6088428700cb9",
-  "blk.10.post_attention_norm.weight": "b841c6308296e8984f3c5f549c6e3a242f4b3e19141e1f54cc08de9c46759c09",
-  "blk.10.post_ffw_norm.weight": "9d66fa05b5c940208f634f5053d809094c99a2a10a1d1e8847c8281fbd99fb49",
-  "blk.11.attn_k.weight": "14adf24ebb2bb17b336ca81cec3e690fd854782f4440ca6c66cc1d7e7bf1c850",
-  "blk.11.attn_norm.weight": "2d2213f311f50414702b5b34f22aafb9d9a0b6787243e7578562583dc40ad195",
-  "blk.11.attn_output.weight": "de1f14cc2a7fff00cf11b229f0576999205f17b9536e97abc9d6de3cc79a7884",
-  "blk.11.attn_q.weight": "2bcc5c147524003109ece0be08b89ac8b25baa71416ffa76573c6c052ffc6eea",
-  "blk.11.attn_v.weight": "2e6ab8573070c22dc1e0d7aebe4d52123226dacf7822dcce06fadbb38fb036a4",
-  "blk.11.ffn_down.weight": "1b86902f4e36868421e5228b9445051f8290b292df22a6d1af836dcecc1f25c3",
-  "blk.11.ffn_gate.weight": "e756e8081bd0a16aea4a9ef5076ad102113524f7a3d50a3a77aaa7f7938b63e8",
-  "blk.11.ffn_norm.weight": "6913887267be227cf9d1991a3dd8db2e7e74bb9b5fbdfcb9ac954fd7d7b95b3b",
-  "blk.11.ffn_up.weight": "619a3ac0609ebdf42c3fb2b6e4b1db48df79e6dd8418d7ab8f1bbff13d8a6a50",
-  "blk.11.post_attention_norm.weight": "e4b4ba92cef7b6a78407e8ab1b0307d47dac6c3df7b6817e28038317ff662d7e",
-  "blk.11.post_ffw_norm.weight": "40aceeec58cb855f0c158c9cc217168fcd5d0e735567d587217b1d78df17bc5f",
-  "blk.12.attn_k.weight": "c54c5a4d4892522022d1aa2204cfc624f0b4042caa536e678967316293fe5cb1",
-  "blk.12.attn_norm.weight": "7cd2ef58298569ffdf244d9b390f3917245276c8206e5780af5f96d8c0bbb446",
-  "blk.12.attn_output.weight": "85495ef9cc8b3deb21f741bde463ff6493acae2be51f02ecdeef952cbdec3375",
-  "blk.12.attn_q.weight": "d19383f83fd119bfb8c0280c9515705c11d8e7d502019fcf8f49efeef0d106d0",
-  "blk.12.attn_v.weight": "869ac669ba49531d9128892a0e27cef15de508ff40cdf80cc1681dde50d09204",
-  "blk.12.ffn_down.weight": "578f39f8f9fc2f09138afc884a952d7cc3a9a31de4216acd10e88e19e0b75f8c",
-  "blk.12.ffn_gate.weight": "e29a0186bc6c4a0720246306e922d3a83f777dadcf4ac80bad468287031cc8b5",
-  "blk.12.ffn_norm.weight": "e1ee95c6584b5cb57fcf1db8ce2bcc03aff91eb389238c094a61c00dde93d1f2",
-  "blk.12.ffn_up.weight": "2a826f06d7cdfb3edc6ae250ff44363ef77a2a9cdf96313e23a331b99ebfa17d",
-  "blk.12.post_attention_norm.weight": "4bafc7699b948d5cbc0d3e09b418b06c6abc4651a61ada9609d9a2f21c7e5607",
-  "blk.12.post_ffw_norm.weight": "bbb8c34a7176bb1a49f9fe2bacca0bd26b673d52c0835b2e90fa11f2962f077f",
-  "blk.13.attn_k.weight": "ffeefccfe8255d1b694382012ff4134eee5fec9d9491c8d0ff0a13832d1a37e8",
-  "blk.13.attn_norm.weight": "35713726529e3887c4135a88e86e8a4d7270ba5b9f2d1ab462622fbf40a7cdce",
-  "blk.13.attn_output.weight": "0d60b7c5cd71190a9ef4b873b0f516be15447c32d83914db2794b14592b0b460",
-  "blk.13.attn_q.weight": "8296069e65bef794cefc61257fc65789b3cb22955e30f3df129205e5041b2222",
-  "blk.13.attn_v.weight": "ca0f4ab9d16a748fc643a5c0c7a19826a811bf2a4e7316a8c935d4bf0ce8abc6",
-  "blk.13.ffn_down.weight": "d5514e0c8e7b3ed1cbcc1605eb5be1733b6ab3514cf8a0508fc72f7d05ed8bcb",
-  "blk.13.ffn_gate.weight": "8108e517a82e08a3aefbbd267bfa50a1668f92a76273280ce8a6bc1f6dd61521",
-  "blk.13.ffn_norm.weight": "5fcb6132d2134bf1f835b904a99820fa501dbc57d2224129f7098bf3cabc1d36",
-  "blk.13.ffn_up.weight": "6d744b7cd390a3cae3aa350dd379b81246acd056a2259996b6aaadece8465ccc",
-  "blk.13.post_attention_norm.weight": "e08b14698912509790e9575b8676971fbb0a4d82d719367e3756c0d0c4ab8cc0",
-  "blk.13.post_ffw_norm.weight": "2b196e4450fc5f1e7367b2cf7fe33a15fe919fbcdd861d11002346f16e980535",
-  "blk.14.attn_k.weight": "120e5f48d7268dfd9ab5f4bc9cc57a7cec63ea9635f56b80d435eb22936e9483",
-  "blk.14.attn_norm.weight": "146367bcce4db72cc894419a2e0145a6f533507dd68e4739c10ee480308c401f",
-  "blk.14.attn_output.weight": "720fa0165e756876c5cb6ad9e2780dd910390933f3f8849e5add5da04266650b",
-  "blk.14.attn_q.weight": "f5183466f56219ca1aca52d8b82c2d966a4198fea40fdd6b39f4d8b06ca2a6dd",
-  "blk.14.attn_v.weight": "24f8ea3d5512cd37c43c8329cb0da0c90d1895aef763ac2dcee3fe5157ec50a2",
-  "blk.14.ffn_down.weight": "e29960965b384ae5ab3d898a4dbaa8fddd28fa0e477ac28bcac49dec12a5ac67",
-  "blk.14.ffn_gate.weight": "6d0d6a74bfe9692e8f8eedff0fc34fc4fa1c8687794f35f2e2b033ab2d7510b8",
-  "blk.14.ffn_norm.weight": "f7036c1a9a71e046c9d2af16e9218fda5dbb0f7241ab44747abed1f0f9d602ca",
-  "blk.14.ffn_up.weight": "7d69ea1424007ffc9c12247dd0308c616e93ac02a59ec341cfa48f92d6ce3b10",
-  "blk.14.post_attention_norm.weight": "65b9712834d9445d4236bec362f3fb795c20d60c541b3dc6dbb7914d9b493e41",
-  "blk.14.post_ffw_norm.weight": "9c6a8da2e4e437d5cfdf3b9097e9f8b64bf07946a048badec20f4d374613f38f",
-  "blk.15.attn_k.weight": "864bc618303a0e4ee67fb1d5e751de61e936cd51e96669dd86f8cd08f2305045",
-  "blk.15.attn_norm.weight": "f9f4187da6eeadc2fc5921d8fe669741697d16c13d71e4aaeb73b82f50dc577e",
-  "blk.15.attn_output.weight": "ce2419a0b097036b2a31f2f4ad731d5814bcc2ef4c511786e24471e5eefd273b",
-  "blk.15.attn_q.weight": "9539db5a970d11ebe99722d1e13fcd635e250033630811efe583d2f97778e4a9",
-  "blk.15.attn_v.weight": "1c834b48ccd88adaeabb7d8bcb6be0bcd6d5ac1354ce88fc28f19a1a96b81ab3",
-  "blk.15.ffn_down.weight": "bc1f97a65dde6fa2c1e5397afb612266944b343f2eaa868b635ddd25829f8a42",
-  "blk.15.ffn_gate.weight": "1b14529d57056b79037f6cb5008132e62cc35992353b38dda59572274623103b",
-  "blk.15.ffn_norm.weight": "9af77458de9ee55c66f93865759f9c2c398557f94f3fa8fa6af30543d7339cde",
-  "blk.15.ffn_up.weight": "41d524a26b61a9595816b4fd53cf57ef50a702e4ef32933ff6136dca9136a267",
-  "blk.15.post_attention_norm.weight": "c60a03cd0e63a7db5c80015e58e9b97ba2208caa19f66a6fef5c4447eca900ce",
-  "blk.15.post_ffw_norm.weight": "34f7f9f96769215bbc3d17084df091864aef96a6645b7d0b3b7d9bd92f1a4b0b",
-  "blk.16.attn_k.weight": "7e27240d9f3a8c6cf0f4a980113d43234f514eadc3e3e1792b86efb29ffb1a6d",
-  "blk.16.attn_norm.weight": "af798acc0899282a30448edec48223b3e8efda177090273e612d8eca5e377301",
-  "blk.16.attn_output.weight": "79df39a3709d3d53e84146291e0944a7a653d06705293d9ccb5648dceadb432c",
-  "blk.16.attn_q.weight": "db58a1c3b83ad294804e5fd7321005719e200659173466df5a52a182b80b7165",
-  "blk.16.attn_v.weight": "2af6d48cbaeb225b5c1a704f76abd89c8ab1521417695b112b4dcc2cbd39b74d",
-  "blk.16.ffn_down.weight": "fc1c813eb5e7da3d6194569d6cb21602fc6eff2dc8e1b0eb753f2d5df148189c",
-  "blk.16.ffn_gate.weight": "7a80bcbc42464bd55df4814a6edbd7b5c153e0428323bbe49de55e2d2add33e7",
-  "blk.16.ffn_norm.weight": "2041685ee926d30f3f2ae4ec35b5688f1cd834167a6359a7d4057eac804c58b2",
-  "blk.16.ffn_up.weight": "8da4b718973ac1d43b928829bc45e062fd101984d6c98dd825bd7c5d08ebfbe3",
-  "blk.16.post_attention_norm.weight": "975c48fe680a6167438a106140a8872eee7765191f152d80e3b8ddf47693e095",
-  "blk.16.post_ffw_norm.weight": "4de2d4d483acfe4fc77860ea929025df2f4e15c10729413f36a18c94eaa6d689",
-  "blk.17.attn_k.weight": "f937e61f0af8c4cd98ee742648eb60e02e579683e21d421071295a3b70aebaad",
-  "blk.17.attn_norm.weight": "c3270583ed28b7e423f5b170c59113234f258169b93a867d9274f4c10b7cb115",
-  "blk.17.attn_output.weight": "b8c1150e81e685e539a5dcf2c19047a24eba2b281fabe166674b1d71ef4612ea",
-  "blk.17.attn_q.weight": "c255100ae2011e7dc7e3bf3bc3ccd96d859fbb98581cae993d7b82c1ba8e8b39",
-  "blk.17.attn_v.weight": "5830bb0a555984c6485348067f70b5d22ae337c011aa9248dac2ff4c95944551",
-  "blk.17.ffn_down.weight": "8ff9a7cccaa3776434a9d895aae4fb5c36c736bf2ec98784226b4c234940fbb0",
-  "blk.17.ffn_gate.weight": "1b52876739712831c272911533da206f407b46034a1a4ae8a88c1f96b6bd5747",
-  "blk.17.ffn_norm.weight": "d0e16ba5e87c91b545334e022058c7d03849665c3b1a6298771b656531366b66",
-  "blk.17.ffn_up.weight": "4dd6211d01dbebbe21052708eddc242b082a58b5f18ed16479e17987c1d3432e",
-  "blk.17.post_attention_norm.weight": "6f49c775c7417dade77ba8268a0f8441c1e5ec28b5d7e4dc5ed07a04d04600c8",
-  "blk.17.post_ffw_norm.weight": "b91a0bb2e6679e9c9be06ad323adae441d00a3d673efb19d7c4954be2aa84b27",
-  "blk.18.attn_k.weight": "22b565ace1b4da8b33865a58625be1d90beea9891f29686a69fa9cf7c93217db",
-  "blk.18.attn_norm.weight": "3e0160d7063c8753de65d2356a66648e47d921efdc5c917efb8209892120f8db",
-  "blk.18.attn_output.weight": "e3180f0bb4ca90b31e9b08158db38e332de62dfbaefe34aa94cc316409331e09",
-  "blk.18.attn_q.weight": "f3a5a83614c3ba7ea41cdd5b1b0819a241ee2a951a381ce4a9e001d3f700ed8f",
-  "blk.18.attn_v.weight": "f3350a5984fb951fc738adcf78147e6d812ff1c576670c460cafc99c253c1654",
-  "blk.18.ffn_down.weight": "9e9d09b13a33525e14bdaee6efc65c551ac7cf7680e534b940ab122a3a7c1ac9",
-  "blk.18.ffn_gate.weight": "ebaec8b4b578a2e8d815baac12f1675c208f80c68074d5a18288a2e1a60680ee",
-  "blk.18.ffn_norm.weight": "33e7687c53a242f2f8dc7093a491c97b18d4a5a8c14d183f02bd586a770f05aa",
-  "blk.18.ffn_up.weight": "78a1816662378ce56cc870e705174492781897b3afd2d4d97a51f10f2f2987c1",
-  "blk.18.post_attention_norm.weight": "a58dde3f12df3e94cbc27d87c8ea86f89af8a388a506446ff6758f05399b05fc",
-  "blk.18.post_ffw_norm.weight": "cebf90cc143577d483cca27b032dfd82031ee59bdf17c0e2cf60a0a3ad5bf996",
-  "blk.19.attn_k.weight": "4683375d0599ac9e2232196aae1e90af13a14cae26e865465de5c8e257bb2055",
-  "blk.19.attn_norm.weight": "f3eba936bfb1814bbcb0a1d62739eb66daac839df8c9c836fe0e94860df88525",
-  "blk.19.attn_output.weight": "51c0f01d38a9dcfe9bdbc4643576fab164c1d9e4b7168b7695c0ee55e6965667",
-  "blk.19.attn_q.weight": "28d15b69b8416f2e7ddc88fe381cb1e2ef2ad705fb1c268139ba96498cc74848",
-  "blk.19.attn_v.weight": "6860f1cd720638e63a981fa2c0b4db900129826bcb9823c9ddf9fb8b1b9f3383",
-  "blk.19.ffn_down.weight": "bc7f2d7827ee01c2dd41401c7b3b1700ad3a4ff620e8bb734f92630d342dcc7f",
-  "blk.19.ffn_gate.weight": "54d03ef69ba373fc410fbca8f1e34a565d58e4296d9a035ff7e48340b9c848e7",
-  "blk.19.ffn_norm.weight": "9178fc796a340ee6e8128ca74c0cb6203d1adbed6927af4e5ac7863da57affc7",
-  "blk.19.ffn_up.weight": "a77bd708026c6e83ad5c79c223278e74621bcf74a9641c7818d96b595daaad20",
-  "blk.19.post_attention_norm.weight": "ae94aa26f4c411bf9496a6fd4a6df64ee589ee1ae9a04b531d45acc95721e582",
-  "blk.19.post_ffw_norm.weight": "9ad210700edeef12133bdcff04bf1c7f62b49f6f4a9ba483c7cdc59857c24a5c",
-  "blk.20.attn_k.weight": "e35bce1e9f4a7a09ef34721f57ea38cfca68c272f52d923fe50af8308f66cfaa",
-  "blk.20.attn_norm.weight": "644800f6926fd34f233795c4dec1151a295d2138ca8cac33e3e48167d26f8b41",
-  "blk.20.attn_output.weight": "8d3758cd236471741e1ad66c0710cb79077dc8c7a3a292d35bc551c0c5abe627",
-  "blk.20.attn_q.weight": "c333b1f0f6f956b5d73891df10b1a0321e55fc31c40d623a24e1f52caa6a998b",
-  "blk.20.attn_v.weight": "8562b418d0c4868a050fb19fa3fcaf50a8cf1c669f537d666c80c7b3a04714e1",
-  "blk.20.ffn_down.weight": "97efb608ac44cc804198faec3ee66eafe56ced6b7ca5359700c6f1df75b7205e",
-  "blk.20.ffn_gate.weight": "5c61151d86f28415c73c73d90ec088c646cbe5c1640197caf58eb501ba7db293",
-  "blk.20.ffn_norm.weight": "24bbe0a701afd4bbeea65b3edde712b3cbb2281043bbc43dbf250582453116ed",
-  "blk.20.ffn_up.weight": "e170cf68e249566aa99eb6f6b265679bf9a5a6b76830ba24e7e130c2515910c4",
-  "blk.20.post_attention_norm.weight": "e092d751cfe20dbf2d348358f3b38397bd83e4ed94d6bbaa6bbaddcd902b2ac4",
-  "blk.20.post_ffw_norm.weight": "219a18a47dcba76e669e4322223a5a9227bd3db1de3fbd3d3cfb22e54a783c5a",
-  "blk.21.attn_k.weight": "c3a095ebddb42c63824f1c98da65263dc88e4d790a26aa1632840b44f5cc7cb1",
-  "blk.21.attn_norm.weight": "ef8bbaded5fbc45ad9cf3985ae02174524e7090fe6362811124f942ef643bec7",
-  "blk.21.attn_output.weight": "668f018aba72baac6252aa3ad58569ddd55ab751a0dd8d7bcc9fb9b6efb4bf53",
-  "blk.21.attn_q.weight": "e759c65663089f3bbbd51847934c185e680c82f1249065d5d487da638e519e6d",
-  "blk.21.attn_v.weight": "2ff57762686cf9ba1f5a6be76503454b97556ce67f4ac98254bd0562231197ba",
-  "blk.21.ffn_down.weight": "3fd106556fb721b1c28ae3f4026bc83eb1b08ed910f2ba5f466c6b5f327d91cb",
-  "blk.21.ffn_gate.weight": "338022d882f4b6619e8054a6fb909696fa3eef3013cf69b65c3cacdfc5b9e42c",
-  "blk.21.ffn_norm.weight": "1e77660c23a3f9653ee721a863d1960f773d87437cabc4dc0a6e17ee3d4e5e44",
-  "blk.21.ffn_up.weight": "7d31b20fbc2e6eba8f350f170069dc36f0cb12f68fbc4206ec5022a74085ebcb",
-  "blk.21.post_attention_norm.weight": "9638bae8d8bdcd7ed68da282979cd84a07c41ff9cabcaea94ebc846a1803db23",
-  "blk.21.post_ffw_norm.weight": "d622ef11115fe0cbe04b727d5a3b6371e7f39bf08c8d5eb9bc6da52e3f3cfb9d",
-  "blk.22.attn_k.weight": "5c321cb29deffbe57de200dd206a62005f1e80acb86c4fd2349dd44c8d3594fd",
-  "blk.22.attn_norm.weight": "198d949705d7170a331d75889d8c7500c3635254dac2cc6aa4dc35d556584536",
-  "blk.22.attn_output.weight": "19805cd5d7025b457e5d41d70db8b3fd63c2dd0e4a94d3ef1704d50ef4e749e8",
-  "blk.22.attn_q.weight": "177836cd583fc87405975ddc21ebfebdaa090a0363799664c72caa3da851ae2c",
-  "blk.22.attn_v.weight": "fea255692483e30d0108f9e4e250eb3ed7dbda8d83f499b06519b8c223ae6096",
-  "blk.22.ffn_down.weight": "00cb8939f03e5817d6d412de8cf2c923c9568d5493e382cec7faf5718fb034eb",
-  "blk.22.ffn_gate.weight": "b0591065b91281b2fbd8a9567f3568d40479f680e1f0a29e27ae213f37642489",
-  "blk.22.ffn_norm.weight": "96b5c5d0737c2ceb8fc869f54adb9e5f46e28cb7b177c40f49fa926b923c00f8",
-  "blk.22.ffn_up.weight": "81f472185b24344ab0594ea8246cc6e200e0dc1cab4943e74fbe4ca19d5a9701",
-  "blk.22.post_attention_norm.weight": "27fa9aa6260aa3071e0391e1a1d49322dcb6e8072315b8a9b7064087108dbd06",
-  "blk.22.post_ffw_norm.weight": "f37e1dcd7f643d9545675ffe9dc527a11eba86eb204989c2f44f636b266d896a",
-  "blk.23.attn_k.weight": "5d82f36658a56c3f94d0bb2d61f65509c966fa6568f81812e0d3e338b380ef8c",
-  "blk.23.attn_norm.weight": "b7983f88d9cad88bc88a528923e6da592ad20e699965b223ebc10840fe1f4fec",
-  "blk.23.attn_output.weight": "59f97f80f430d71606aab0158a195aed29ccd3405e6c0a5c41c809be8eb01898",
-  "blk.23.attn_q.weight": "53ac4789fe958919cc02ea4222bcd64c0ea1b4baa54304bff46635bdf42f7490",
-  "blk.23.attn_v.weight": "ec8abe09b9e84dbb52c7a068094657c6d3c62fe551ba8d7c3a3f23da622e9756",
-  "blk.23.ffn_down.weight": "3cf547eccb1b82aa64f208cee9682d7f558ca84e0aead7d9d3d1420d90f3d992",
-  "blk.23.ffn_gate.weight": "366aa2486d911ba81eb519119e13807deacf7e9908bc1975a2a63e00d6b10124",
-  "blk.23.ffn_norm.weight": "6d1d4a4af34bb7dc090ac87d6457d398c3e0fb68bd2e2b60b099dc318b6cfac3",
-  "blk.23.ffn_up.weight": "53f76692e253f5d2420b3f200c731b9f3b7a83e379920b4a067c729b4674aa4d",
-  "blk.23.post_attention_norm.weight": "7c952fa0efa76b3f048c8c4c9e8dcb5e3724d231327eda6423a34d3f3d3367de",
-  "blk.23.post_ffw_norm.weight": "7ab188cfe61f0a91b40309a0ab6bfa99f19d0ff2a37b6ac10e5f0c7f44eb5270",
-  "blk.24.attn_k.weight": "225798792f9bfdd10eff0505ebe61e0aad0209c17b431f6044ee7968ffe8c198",
-  "blk.24.attn_norm.weight": "635e3c1ebf5219bbebfc40ef164bc32d2b726ef595a94da64ac524ae878e2915",
-  "blk.24.attn_output.weight": "482f5bb2db8d9ed22b253d9a3296333b239efe698e5992e5d77e7e12dc2a5cf5",
-  "blk.24.attn_q.weight": "43805bbccddb65d58fffc4be9b5c374d4e1df1395ec1e1ffb4bcff03e98d5adb",
-  "blk.24.attn_v.weight": "fa741af54b4a3b1775d32f59134756090c5df2e7345a12a2d8db94fe289667a7",
-  "blk.24.ffn_down.weight": "83c6351e3162626b276f524a57836144625c2556dbe321b57cbd8fd486a68fab",
-  "blk.24.ffn_gate.weight": "fbe66be0d84d12cea5176cc7eaef64382ffc7324cd9d6266a3342dc43442f2ac",
-  "blk.24.ffn_norm.weight": "77c1445a8639ad24938bdf0280233eea2362d47391421833dfa72ec756dfc1e8",
-  "blk.24.ffn_up.weight": "78235ac729ee23c1cf1ae543751e3af32776d8808cee6e529c2a625a1f027654",
-  "blk.24.post_attention_norm.weight": "161f71b6d07628d43e4ae51a4c9088ec6ca2db123a17986a14505d83fdd04dad",
-  "blk.24.post_ffw_norm.weight": "cf1ba692aa683368b02ac413e69b2521b98c69a5274eacbb54165b53bf38a8b2",
-  "blk.25.attn_k.weight": "057a56bd8c8d2b41608d1f71faa3052902152ddf85e47669ad950c1c3e77c33f",
-  "blk.25.attn_norm.weight": "b7179fe02c334da556ddcf6c1b502245639a728c4cbba8b552d8e1df4565ee9d",
-  "blk.25.attn_output.weight": "4fed8b05b08a0ff75ffd022701bbeb52f17b23d09332a1ddcba737244bd0d3b0",
-  "blk.25.attn_q.weight": "c52e99f5d38bf7538d6106a0bbf38ac6dc6296bca9a3f849afa384ea67b4af01",
-  "blk.25.attn_v.weight": "c49c23d8e1cfa6a8eb971eb69942204890c6d7d830dc8774c84b108a80598912",
-  "blk.25.ffn_down.weight": "c08d4dc8412b19fdc870c164b83c341b236ec6fe7bb4a9bcfe0dc100faa20286",
-  "blk.25.ffn_gate.weight": "1a4cb3f36735d59181721471452807903006539e5e1b5ceb4f72d1d7ae134127",
-  "blk.25.ffn_norm.weight": "8fd6bd0dcec5198761525a36992a57c9ec5e9da60a22092839a84ae8c4e87f26",
-  "blk.25.ffn_up.weight": "3a00f39bdd5f31dc5e3b281d2002e1ac4f2475d49a0ac1d7720a25b377dcd04a",
-  "blk.25.post_attention_norm.weight": "e5f31a648612c859b6d21c9ee426e87a86cb1973dfdd86276c767371d9cef5ad",
-  "blk.25.post_ffw_norm.weight": "553c3bd774922c99c2384380a142d019881d30dbf0fe3bf9430dabfb3f6cbd33",
-  "output_norm.weight": "49445c4585ab0a8135717a0bdb1cda4a062a030177d0119561d91542aec5744b"
-}
--- a/docs/api.md
+++ b/docs/api.md
@@ -69,7 +69,7 @@ Enable JSON mode by setting the `format` parameter to `json`. This will structur

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "prompt": "Why is the sky blue?"
 }'
 ```
@@ -80,7 +80,7 @@ A stream of JSON objects is returned:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T08:52:19.385406455-07:00",
  "response": "The",
  "done": false
@@ -102,7 +102,7 @@ To calculate how fast the response is generated in tokens per second (token/s),

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T19:22:45.499127Z",
  "response": "",
  "done": true,
@@ -124,7 +124,7 @@ A response can be received in one reply when streaming is off.

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "prompt": "Why is the sky blue?",
  "stream": false
 }'
@@ -136,7 +136,7 @@ If `stream` is set to `false`, the response will be a single JSON object:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T19:22:45.499127Z",
  "response": "The sky is blue because it is the color of the sky.",
  "done": true,
@@ -194,7 +194,7 @@ curl http://localhost:11434/api/generate -d '{

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "prompt": "What color is the sky at different times of the day? Respond using JSON",
  "format": "json",
  "stream": false
@@ -205,7 +205,7 @@ curl http://localhost:11434/api/generate -d '{

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-11-09T21:07:55.186497Z",
  "response": "{\n\"morning\": {\n\"color\": \"blue\"\n},\n\"noon\": {\n\"color\": \"blue-gray\"\n},\n\"afternoon\": {\n\"color\": \"warm gray\"\n},\n\"evening\": {\n\"color\": \"orange\"\n}\n}\n",
  "done": true,
@@ -327,7 +327,7 @@ If you want to set custom options for the model at runtime rather than in the Mo

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "prompt": "Why is the sky blue?",
  "stream": false,
  "options": {
@@ -368,7 +368,7 @@ curl http://localhost:11434/api/generate -d '{

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T19:22:45.499127Z",
  "response": "The sky is blue because it is the color of the sky.",
  "done": true,
@@ -390,7 +390,7 @@ If an empty prompt is provided, the model will be loaded into memory.

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1"
+  "model": "llama3"
 }'
 ```

@@ -400,40 +400,13 @@ A single JSON object is returned:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-12-18T19:52:07.071755Z",
  "response": "",
  "done": true
 }
 ```

-#### Unload a model
-
-If an empty prompt is provided and the `keep_alive` parameter is set to `0`, a model will be unloaded from memory.
-
-##### Request
-
-```shell
-curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1",
-  "keep_alive": 0
-}'
-```
-
-##### Response
-
-A single JSON object is returned:
-
-```json
-{
-  "model": "llama3.1",
-  "created_at": "2024-09-12T03:54:03.516566Z",
-  "response": "",
-  "done": true,
-  "done_reason": "unload"
-}
-```
-
 ## Generate a chat completion

 ```shell
@@ -472,7 +445,7 @@ Send a chat message with a streaming response.

 ```shell
 curl http://localhost:11434/api/chat -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "messages": [
    {
      "role": "user",
@@ -488,7 +461,7 @@ A stream of JSON objects is returned:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T08:52:19.385406455-07:00",
  "message": {
    "role": "assistant",
@@ -503,7 +476,7 @@ Final response:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T19:22:45.499127Z",
  "done": true,
  "total_duration": 4883583458,
@@ -521,7 +494,7 @@ Final response:

 ```shell
 curl http://localhost:11434/api/chat -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "messages": [
    {
      "role": "user",
@@ -536,7 +509,7 @@ curl http://localhost:11434/api/chat -d '{

 ```json
 {
-  "model": "llama3.1",
+  "model": "registry.ollama.ai/library/llama3:latest",
  "created_at": "2023-12-12T14:13:43.416799Z",
  "message": {
    "role": "assistant",
@@ -560,7 +533,7 @@ Send a chat message with a conversation history. You can use this same approach

 ```shell
 curl http://localhost:11434/api/chat -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "messages": [
    {
      "role": "user",
@@ -584,7 +557,7 @@ A stream of JSON objects is returned:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T08:52:19.385406455-07:00",
  "message": {
    "role": "assistant",
@@ -598,7 +571,7 @@ Final response:

 ```json
 {
-  "model": "llama3.1",
+  "model": "llama3",
  "created_at": "2023-08-04T19:22:45.499127Z",
  "done": true,
  "total_duration": 8113331500,
@@ -656,7 +629,7 @@ curl http://localhost:11434/api/chat -d '{

 ```shell
 curl http://localhost:11434/api/chat -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "messages": [
    {
      "role": "user",
@@ -674,7 +647,7 @@ curl http://localhost:11434/api/chat -d '{

 ```json
 {
-  "model": "llama3.1",
+  "model": "registry.ollama.ai/library/llama3:latest",
  "created_at": "2023-12-12T14:13:43.416799Z",
  "message": {
    "role": "assistant",
@@ -763,64 +736,6 @@ curl http://localhost:11434/api/chat -d '{
 }
 ```

-#### Load a model
-
-If the messages array is empty, the model will be loaded into memory.
-
-##### Request
-
-```
-curl http://localhost:11434/api/chat -d '{
-  "model": "llama3.1",
-  "messages": []
-}'
-```
-
-##### Response
-```json
-{
-  "model": "llama3.1",
-  "created_at":"2024-09-12T21:17:29.110811Z",
-  "message": {
-    "role": "assistant",
-    "content": ""
-  },
-  "done_reason": "load",
-  "done": true
-}
-```
-
-#### Unload a model
-
-If the messages array is empty and the `keep_alive` parameter is set to `0`, a model will be unloaded from memory.
-
-##### Request
-
-```
-curl http://localhost:11434/api/chat -d '{
-  "model": "llama3.1",
-  "messages": [],
-  "keep_alive": 0
-}'
-```
-
-##### Response
-
-A single JSON object is returned:
-
-```json
-{
-  "model": "llama3.1",
-  "created_at":"2024-09-12T21:33:17.547535Z",
-  "message": {
-    "role": "assistant",
-    "content": ""
-  },
-  "done_reason": "unload",
-  "done": true
-}
-```
-
 ## Create a Model

 ```shell
@@ -989,7 +904,7 @@ Show information about a model including details, modelfile, template, parameter

 ```shell
 curl http://localhost:11434/api/show -d '{
-  "name": "llama3.1"
+  "name": "llama3"
 }'
 ```

@@ -1050,7 +965,7 @@ Copy a model. Creates a model with another name from an existing model.

 ```shell
 curl http://localhost:11434/api/copy -d '{
-  "source": "llama3.1",
+  "source": "llama3",
  "destination": "llama3-backup"
 }'
 ```
@@ -1105,7 +1020,7 @@ Download a model from the ollama library. Cancelled pulls are resumed from where

 ```shell
 curl http://localhost:11434/api/pull -d '{
-  "name": "llama3.1"
+  "name": "llama3"
 }'
 ```

--- a/docs/development.md
+++ b/docs/development.md
@@ -148,22 +148,3 @@ In addition to the common Windows development tools described above, install AMD
 - [Strawberry Perl](https://strawberryperl.com/)

 Lastly, add `ninja.exe` included with MSVC to the system path (e.g. `C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\Common7\IDE\CommonExtensions\Microsoft\CMake\Ninja`).
-
-#### Windows arm64
-
-The default `Developer PowerShell for VS 2022` may default to x86 which is not what you want.  To ensure you get an arm64 development environment, start a plain PowerShell terminal and run:
-
-```powershell
-import-module 'C:\\Program Files\\Microsoft Visual Studio\\2022\\Community\\Common7\\Tools\\Microsoft.VisualStudio.DevShell.dll'
-Enter-VsDevShell -Arch arm64 -vsinstallpath 'C:\\Program Files\\Microsoft Visual Studio\\2022\\Community' -skipautomaticlocation
-```
-
-You can confirm with `write-host $env:VSCMD_ARG_TGT_ARCH`
-
-Follow the instructions at https://www.msys2.org/wiki/arm64/ to set up an arm64 msys2 environment.  Ollama requires gcc and mingw32-make to compile, which is not currently available on Windows arm64, but a gcc compatibility adapter is available via `mingw-w64-clang-aarch64-gcc-compat`. At a minimum you will need to install the following:
-
-```
-pacman -S mingw-w64-clang-aarch64-clang mingw-w64-clang-aarch64-gcc-compat mingw-w64-clang-aarch64-make make
-```
-
-You will need to ensure your PATH includes go, cmake, gcc and clang mingw32-make to build ollama from source. (typically `C:\msys64\clangarm64\bin\`)
--- a/docs/faq.md
+++ b/docs/faq.md
@@ -32,7 +32,7 @@ When using the API, specify the `num_ctx` parameter:

 ```shell
 curl http://localhost:11434/api/generate -d '{
-  "model": "llama3.1",
+  "model": "llama3",
  "prompt": "Why is the sky blue?",
  "options": {
    "num_ctx": 4096
@@ -237,13 +237,9 @@ ollama run llama3.1 ""

 ## How do I keep a model loaded in memory or make it unload immediately?

-By default models are kept in memory for 5 minutes before being unloaded. This allows for quicker response times if you're making numerous requests to the LLM. If you want to immediately unload a model from memory, use the `ollama stop` command:
+By default models are kept in memory for 5 minutes before being unloaded. This allows for quicker response times if you are making numerous requests to the LLM. You may, however, want to free up the memory before the 5 minutes have elapsed or keep the model loaded indefinitely. Use the `keep_alive` parameter with either the `/api/generate` and `/api/chat` API endpoints to control how long the model is left in memory.

-```shell
-ollama stop llama3.1
-```
-
-If you're using the API, use the `keep_alive` parameter with the `/api/generate` and `/api/chat` endpoints to set the amount of time that a model stays in memory. The `keep_alive` parameter can be set to:
+The `keep_alive` parameter can be set to:
 * a duration string (such as "10m" or "24h")
 * a number in seconds (such as 3600)
 * any negative number which will keep the model loaded in memory (e.g. -1 or "-1m")
@@ -251,17 +247,17 @@ If you're using the API, use the `keep_alive` parameter with the `/api/generate`

 For example, to preload a model and leave it in memory use:
 ```shell
-curl http://localhost:11434/api/generate -d '{"model": "llama3.1", "keep_alive": -1}'
+curl http://localhost:11434/api/generate -d '{"model": "llama3", "keep_alive": -1}'
 ```

 To unload the model and free up memory use:
 ```shell
-curl http://localhost:11434/api/generate -d '{"model": "llama3.1", "keep_alive": 0}'
+curl http://localhost:11434/api/generate -d '{"model": "llama3", "keep_alive": 0}'
 ```

-Alternatively, you can change the amount of time all models are loaded into memory by setting the `OLLAMA_KEEP_ALIVE` environment variable when starting the Ollama server. The `OLLAMA_KEEP_ALIVE` variable uses the same parameter types as the `keep_alive` parameter types mentioned above. Refer to the section explaining [how to configure the Ollama server](#how-do-i-configure-ollama-server) to correctly set the environment variable.
+Alternatively, you can change the amount of time all models are loaded into memory by setting the `OLLAMA_KEEP_ALIVE` environment variable when starting the Ollama server. The `OLLAMA_KEEP_ALIVE` variable uses the same parameter types as the `keep_alive` parameter types mentioned above. Refer to section explaining [how to configure the Ollama server](#how-do-i-configure-ollama-server) to correctly set the environment variable.

-The `keep_alive` API parameter with the `/api/generate` and `/api/chat` API endpoints will override the `OLLAMA_KEEP_ALIVE` setting.
+If you wish to override the `OLLAMA_KEEP_ALIVE` setting, use the `keep_alive` API parameter with the `/api/generate` or `/api/chat` API endpoints.

 ## How do I manage the maximum number of requests the Ollama server can queue?

--- a/docs/gpu.md
+++ b/docs/gpu.md
@@ -10,7 +10,7 @@ Check your compute compatibility to see if your card is supported:
 | 9.0                | NVIDIA              | `H100`                                                                                                      |
 | 8.9                | GeForce RTX 40xx    | `RTX 4090` `RTX 4080 SUPER` `RTX 4080` `RTX 4070 Ti SUPER` `RTX 4070 Ti` `RTX 4070 SUPER` `RTX 4070` `RTX 4060 Ti` `RTX 4060`  |
 |                    | NVIDIA Professional | `L4` `L40` `RTX 6000`                                                                                       |
-| 8.6                | GeForce RTX 30xx    | `RTX 3090 Ti` `RTX 3090` `RTX 3080 Ti` `RTX 3080` `RTX 3070 Ti` `RTX 3070` `RTX 3060 Ti` `RTX 3060` `RTX 3050 Ti` `RTX 3050`   |
+| 8.6                | GeForce RTX 30xx    | `RTX 3090 Ti` `RTX 3090` `RTX 3080 Ti` `RTX 3080` `RTX 3070 Ti` `RTX 3070` `RTX 3060 Ti` `RTX 3060`         |
 |                    | NVIDIA Professional | `A40` `RTX A6000` `RTX A5000` `RTX A4000` `RTX A3000` `RTX A2000` `A10` `A16` `A2`                          |
 | 8.0                | NVIDIA              | `A100` `A30`                                                                                                |
 | 7.5                | GeForce GTX/RTX     | `GTX 1650 Ti` `TITAN RTX` `RTX 2080 Ti` `RTX 2080` `RTX 2070` `RTX 2060`                                    |
--- a/docs/import.md
+++ b/docs/import.md
@@ -38,7 +38,7 @@ Ollama supports importing adapters based on several different model architecture

 You can create the adapter using a fine tuning framework or tool which can output adapters in the Safetensors format, such as:

-  * Hugging Face [fine tuning framework](https://huggingface.co/docs/transformers/en/training)
+  * Hugging Face [fine tuning framework] (https://huggingface.co/docs/transformers/en/training)
  * [Unsloth](https://github.com/unslothai/unsloth)
  * [MLX](https://github.com/ml-explore/mlx)

--- a/docs/linux.md
+++ b/docs/linux.md
@@ -1,59 +1,44 @@
-# Linux
+# Ollama on Linux

 ## Install

-To install Ollama, run the following command:
+Install Ollama running this one-liner:

-```shell
+>
+
+```bash
 curl -fsSL https://ollama.com/install.sh | sh
 ```

+## AMD Radeon GPU support
+
+While AMD has contributed the `amdgpu` driver upstream to the official linux
+kernel source, the version is older and may not support all ROCm features. We
+recommend you install the latest driver from
+https://www.amd.com/en/support/linux-drivers for best support of your Radeon
+GPU.
+
 ## Manual install

-Download and extract the package:
+### Download `ollama`

-```shell
-curl -L https://ollama.com/download/ollama-linux-amd64.tgz -o ollama-linux-amd64.tgz
-sudo tar -C /usr -xzf ollama-linux-amd64.tgz
+Download and extract the Linux package:
+
+```bash
+curl -fsSL https://ollama.com/download/ollama-linux-amd64.tgz | sudo tar zx -C /usr
 ```

-Start Ollama:
-
-```shell
-ollama serve
-```
-
-In another terminal, verify that Ollama is running:
-
-```shell
-ollama -v
-```
-
-### AMD GPU install
-
-If you have an AMD GPU, also download and extract the additional ROCm package:
-
-```shell
-curl -L https://ollama.com/download/ollama-linux-amd64-rocm.tgz -o ollama-linux-amd64-rocm.tgz
-sudo tar -C /usr -xzf ollama-linux-amd64-rocm.tgz
-```
-
-### ARM64 install
-
-Download and extract the ARM64-specific package:
-
-```shell
-curl -L https://ollama.com/download/ollama-linux-arm64.tgz -o ollama-linux-arm64.tgz
-sudo tar -C /usr -xzf ollama-linux-arm64.tgz
+If you have an AMD GPU, also download and extract the ROCm package into the same location
+```bash
+curl -fsSL https://ollama.com/download/ollama-linux-amd64-rocm.tgz | sudo tar zx -C /usr
 ```

 ### Adding Ollama as a startup service (recommended)

-Create a user and group for Ollama:
+Create a user for Ollama:

-```shell
-sudo useradd -r -s /bin/false -U -m -d /usr/share/ollama ollama
-sudo usermod -a -G ollama $(whoami)
+```bash
+sudo useradd -r -s /bin/false -m -d /usr/share/ollama ollama
 ```

 Create a service file in `/etc/systemd/system/ollama.service`:
@@ -69,7 +54,6 @@ User=ollama
 Group=ollama
 Restart=always
 RestartSec=3
-Environment="PATH=$PATH"

 [Install]
 WantedBy=default.target
@@ -77,54 +61,46 @@ WantedBy=default.target

 Then start the service:

-```shell
+```bash
 sudo systemctl daemon-reload
 sudo systemctl enable ollama
 ```

-### Install CUDA drivers (optional)
+### Install CUDA drivers (optional – for Nvidia GPUs)

 [Download and install](https://developer.nvidia.com/cuda-downloads) CUDA.

 Verify that the drivers are installed by running the following command, which should print details about your GPU:

-```shell
+```bash
 nvidia-smi
 ```

-### Install AMD ROCm drivers (optional)
+### Install ROCm (optional - for Radeon GPUs)
+[Download and Install](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/tutorial/quick-start.html)

-[Download and Install](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/tutorial/quick-start.html) ROCm v6.
+Make sure to install ROCm v6

 ### Start Ollama

-Start Ollama and verify it is running:
+Start Ollama using `systemd`:

-```shell
+```bash
 sudo systemctl start ollama
-sudo systemctl status ollama
 ```

-> [!NOTE]
-> While AMD has contributed the `amdgpu` driver upstream to the official linux
-> kernel source, the version is older and may not support all ROCm features. We
-> recommend you install the latest driver from
-> https://www.amd.com/en/support/linux-drivers for best support of your Radeon
-> GPU.
+## Update

-## Updating
+Update ollama by running the install script again:

-Update Ollama by running the install script again:
-
-```shell
+```bash
 curl -fsSL https://ollama.com/install.sh | sh
 ```

-Or by re-downloading Ollama:
+Or by downloading the ollama binary:

-```shell
-curl -L https://ollama.com/download/ollama-linux-amd64.tgz -o ollama-linux-amd64.tgz
-sudo tar -C /usr -xzf ollama-linux-amd64.tgz
+```bash
+curl -fsSL https://ollama.com/download/ollama-linux-amd64.tgz | sudo tar zx -C /usr
 ```

 ## Installing specific versions
@@ -133,15 +109,15 @@ Use `OLLAMA_VERSION` environment variable with the install script to install a s

 For example:

-```shell
-curl -fsSL https://ollama.com/install.sh | OLLAMA_VERSION=0.3.9 sh
+```
+curl -fsSL https://ollama.com/install.sh | OLLAMA_VERSION=0.1.32 sh
 ```

 ## Viewing logs

 To view logs of Ollama running as a startup service, run:

-```shell
+```bash
 journalctl -e -u ollama
 ```

@@ -149,7 +125,7 @@ journalctl -e -u ollama

 Remove the ollama service:

-```shell
+```bash
 sudo systemctl stop ollama
 sudo systemctl disable ollama
 sudo rm /etc/systemd/system/ollama.service
@@ -157,13 +133,13 @@ sudo rm /etc/systemd/system/ollama.service

 Remove the ollama binary from your bin directory (either `/usr/local/bin`, `/usr/bin`, or `/bin`):

-```shell
+```bash
 sudo rm $(which ollama)
 ```

 Remove the downloaded models and Ollama service user and group:

-```shell
+```bash
 sudo rm -r /usr/share/ollama
 sudo userdel ollama
 sudo groupdel ollama
--- a/docs/modelfile.md
+++ b/docs/modelfile.md
@@ -11,7 +11,7 @@ A model file is the blueprint to create and share models with Ollama.
 - [Examples](#examples)
 - [Instructions](#instructions)
  - [FROM (Required)](#from-required)
-    - [Build from existing model](#build-from-existing-model)
+    - [Build from llama3.1](#build-from-llama31)
    - [Build from a Safetensors model](#build-from-a-safetensors-model)
    - [Build from a GGUF file](#build-from-a-gguf-file)
  - [PARAMETER](#parameter)
@@ -50,7 +50,7 @@ INSTRUCTION arguments
 An example of a `Modelfile` creating a mario blueprint:

 ```modelfile
-FROM llama3.1
+FROM llama3
 # sets the temperature to 1 [higher is more creative, lower is more coherent]
 PARAMETER temperature 1
 # sets the context window size to 4096, this controls how many tokens the LLM can use as context to generate the next token
@@ -72,10 +72,10 @@ More examples are available in the [examples directory](../examples).
 To view the Modelfile of a given model, use the `ollama show --modelfile` command.

  ```bash
-  > ollama show --modelfile llama3.1
+  > ollama show --modelfile llama3
  # Modelfile generated by "ollama show"
  # To build a new Modelfile based on this one, replace the FROM line with:
-  # FROM llama3.1:latest
+  # FROM llama3:latest
  FROM /Users/pdevine/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29
  TEMPLATE """{{ if .System }}<|start_header_id|>system<|end_header_id|>

@@ -100,7 +100,7 @@ The `FROM` instruction defines the base model to use when creating a model.
 FROM <model name>:<tag>
 ```

-#### Build from existing model
+#### Build from llama3.1

 ```modelfile
 FROM llama3.1
--- a/docs/openai.md
+++ b/docs/openai.md
@@ -25,7 +25,7 @@ chat_completion = client.chat.completions.create(
            'content': 'Say this is a test',
        }
    ],
-    model='llama3.1',
+    model='llama3',
 )

 response = client.chat.completions.create(
@@ -46,13 +46,13 @@ response = client.chat.completions.create(
 )

 completion = client.completions.create(
-    model="llama3.1",
+    model="llama3",
    prompt="Say this is a test",
 )

 list_completion = client.models.list()

-model = client.models.retrieve("llama3.1")
+model = client.models.retrieve("llama3")

 embeddings = client.embeddings.create(
    model="all-minilm",
@@ -74,7 +74,7 @@ const openai = new OpenAI({

 const chatCompletion = await openai.chat.completions.create({
    messages: [{ role: 'user', content: 'Say this is a test' }],
-    model: 'llama3.1',
+    model: 'llama3',
 })

 const response = await openai.chat.completions.create({
@@ -94,13 +94,13 @@ const response = await openai.chat.completions.create({
 })

 const completion = await openai.completions.create({
-    model: "llama3.1",
+    model: "llama3",
    prompt: "Say this is a test.",
 })

 const listCompletion = await openai.models.list()

-const model = await openai.models.retrieve("llama3.1")
+const model = await openai.models.retrieve("llama3")

 const embedding = await openai.embeddings.create({
  model: "all-minilm",
@@ -114,7 +114,7 @@ const embedding = await openai.embeddings.create({
 curl http://localhost:11434/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
-        "model": "llama3.1",
+        "model": "llama3",
        "messages": [
            {
                "role": "system",
@@ -154,13 +154,13 @@ curl http://localhost:11434/v1/chat/completions \
 curl http://localhost:11434/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
-        "model": "llama3.1",
+        "model": "llama3",
        "prompt": "Say this is a test"
    }'

 curl http://localhost:11434/v1/models

-curl http://localhost:11434/v1/models/llama3.1
+curl http://localhost:11434/v1/models/llama3

 curl http://localhost:11434/v1/embeddings \
    -H "Content-Type: application/json" \
@@ -274,7 +274,7 @@ curl http://localhost:11434/v1/embeddings \
 Before using a model, pull it locally `ollama pull`:

 ```shell
-ollama pull llama3.1
+ollama pull llama3
 ```

 ### Default model names
@@ -282,7 +282,7 @@ ollama pull llama3.1
 For tooling that relies on default OpenAI model names such as `gpt-3.5-turbo`, use `ollama cp` to copy an existing model name to a temporary name:

 ```
-ollama cp llama3.1 gpt-3.5-turbo
+ollama cp llama3 gpt-3.5-turbo
 ```

 Afterwards, this new model name can be specified the `model` field:
--- a/docs/template.md
+++ b/docs/template.md
@@ -33,7 +33,7 @@ Omitting a template in these models puts the responsibility of correctly templat
 To add templates in your model, you'll need to add a `TEMPLATE` command to the Modelfile. Here's an example using Meta's Llama 3.

 ```dockerfile
-FROM llama3.1
+FROM llama3

 TEMPLATE """{{- if .System }}<|start_header_id|>system<|end_header_id|>

--- a/docs/troubleshooting.md
+++ b/docs/troubleshooting.md
@@ -91,17 +91,6 @@ If none of those resolve the problem, gather additional information and file an
 - Check dmesg for any errors `sudo dmesg | grep -i nvrm` and `sudo dmesg | grep -i nvidia`


-## AMD GPU Discovery
-
-On linux, AMD GPU access typically requires `video` and/or `render` group membership to access the `/dev/kfd` device.  If permissions are not set up correctly, Ollama will detect this and report an error in the server log.
-
-When running in a container, in some Linux distributions and container runtimes, the ollama process may be unable to access the GPU.  Use `ls -ld /dev/kfd /dev/dri /dev/dri/*` on the host system to determine the group assignments on your system, and pass additional `--group-add ...` arguments to the container so it can access the required devices.
-
-If you are experiencing problems getting Ollama to correctly discover or use your GPU for inference, the following may help isolate the failure.
- `AMD_LOG_LEVEL=3` Enable info log levels in the AMD HIP/ROCm libraries.  This can help show more detailed error codes that can help troubleshoot problems
- `OLLAMA_DEBUG=1` During GPU discovery additional information will be reported
- Check dmesg for any errors from amdgpu or kfd drivers `sudo dmesg | grep -i amdgpu` and `sudo dmesg | grep -i kfd`
-
 ## Windows Terminal Errors

 Older versions of Windows 10 (e.g., 21H1) are known to have a bug where the standard terminal program does not display control characters correctly.  This can result in a long string of strings like `←[?25h←[?25l` being displayed, sometimes erroring with `The parameter is incorrect`  To resolve this problem, please update to Win 10 22H1 or newer.
--- a/docs/windows.md
+++ b/docs/windows.md
@@ -29,7 +29,7 @@ Ollama uses unicode characters for progress indication, which may render as unkn

 Here's a quick example showing API access from `powershell`
 ```powershell
-(Invoke-WebRequest -method POST -Body '{"model":"llama3.1", "prompt":"Why is the sky blue?", "stream": false}' -uri http://localhost:11434/api/generate ).Content | ConvertFrom-json
+(Invoke-WebRequest -method POST -Body '{"model":"llama3", "prompt":"Why is the sky blue?", "stream": false}' -uri http://localhost:11434/api/generate ).Content | ConvertFrom-json
 ```

 ## Troubleshooting
@@ -48,9 +48,6 @@ the explorer window by hitting `<cmd>+R` and type in:
 - `explorer %HOMEPATH%\.ollama` contains models and configuration
 - `explorer %TEMP%` contains temporary executable files in one or more `ollama*` directories

-## Uninstall
-
-The Ollama Windows installer registers an Uninstaller application.  Under `Add or remove programs` in Windows Settings, you can uninstall Ollama.

 ## Standalone CLI

--- a/envconfig/config.go
+++ b/envconfig/config.go
@@ -112,26 +112,6 @@ func KeepAlive() (keepAlive time.Duration) {
 	return keepAlive
 }

-// LoadTimeout returns the duration for stall detection during model loads. LoadTimeout can be configured via the OLLAMA_LOAD_TIMEOUT environment variable.
-// Zero or Negative values are treated as infinite.
-// Default is 5 minutes.
-func LoadTimeout() (loadTimeout time.Duration) {
-	loadTimeout = 5 * time.Minute
-	if s := Var("OLLAMA_LOAD_TIMEOUT"); s != "" {
-		if d, err := time.ParseDuration(s); err == nil {
-			loadTimeout = d
-		} else if n, err := strconv.ParseInt(s, 10, 64); err == nil {
-			loadTimeout = time.Duration(n) * time.Second
-		}
-	}
-
-	if loadTimeout <= 0 {
-		return time.Duration(math.MaxInt64)
-	}
-
-	return loadTimeout
-}
-
 func Bool(k string) func() bool {
 	return func() bool {
 		if s := Var(k); s != "" {
@@ -160,6 +140,8 @@ var (
 	SchedSpread = Bool("OLLAMA_SCHED_SPREAD")
 	// IntelGPU enables experimental Intel GPU detection.
 	IntelGPU = Bool("OLLAMA_INTEL_GPU")
+	// Set via OLLAMA_NEW_RUNNERS in the environment
+	NewRunners = Bool("OLLAMA_NEW_RUNNERS")
 )

 func String(s string) func() string {
@@ -179,6 +161,53 @@ var (
 	HsaOverrideGfxVersion = String("HSA_OVERRIDE_GFX_VERSION")
 )

+func RunnersDir() (p string) {
+	if p := Var("OLLAMA_RUNNERS_DIR"); p != "" {
+		return p
+	}
+
+	if runtime.GOOS != "windows" {
+		return
+	}
+
+	defer func() {
+		if p == "" {
+			slog.Error("unable to locate llm runner directory. Set OLLAMA_RUNNERS_DIR to the location of 'ollama/runners'")
+		}
+	}()
+
+	// On Windows we do not carry the payloads inside the main executable
+	exe, err := os.Executable()
+	if err != nil {
+		return
+	}
+
+	cwd, err := os.Getwd()
+	if err != nil {
+		return
+	}
+
+	var paths []string
+	for _, root := range []string{filepath.Dir(exe), filepath.Join(filepath.Dir(exe), LibRelativeToExe()), cwd} {
+		paths = append(paths,
+			root,
+			filepath.Join(root, runtime.GOOS+"-"+runtime.GOARCH),
+			filepath.Join(root, "dist", runtime.GOOS+"-"+runtime.GOARCH),
+		)
+	}
+
+	// Try a few variations to improve developer experience when building from source in the local tree
+	for _, path := range paths {
+		candidate := filepath.Join(path, "lib", "ollama", "runners")
+		if _, err := os.Stat(candidate); err == nil {
+			p = candidate
+			break
+		}
+	}
+
+	return p
+}
+
 func Uint(key string, defaultValue uint) func() uint {
 	return func() uint {
 		if s := Var(key); s != "" {
@@ -204,23 +233,6 @@ var (
 	MaxVRAM = Uint("OLLAMA_MAX_VRAM", 0)
 )

-func Uint64(key string, defaultValue uint64) func() uint64 {
-	return func() uint64 {
-		if s := Var(key); s != "" {
-			if n, err := strconv.ParseUint(s, 10, 64); err != nil {
-				slog.Warn("invalid environment variable, using default", "key", key, "value", s, "default", defaultValue)
-			} else {
-				return n
-			}
-		}
-
-		return defaultValue
-	}
-}
-
-// Set aside VRAM per GPU
-var GpuOverhead = Uint64("OLLAMA_GPU_OVERHEAD", 0)
-
 type EnvVar struct {
 	Name        string
 	Value       any
@@ -231,34 +243,21 @@ func AsMap() map[string]EnvVar {
 	ret := map[string]EnvVar{
 		"OLLAMA_DEBUG":             {"OLLAMA_DEBUG", Debug(), "Show additional debug information (e.g. OLLAMA_DEBUG=1)"},
 		"OLLAMA_FLASH_ATTENTION":   {"OLLAMA_FLASH_ATTENTION", FlashAttention(), "Enabled flash attention"},
-		"OLLAMA_GPU_OVERHEAD":      {"OLLAMA_GPU_OVERHEAD", GpuOverhead(), "Reserve a portion of VRAM per GPU (bytes)"},
 		"OLLAMA_HOST":              {"OLLAMA_HOST", Host(), "IP Address for the ollama server (default 127.0.0.1:11434)"},
 		"OLLAMA_KEEP_ALIVE":        {"OLLAMA_KEEP_ALIVE", KeepAlive(), "The duration that models stay loaded in memory (default \"5m\")"},
 		"OLLAMA_LLM_LIBRARY":       {"OLLAMA_LLM_LIBRARY", LLMLibrary(), "Set LLM library to bypass autodetection"},
-		"OLLAMA_LOAD_TIMEOUT":      {"OLLAMA_LOAD_TIMEOUT", LoadTimeout(), "How long to allow model loads to stall before giving up (default \"5m\")"},
 		"OLLAMA_MAX_LOADED_MODELS": {"OLLAMA_MAX_LOADED_MODELS", MaxRunners(), "Maximum number of loaded models per GPU"},
 		"OLLAMA_MAX_QUEUE":         {"OLLAMA_MAX_QUEUE", MaxQueue(), "Maximum number of queued requests"},
 		"OLLAMA_MODELS":            {"OLLAMA_MODELS", Models(), "The path to the models directory"},
 		"OLLAMA_NOHISTORY":         {"OLLAMA_NOHISTORY", NoHistory(), "Do not preserve readline history"},
 		"OLLAMA_NOPRUNE":           {"OLLAMA_NOPRUNE", NoPrune(), "Do not prune model blobs on startup"},
 		"OLLAMA_NUM_PARALLEL":      {"OLLAMA_NUM_PARALLEL", NumParallel(), "Maximum number of parallel requests"},
+		"OLLAMA_NEW_RUNNERS":       {"OLLAMA_NEW_RUNNERS", NewRunners(), "Enable new experimental runners"},
 		"OLLAMA_ORIGINS":           {"OLLAMA_ORIGINS", Origins(), "A comma separated list of allowed origins"},
+		"OLLAMA_RUNNERS_DIR":       {"OLLAMA_RUNNERS_DIR", RunnersDir(), "Location for runners"},
 		"OLLAMA_SCHED_SPREAD":      {"OLLAMA_SCHED_SPREAD", SchedSpread(), "Always schedule model across all GPUs"},
 		"OLLAMA_TMPDIR":            {"OLLAMA_TMPDIR", TmpDir(), "Location for temporary files"},
-
-		// Informational
-		"HTTP_PROXY":  {"HTTP_PROXY", String("HTTP_PROXY")(), "HTTP proxy"},
-		"HTTPS_PROXY": {"HTTPS_PROXY", String("HTTPS_PROXY")(), "HTTPS proxy"},
-		"NO_PROXY":    {"NO_PROXY", String("NO_PROXY")(), "No proxy"},
 	}
-
-	if runtime.GOOS != "windows" {
-		// Windows environment variables are case-insensitive so there's no need to duplicate them
-		ret["http_proxy"] = EnvVar{"http_proxy", String("http_proxy")(), "HTTP proxy"}
-		ret["https_proxy"] = EnvVar{"https_proxy", String("https_proxy")(), "HTTPS proxy"}
-		ret["no_proxy"] = EnvVar{"no_proxy", String("no_proxy")(), "No proxy"}
-	}
-
 	if runtime.GOOS != "darwin" {
 		ret["CUDA_VISIBLE_DEVICES"] = EnvVar{"CUDA_VISIBLE_DEVICES", CudaVisibleDevices(), "Set which NVIDIA devices are visible"}
 		ret["HIP_VISIBLE_DEVICES"] = EnvVar{"HIP_VISIBLE_DEVICES", HipVisibleDevices(), "Set which AMD devices are visible"}
@@ -267,7 +266,6 @@ func AsMap() map[string]EnvVar {
 		ret["HSA_OVERRIDE_GFX_VERSION"] = EnvVar{"HSA_OVERRIDE_GFX_VERSION", HsaOverrideGfxVersion(), "Override the gfx used for all detected AMD GPUs"}
 		ret["OLLAMA_INTEL_GPU"] = EnvVar{"OLLAMA_INTEL_GPU", IntelGPU(), "Enable experimental Intel GPU detection"}
 	}
-
 	return ret
 }

--- a/envconfig/config_test.go
+++ b/envconfig/config_test.go
@@ -215,40 +215,6 @@ func TestKeepAlive(t *testing.T) {
 	}
 }

-func TestLoadTimeout(t *testing.T) {
-	defaultTimeout := 5 * time.Minute
-	cases := map[string]time.Duration{
-		"":       defaultTimeout,
-		"1s":     time.Second,
-		"1m":     time.Minute,
-		"1h":     time.Hour,
-		"5m0s":   defaultTimeout,
-		"1h2m3s": 1*time.Hour + 2*time.Minute + 3*time.Second,
-		"0":      time.Duration(math.MaxInt64),
-		"60":     60 * time.Second,
-		"120":    2 * time.Minute,
-		"3600":   time.Hour,
-		"-0":     time.Duration(math.MaxInt64),
-		"-1":     time.Duration(math.MaxInt64),
-		"-1m":    time.Duration(math.MaxInt64),
-		// invalid values
-		" ":   defaultTimeout,
-		"???": defaultTimeout,
-		"1d":  defaultTimeout,
-		"1y":  defaultTimeout,
-		"1w":  defaultTimeout,
-	}
-
-	for tt, expect := range cases {
-		t.Run(tt, func(t *testing.T) {
-			t.Setenv("OLLAMA_LOAD_TIMEOUT", tt)
-			if actual := LoadTimeout(); actual != expect {
-				t.Errorf("%s: expected %s, got %s", tt, expect, actual)
-			}
-		})
-	}
-}
-
 func TestVar(t *testing.T) {
 	cases := map[string]string{
 		"value":       "value",
--- a/examples/langchain-python-rag-privategpt/requirements.txt
+++ b/examples/langchain-python-rag-privategpt/requirements.txt
@@ -1,6 +1,6 @@
 langchain==0.0.274
 gpt4all==1.0.8
-chromadb==0.5.0
+chromadb==0.4.7
 llama-cpp-python==0.1.81
 urllib3==2.0.4
 PyMuPDF==1.23.5
@@ -12,4 +12,4 @@ pandoc==2.3
 pypandoc==1.11
 tqdm==4.66.1
 sentence_transformers==2.2.2
-numpy>=1.22.2 # not directly required, pinned by Snyk to avoid a vulnerability
+numpy>=1.22.2 # not directly required, pinned by Snyk to avoid a vulnerability
--- a/examples/python-grounded-factuality-rag-check/README.md
+++ b/examples/python-grounded-factuality-rag-check/README.md
@@ -1,93 +0,0 @@
-# RAG Hallucination Checker using Bespoke-Minicheck
-
-This example allows the user to ask questions related to a document, which can be specified via an article url. Relevant chunks are retreived from the document and given to `llama3.1` as context to answer the question. Then each sentence in the answer is checked against the retrieved chunks using `bespoke-minicheck` to ensure that the answer does not contain hallucinations. 
-
-## Running the Example
-
-1. Ensure `all-minilm` (embedding) `llama3.1` (chat) and `bespoke-minicheck` (check) models installed:
-
-   ```bash
-   ollama pull all-minilm
-   ollama pull llama3.1
-   ollama pull bespoke-minicheck
-   ```
-
-2. Install the dependencies.
-
-   ```bash
-   pip install -r requirements.txt
-   ```
-
-3. Run the example:
-
-   ```bash
-   python main.py
-   ```
-
-## Expected Output
-
-```text
-Enter the URL of an article you want to chat with, or press Enter for default example:
-
-Loaded, chunked, and embedded text from https://www.theverge.com/2024/9/12/24242439/openai-o1-model-reasoning-strawberry-chatgpt.
-
-Enter your question or type quit: Who is the CEO of openai?
-
-Retrieved chunks:
-OpenAI is releasing a new model called o1 , the first in a planned series of “ reasoning ” models that have been trained to answer more complex questions , faster than a human can . It ’ s being released alongside o1-mini , a smaller , cheaper version . And yes , if you ’ re steeped in AI rumors : this is , in fact , the extremely hyped Strawberry model . For OpenAI , o1 represents a step toward its broader goal of human-like artificial intelligence .
-
-OpenAI is releasing a new model called o1 , the first in a planned series of “ reasoning ” models that have been trained to answer more complex questions , faster than a human can . It ’ s being released alongside o1-mini , a smaller , cheaper version . And yes , if you ’ re steeped in AI rumors : this is , in fact , the extremely hyped Strawberry model . For OpenAI , o1 represents a step toward its broader goal of human-like artificial intelligence . More practically , it does a better job at writing code and solving multistep problems than previous models . But it ’ s also more expensive and slower to use than GPT-4o . OpenAI is calling this release of o1 a “ preview ” to emphasize how nascent it is . ChatGPT Plus and Team users get access to both o1-preview and o1-mini starting today , while Enterprise and Edu users will get access early next week .
-
-More practically , it does a better job at writing code and solving multistep problems than previous models . But it ’ s also more expensive and slower to use than GPT-4o . OpenAI is calling this release of o1 a “ preview ” to emphasize how nascent it is . ChatGPT Plus and Team users get access to both o1-preview and o1-mini starting today , while Enterprise and Edu users will get access early next week . OpenAI says it plans to bring o1-mini access to all the free users of ChatGPT but hasn ’ t set a release date yet . Developer access to o1 is really expensive : In the API , o1-preview is $ 15 per 1 million input tokens , or chunks of text parsed by the model , and $ 60 per 1 million output tokens . For comparison , GPT-4o costs $ 5 per 1 million input tokens and $ 15 per 1 million output tokens .
-
-OpenAI says it plans to bring o1-mini access to all the free users of ChatGPT but hasn ’ t set a release date yet . Developer access to o1 is really expensive : In the API , o1-preview is $ 15 per 1 million input tokens , or chunks of text parsed by the model , and $ 60 per 1 million output tokens . For comparison , GPT-4o costs $ 5 per 1 million input tokens and $ 15 per 1 million output tokens . The training behind o1 is fundamentally different from its predecessors , OpenAI ’ s research lead , Jerry Tworek , tells me , though the company is being vague about the exact details . He says o1 “ has been trained using a completely new optimization algorithm and a new training dataset specifically tailored for it. ” Image : OpenAI OpenAI taught previous GPT models to mimic patterns from its training data .
-
-LLM Answer:
-The text does not mention the CEO of OpenAI. It only discusses the release of a new model called o1 and some details about it, but does not provide information on the company's leadership.
-
-LLM Claim: The text does not mention the CEO of OpenAI.
-Is this claim supported by the context according to bespoke-minicheck? Yes
-
-LLM Claim: It only discusses the release of a new model called o1 and some details about it, but does not provide information on the company's leadership.
-Is this claim supported by the context according to bespoke-minicheck? No
-```
-
-The second claim is unsupported since the text mentions the research lead. 
-
-Another tricky example:
-
-```text
-
-Enter your question or type quit: what sets o1 apart from gpt-4o?
-
-Retrieved chunks: 
-OpenAI says it plans to bring o1-mini access to all the free users of ChatGPT but hasn ’ t set a release date yet . Developer access to o1 is really expensive : In the API , o1-preview is $ 15 per 1 million input tokens , or chunks of text parsed by the model , and $ 60 per 1 million output tokens . For comparison , GPT-4o costs $ 5 per 1 million input tokens and $ 15 per 1 million output tokens . The training behind o1 is fundamentally different from its predecessors , OpenAI ’ s research lead , Jerry Tworek , tells me , though the company is being vague about the exact details . He says o1 “ has been trained using a completely new optimization algorithm and a new training dataset specifically tailored for it. ” Image : OpenAI OpenAI taught previous GPT models to mimic patterns from its training data .
-
-He says OpenAI also tested o1 against a qualifying exam for the International Mathematics Olympiad , and while GPT-4o only correctly solved only 13 percent of problems , o1 scored 83 percent . “ We can ’ t say we solved hallucinations ” In online programming contests known as Codeforces competitions , this new model reached the 89th percentile of participants , and OpenAI claims the next update of this model will perform “ similarly to PhD students on challenging benchmark tasks in physics , chemistry and biology. ” At the same time , o1 is not as capable as GPT-4o in a lot of areas . It doesn ’ t do as well on factual knowledge about the world .
-
-More practically , it does a better job at writing code and solving multistep problems than previous models . But it ’ s also more expensive and slower to use than GPT-4o . OpenAI is calling this release of o1 a “ preview ” to emphasize how nascent it is . ChatGPT Plus and Team users get access to both o1-preview and o1-mini starting today , while Enterprise and Edu users will get access early next week . OpenAI says it plans to bring o1-mini access to all the free users of ChatGPT but hasn ’ t set a release date yet . Developer access to o1 is really expensive : In the API , o1-preview is $ 15 per 1 million input tokens , or chunks of text parsed by the model , and $ 60 per 1 million output tokens . For comparison , GPT-4o costs $ 5 per 1 million input tokens and $ 15 per 1 million output tokens .
-
-“ We can ’ t say we solved hallucinations ” In online programming contests known as Codeforces competitions , this new model reached the 89th percentile of participants , and OpenAI claims the next update of this model will perform “ similarly to PhD students on challenging benchmark tasks in physics , chemistry and biology. ” At the same time , o1 is not as capable as GPT-4o in a lot of areas . It doesn ’ t do as well on factual knowledge about the world . It also doesn ’ t have the ability to browse the web or process files and images . Still , the company believes it represents a brand-new class of capabilities . It was named o1 to indicate “ resetting the counter back to 1. ” “ I ’ m gon na be honest : I think we ’ re terrible at naming , traditionally , ” McGrew says .
-LLM Answer: According to the text, several things set o1 apart from GPT-4o:
-
-* In online programming contests (Codeforces competitions), o1 scored 83% correct solutions compared to GPT-4o's 13%.
-* The training behind o1 is "fundamentally different" from its predecessors, including a completely new optimization algorithm and a new training dataset specifically tailored for it.
-* o1 has been shown to perform similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology, while GPT-4o does not have this level of performance.
-* However, o1 is also less capable than GPT-4o in some areas, particularly with regard to factual knowledge about the world.
-
-LLM Claim: According to the text, several things set o1 apart from GPT-4o:
-
-* In online programming contests (Codeforces competitions), o1 scored 83% correct solutions compared to GPT-4o's 13%.
-Is this claim supported by the context according to bespoke-minicheck? Yes
-
-LLM Claim: * The training behind o1 is "fundamentally different" from its predecessors, including a completely new optimization algorithm and a new training dataset specifically tailored for it.
-Is this claim supported by the context according to bespoke-minicheck? Yes
-
-LLM Claim: * o1 has been shown to perform similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology, while GPT-4o does not have this level of performance.
-Is this claim supported by the context according to bespoke-minicheck? No
-
-LLM Claim: * However, o1 is also less capable than GPT-4o in some areas, particularly with regard to factual knowledge about the world.
-Is this claim supported by the context according to bespoke-minicheck? Yes
-```
-
-We see that the third claim "* o1 has been shown to perform similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology, while GPT-4o does not have this level of performance." is not supported by the context. This is because the context only mentions that o1 "is claimed to perform" which is different from "has been shown to perform".
--- a/examples/python-grounded-factuality-rag-check/main.py
+++ b/examples/python-grounded-factuality-rag-check/main.py
@@ -1,137 +0,0 @@
-import ollama
-import warnings
-from mattsollamatools import chunker
-from newspaper import Article
-import numpy as np
-from sklearn.neighbors import NearestNeighbors
-import nltk
-
-warnings.filterwarnings(
-    "ignore", category=FutureWarning, module="transformers.tokenization_utils_base"
-)
-nltk.download("punkt_tab", quiet=True)
-
-
-def getArticleText(url):
-    """Gets the text of an article from a URL.
-
-    Often there are a bunch of ads and menus on pages for a news article.
-    This uses newspaper3k to get just the text of just the article.
-    """
-    article = Article(url)
-    article.download()
-    article.parse()
-    return article.text
-
-
-def knn_search(question_embedding, embeddings, k=5):
-    """Performs K-nearest neighbors (KNN) search"""
-    X = np.array(
-        [item["embedding"] for article in embeddings for item in article["embeddings"]]
-    )
-    source_texts = [
-        item["source"] for article in embeddings for item in article["embeddings"]
-    ]
-
-    # Fit a KNN model on the embeddings
-    knn = NearestNeighbors(n_neighbors=k, metric="cosine")
-    knn.fit(X)
-
-    # Find the indices and distances of the k-nearest neighbors.
-    _, indices = knn.kneighbors(question_embedding, n_neighbors=k)
-
-    # Get the indices and source texts of the best matches
-    best_matches = [(indices[0][i], source_texts[indices[0][i]]) for i in range(k)]
-
-    return best_matches
-
-
-def check(document, claim):
-    """Checks if the claim is supported by the document by calling bespoke-minicheck.
-
-    Returns Yes/yes if the claim is supported by the document, No/no otherwise.
-    Support for logits will be added in the future.
-
-    bespoke-minicheck's system prompt is defined as:
-      'Determine whether the provided claim is consistent with the corresponding
-      document. Consistency in this context implies that all information presented in the claim
-      is substantiated by the document. If not, it should be considered inconsistent. Please
-      assess the claim's consistency with the document by responding with either "Yes" or "No".'
-
-    bespoke-minicheck's user prompt is defined as:
-      "Document: {document}\nClaim: {claim}"
-    """
-    prompt = f"Document: {document}\nClaim: {claim}"
-    response = ollama.generate(
-        model="bespoke-minicheck", prompt=prompt, options={"num_predict": 2, "temperature": 0.0}
-    )
-    return response["response"].strip()
-
-
-if __name__ == "__main__":
-    allEmbeddings = []
-    default_url = "https://www.theverge.com/2024/9/12/24242439/openai-o1-model-reasoning-strawberry-chatgpt"
-    user_input = input(
-        "Enter the URL of an article you want to chat with, or press Enter for default example: "
-    )
-    article_url = user_input.strip() if user_input.strip() else default_url
-    article = {}
-    article["embeddings"] = []
-    article["url"] = article_url
-    text = getArticleText(article_url)
-    chunks = chunker(text)
-
-    # Embed (batch) chunks using ollama
-    embeddings = ollama.embed(model="all-minilm", input=chunks)["embeddings"]
-
-    for chunk, embedding in zip(chunks, embeddings):
-        item = {}
-        item["source"] = chunk
-        item["embedding"] = embedding
-        item["sourcelength"] = len(chunk)
-        article["embeddings"].append(item)
-
-    allEmbeddings.append(article)
-
-    print(f"\nLoaded, chunked, and embedded text from {article_url}.\n")
-
-    while True:
-        # Input a question from the user
-        # For example, "Who is the chief research officer?"
-        question = input("Enter your question or type quit: ")
-
-        if question.lower() == "quit":
-            break
-
-        # Embed the user's question using ollama.embed
-        question_embedding = ollama.embed(model="all-minilm", input=question)[
-            "embeddings"
-        ]
-
-        # Perform KNN search to find the best matches (indices and source text)
-        best_matches = knn_search(question_embedding, allEmbeddings, k=4)
-
-        sourcetext = "\n\n".join([source_text for (_, source_text) in best_matches])
-
-        print(f"\nRetrieved chunks: \n{sourcetext}\n")
-
-        # Give the retreived chunks and question to the chat model
-        system_prompt = f"Only use the following information to answer the question. Do not use anything else: {sourcetext}"
-
-        ollama_response = ollama.generate(
-            model="llama3.1",
-            prompt=question,
-            system=system_prompt,
-            options={"stream": False},
-        )
-
-        answer = ollama_response["response"]
-        print(f"LLM Answer:\n{answer}\n")
-
-        # Check each sentence in the response for grounded factuality
-        if answer:
-            for claim in nltk.sent_tokenize(answer):
-                print(f"LLM Claim: {claim}")
-                print(
-                    f"Is this claim supported by the context according to bespoke-minicheck? {check(sourcetext, claim)}\n"
-                )
--- a/examples/python-grounded-factuality-rag-check/requirements.txt
+++ b/examples/python-grounded-factuality-rag-check/requirements.txt
@@ -1,8 +0,0 @@
-ollama
-lxml==5.3.0
-lxml_html_clean==0.2.2
-mattsollamatools==0.0.25
-newspaper3k==0.2.8
-nltk==3.9.1
-numpy==1.26.4
-scikit-learn==1.5.2
--- a/examples/python-grounded-factuality-simple-check/main.py
+++ b/examples/python-grounded-factuality-simple-check/main.py
@@ -1,53 +0,0 @@
-"""Simple example to demonstrate how to use the bespoke-minicheck model."""
-
-import ollama
-
-# NOTE: ollama must be running for this to work, start the ollama app or run `ollama serve`
-
-
-def check(document, claim):
-    """Checks if the claim is supported by the document by calling bespoke-minicheck.
-
-    Returns Yes/yes if the claim is supported by the document, No/no otherwise.
-    Support for logits will be added in the future.
-
-    bespoke-minicheck's system prompt is defined as:
-      'Determine whether the provided claim is consistent with the corresponding
-      document. Consistency in this context implies that all information presented in the claim
-      is substantiated by the document. If not, it should be considered inconsistent. Please
-      assess the claim's consistency with the document by responding with either "Yes" or "No".'
-
-    bespoke-minicheck's user prompt is defined as:
-      "Document: {document}\nClaim: {claim}"
-    """
-    prompt = f"Document: {document}\nClaim: {claim}"
-    response = ollama.generate(
-        model="bespoke-minicheck", prompt=prompt, options={"num_predict": 2, "temperature": 0.0}
-    )
-    return response["response"].strip()
-
-
-def get_user_input(prompt):
-    user_input = input(prompt)
-    if not user_input:
-        exit()
-    print()
-    return user_input
-
-
-def main():
-    while True:
-        # Get a document from the user (e.g. "Ryan likes running and biking.")
-        document = get_user_input("Enter a document: ")
-        # Get a claim from the user (e.g. "Ryan likes to run.")
-        claim = get_user_input("Enter a claim: ")
-        # Check if the claim is supported by the document
-        grounded_factuality_check = check(document, claim)
-        print(
-            f"Is the claim supported by the document according to bespoke-minicheck? {grounded_factuality_check}"
-        )
-        print("\n\n")
-
-
-if __name__ == "__main__":
-    main()
--- a/examples/python-grounded-factuality-simple-check/readme.md
+++ b/examples/python-grounded-factuality-simple-check/readme.md
@@ -1,54 +0,0 @@
-# Simple Bespoke-Minicheck Example
-
-`bespoke-minicheck` is a model for checking if a claim is supported by a document. It is used through the **generate** endpoint, which is called in this example with a `prompt` that includes the expected formatting of the user input. 
-
-## Running the Example
-
-1. Ensure you have the `bespoke-minicheck` model installed:
-
-   ```bash
-   ollama pull bespoke-minicheck
-   ```
-
-2. Install the dependencies:
-
-   ```bash
-   pip install -r requirements.txt
-   ```
-
-3. Run the program:
-
-   ```bash
-   python main.py
-   ```
-
-4. Enter a document and a claim when prompted:
-
-   ```bash
-   Enter a document: Roses are red.
-
-   Enter a claim: Roses are blue. 
-   ```
-
-   The claim and document are then given to the `bespoke-minicheck` as inputs, which then generates a response (Yes or No) on whether the claim is supported by the document.
-
-   ```bash
-   Is the claim supported by the document according to bespoke-minicheck? No
-   ```
-
-## More Examples
-
-Document ([source](https://en.wikipedia.org/wiki/Apple_I)): 
-> The Apple Computer 1 (Apple-1[a]), later known predominantly as the Apple I(written with a Roman numeral),[b] is an 8-bit motherboard-only personal computer designed by Steve Wozniak[5][6] and released by the Apple Computer Company (now Apple Inc.) in 1976. The company was initially formed to sell the Apple I – its first product – and would later become the world's largest technology company.[7] The idea of starting a company and selling the computer came from Wozniak's friend and Apple co-founder Steve Jobs.[8][9] One of the main innovations of the Apple I was that it included video display terminal circuitry on its circuit board, allowing it to connect to a low-cost composite video monitor or television, instead of an expensive computer terminal, compared to most existing computers at the time.
-
-Claim: 
->The Apple I is a 16-bit computer.
-
-Expected output:
->Is the claim supported by the document according to bespoke-minicheck? **No**
-
-Claim: 
->Apple was originally called the Apple Computer Company.
-
-Expected output:
->Is the claim supported by the document according to bespoke-minicheck? **Yes**
--- a/examples/python-grounded-factuality-simple-check/requirements.txt
+++ b/examples/python-grounded-factuality-simple-check/requirements.txt
@@ -1 +0,0 @@
-ollama
--- a/examples/python-loganalysis/Modelfile
+++ b/examples/python-loganalysis/Modelfile
@@ -4,5 +4,5 @@ SYSTEM """
 You are a log file analyzer. You will receive a set of lines from a log file for some software application, find the errors and other interesting aspects of the logs, and explain them so a new user can understand what they mean. If there are any steps they can do to resolve them, list the steps in your answer.
 """

-PARAMETER temperature 0.3
+PARAMETER TEMPERATURE 0.3

--- a/examples/python-loganalysis/readme.md
+++ b/examples/python-loganalysis/readme.md
@@ -21,8 +21,6 @@ You can try this with the `logtest.logfile` file included in this directory.
 2. Install the Python Requirements.

   ```bash
-   python3 -m venv .venv
-   source .venv/bin/activate
   pip install -r requirements.txt
   ```

--- a/examples/python-loganalysis/requirements.txt
+++ b/examples/python-loganalysis/requirements.txt
@@ -1 +1 @@
-Requests>=2.32.3
+Requests==2.31.0
--- a/gpu/amd_linux.go
+++ b/gpu/amd_linux.go
@@ -5,7 +5,6 @@ import (
 	"errors"
 	"fmt"
 	"io"
-	"io/fs"
 	"log/slog"
 	"os"
 	"path/filepath"
@@ -360,10 +359,6 @@ func AMDGetGPUInfo() []RocmGPUInfo {
 	if len(resp) == 0 {
 		slog.Info("no compatible amdgpu devices detected")
 	}
-	if err := verifyKFDDriverAccess(); err != nil {
-		slog.Error("amdgpu devices detected but permission problems block access", "error", err)
-		return nil
-	}
 	return resp
 }

@@ -460,19 +455,3 @@ func getFreeMemory(usedFile string) (uint64, error) {
 	}
 	return usedMemory, nil
 }
-
-func verifyKFDDriverAccess() error {
-	// Verify we have permissions - either running as root, or we have group access to the driver
-	fd, err := os.OpenFile("/dev/kfd", os.O_RDWR, 0o666)
-	if err != nil {
-		if errors.Is(err, fs.ErrPermission) {
-			return fmt.Errorf("permissions not set up properly.  Either run ollama as root, or add you user account to the render group. %w", err)
-		} else if errors.Is(err, fs.ErrNotExist) {
-			// Container runtime failure?
-			return fmt.Errorf("kfd driver not loaded.  If running in a container, remember to include '--device /dev/kfd --device /dev/dri'")
-		}
-		return fmt.Errorf("failed to check permission on /dev/kfd: %w", err)
-	}
-	fd.Close()
-	return nil
-}
--- a/gpu/assets.go
+++ b/gpu/assets.go
@@ -0,0 +1,148 @@
+package gpu
+
+import (
+	"errors"
+	"fmt"
+	"log/slog"
+	"os"
+	"path/filepath"
+	"runtime"
+	"strconv"
+	"strings"
+	"sync"
+	"syscall"
+	"time"
+
+	"github.com/ollama/ollama/envconfig"
+)
+
+var (
+	lock        sync.Mutex
+	payloadsDir = ""
+)
+
+func PayloadsDir() (string, error) {
+	lock.Lock()
+	defer lock.Unlock()
+	var err error
+	if payloadsDir == "" {
+		runnersDir := envconfig.RunnersDir()
+
+		if runnersDir != "" {
+			payloadsDir = runnersDir
+			return payloadsDir, nil
+		}
+
+		// The remainder only applies on non-windows where we still carry payloads in the main executable
+		cleanupTmpDirs()
+		tmpDir := envconfig.TmpDir()
+		if tmpDir == "" {
+			tmpDir, err = os.MkdirTemp("", "ollama")
+			if err != nil {
+				return "", fmt.Errorf("failed to generate tmp dir: %w", err)
+			}
+		} else {
+			err = os.MkdirAll(tmpDir, 0o755)
+			if err != nil {
+				return "", fmt.Errorf("failed to generate tmp dir %s: %w", tmpDir, err)
+			}
+		}
+
+		// Track our pid so we can clean up orphaned tmpdirs
+		n := filepath.Join(tmpDir, "ollama.pid")
+		if err := os.WriteFile(n, []byte(strconv.Itoa(os.Getpid())), 0o644); err != nil {
+			return "", fmt.Errorf("failed to write pid file %s: %w", n, err)
+		}
+
+		// We create a distinct subdirectory for payloads within the tmpdir
+		// This will typically look like /tmp/ollama3208993108/runners on linux
+		payloadsDir = filepath.Join(tmpDir, "runners")
+	}
+	return payloadsDir, nil
+}
+
+// Best effort to clean up prior tmpdirs
+func cleanupTmpDirs() {
+	matches, err := filepath.Glob(filepath.Join(os.TempDir(), "ollama*", "ollama.pid"))
+	if err != nil {
+		return
+	}
+
+	for _, match := range matches {
+		raw, err := os.ReadFile(match)
+		if errors.Is(err, os.ErrNotExist) {
+			slog.Debug("not a ollama runtime directory, skipping", "path", match)
+			continue
+		} else if err != nil {
+			slog.Warn("could not read ollama.pid, skipping", "path", match, "error", err)
+			continue
+		}
+
+		pid, err := strconv.Atoi(string(raw))
+		if err != nil {
+			slog.Warn("invalid pid, skipping", "path", match, "error", err)
+			continue
+		}
+
+		p, err := os.FindProcess(pid)
+		if err == nil && !errors.Is(p.Signal(syscall.Signal(0)), os.ErrProcessDone) {
+			slog.Warn("process still running, skipping", "pid", pid, "path", match)
+			continue
+		}
+
+		if err := os.Remove(match); err != nil {
+			slog.Warn("could not cleanup stale pidfile", "path", match, "error", err)
+		}
+
+		runners := filepath.Join(filepath.Dir(match), "runners")
+		if err := os.RemoveAll(runners); err != nil {
+			slog.Warn("could not cleanup stale runners", "path", runners, "error", err)
+		}
+
+		if err := os.Remove(filepath.Dir(match)); err != nil {
+			slog.Warn("could not cleanup stale tmpdir", "path", filepath.Dir(match), "error", err)
+		}
+	}
+}
+
+func Cleanup() {
+	lock.Lock()
+	defer lock.Unlock()
+	runnersDir := envconfig.RunnersDir()
+	if payloadsDir != "" && runnersDir == "" && runtime.GOOS != "windows" {
+		// We want to fully clean up the tmpdir parent of the payloads dir
+		tmpDir := filepath.Clean(filepath.Join(payloadsDir, ".."))
+		slog.Debug("cleaning up", "dir", tmpDir)
+		err := os.RemoveAll(tmpDir)
+		if err != nil {
+			// On windows, if we remove too quickly the llama.dll may still be in-use and fail to remove
+			time.Sleep(1000 * time.Millisecond)
+			err = os.RemoveAll(tmpDir)
+			if err != nil {
+				slog.Warn("failed to clean up", "dir", tmpDir, "err", err)
+			}
+		}
+	}
+}
+
+func UpdatePath(dir string) {
+	if runtime.GOOS == "windows" {
+		tmpDir := filepath.Dir(dir)
+		pathComponents := strings.Split(os.Getenv("PATH"), ";")
+		i := 0
+		for _, comp := range pathComponents {
+			if strings.EqualFold(comp, dir) {
+				return
+			}
+			// Remove any other prior paths to our temp dir
+			if !strings.HasPrefix(strings.ToLower(comp), strings.ToLower(tmpDir)) {
+				pathComponents[i] = comp
+				i++
+			}
+		}
+		newPath := strings.Join(append([]string{dir}, pathComponents...), ";")
+		slog.Info("updating", "PATH", newPath)
+		os.Setenv("PATH", newPath)
+	}
+	// linux and darwin rely on rpath
+}
--- a/gpu/gpu.go
+++ b/gpu/gpu.go
@@ -93,9 +93,10 @@ func initCudaHandles() *cudaHandles {
 		localAppData := os.Getenv("LOCALAPPDATA")
 		cudartMgmtPatterns = []string{filepath.Join(localAppData, "Programs", "Ollama", CudartMgmtName)}
 	}
-	libDir := LibraryDir()
-	if libDir != "" {
-		cudartMgmtPatterns = []string{filepath.Join(libDir, CudartMgmtName)}
+	tmpDir, _ := PayloadsDir()
+	if tmpDir != "" {
+		// TODO - add "payloads" for subprocess
+		cudartMgmtPatterns = []string{filepath.Join(tmpDir, "cuda*", CudartMgmtName)}
 	}
 	cudartMgmtPatterns = append(cudartMgmtPatterns, CudartGlobs...)

@@ -205,16 +206,13 @@ func GetGPUInfo() GpuInfoList {
 		if err != nil {
 			slog.Warn("error looking up system memory", "error", err)
 		}
-		depPath := LibraryDir()
-
 		cpus = []CPUInfo{
 			{
 				GpuInfo: GpuInfo{
-					memInfo:        mem,
-					Library:        "cpu",
-					Variant:        cpuCapability.String(),
-					ID:             "0",
-					DependencyPath: depPath,
+					memInfo: mem,
+					Library: "cpu",
+					Variant: cpuCapability.String(),
+					ID:      "0",
 				},
 			},
 		}
@@ -227,6 +225,8 @@ func GetGPUInfo() GpuInfoList {
 			return GpuInfoList{cpus[0].GpuInfo}
 		}

+		depPath := LibraryDir()
+
 		// Load ALL libraries
 		cHandles = initCudaHandles()

--- a/integration/concurrency_test.go
+++ b/integration/concurrency_test.go
@@ -42,7 +42,7 @@ func TestMultiModelConcurrency(t *testing.T) {
 		}
 		resp = [2][]string{
 			{"sunlight"},
-			{"england", "english", "massachusetts", "pilgrims", "british"},
+			{"england", "english", "massachusetts", "pilgrims", "british", "festival"},
 		}
 	)
 	var wg sync.WaitGroup
--- a/integration/utils_test.go
+++ b/integration/utils_test.go
@@ -275,7 +275,7 @@ func DoGenerate(ctx context.Context, t *testing.T, client *api.Client, genReq ap
 				break
 			}
 		}
-		require.True(t, atLeastOne, "none of %v found in %s", anyResp, response)
+		require.True(t, atLeastOne, "%s: none of %v found in %s", genReq.Model, anyResp, response)
 		slog.Info("test pass", "model", genReq.Model, "prompt", genReq.Prompt, "contains", anyResp, "response", response)
 	case <-ctx.Done():
 		t.Error("outer test context done while waiting for generate")
--- a/llama/.gitignore
+++ b/llama/.gitignore
@@ -0,0 +1,3 @@
+*.bin
+*.gguf
+build/
--- a/llama/Makefile
+++ b/llama/Makefile
@@ -0,0 +1,388 @@
+OS := $(shell uname -s)
+ARCH := $(or $(ARCH), $(shell uname -m))
+ifeq ($(ARCH),x86_64)
+	ARCH := amd64
+endif
+ifneq (,$(findstring MINGW,$(OS))$(findstring MSYS,$(OS)))
+	OS := windows
+else ifeq ($(OS),Linux)
+	OS := linux
+else ifeq ($(OS),Darwin)
+	OS := darwin
+endif
+comma:= ,
+empty:=
+space:= $(empty) $(empty)
+
+export CGO_CFLAGS_ALLOW = -mfma|-mf16c
+export CGO_CXXFLAGS_ALLOW = -mfma|-mf16c
+export HIP_PLATFORM = amd
+
+SRC_DIR := $(dir $(abspath $(lastword $(MAKEFILE_LIST))))
+BUILD_DIR = $(SRC_DIR)build/$(OS)-$(ARCH)
+DIST_BASE = $(abspath $(SRC_DIR)/../dist/$(OS)-$(ARCH))
+RUNNERS_DIST_DIR = $(DIST_BASE)/ollama_runners
+RUNNERS_PAYLOAD_DIR = $(abspath $(SRC_DIR)/../llm/build/$(OS)/$(patsubst amd64,x86_64,$(ARCH)))
+RUNNERS_BUILD_DIR = $(BUILD_DIR)/ollama_runners
+DEFAULT_RUNNER := $(if $(and $(filter darwin,$(OS)),$(filter arm64,$(ARCH))),metal,cpu)
+
+CUDA_LIBS_SHORT := cublas cudart cublasLt
+ROCM_LIBS_SHORT := hipblas rocblas
+
+ifeq ($(OS),windows)
+	SRC_DIR := $(shell cygpath -m -s "$(SRC_DIR)")
+	OBJ_EXT := obj
+	SHARED_EXT := dll
+	EXE_EXT := .exe
+	SHARED_PREFIX := 
+	
+	# TODO needs work for multiple cuda versions on windows
+
+	CUDA_BASE_DIR := $(dir $(shell cygpath -m -s "$(CUDA_PATH)\.."))
+	CUDA_11=$(shell ls -d $(CUDA_BASE_DIR)/v11.? 2>/dev/null)
+	CUDA_12=$(shell ls -d $(CUDA_BASE_DIR)/v12.? 2>/dev/null)
+	CUDA_11_LIB_DIR := $(CUDA_11)/bin
+	CUDA_12_LIB_DIR := $(CUDA_12)/bin
+
+
+	NVCC := $(shell X=$$(which nvcc 2>/dev/null) && cygpath -m -s "$$X")
+	ifneq ($(HIP_PATH),)
+		HIP_LIB_DIR := $(shell cygpath -m -s "$(HIP_PATH)\bin")
+		# If HIP_PATH has spaces, hipcc trips over them when subprocessing
+		HIP_PATH := $(shell cygpath -m -s "$(HIP_PATH)\")
+		export HIP_PATH
+		HIPCC := $(HIP_PATH)bin/hipcc.bin.exe
+	endif
+	CP := cp
+	CUDA_LIBS = $(wildcard $(addsuffix 64*.$(SHARED_EXT),$(addprefix $(CUDA_LIB_DIR)/$(SHARED_PREFIX),$(CUDA_LIBS_SHORT))))
+else ifeq ($(OS),linux)
+	CP := cp -a
+	OBJ_EXT := o
+	SHARED_EXT := so
+	SHARED_PREFIX := lib
+	HIP_PATH?=/opt/rocm
+	HIP_LIB_DIR := $(HIP_PATH)/lib
+	HIPCC := $(shell X=$$(which hipcc 2>/dev/null) && echo $$X)
+	CUDA_PATH?=/usr/local/cuda
+	CUDA_11=$(shell ls -d $(CUDA_PATH)-11 2>/dev/null)
+	CUDA_12=$(shell ls -d $(CUDA_PATH)-12 2>/dev/null)
+	CUDA_11_LIB_DIR := $(CUDA_11)/lib64
+	CUDA_12_LIB_DIR := $(CUDA_12)/lib64
+else
+	OBJ_EXT := o
+	SHARED_EXT := so
+	CP := cp -a
+endif
+
+CUDA_11_LIBS = $(wildcard $(addsuffix .$(SHARED_EXT).*,$(addprefix $(CUDA_11_LIB_DIR)/$(SHARED_PREFIX),$(CUDA_LIBS_SHORT))))
+CUDA_12_LIBS = $(wildcard $(addsuffix .$(SHARED_EXT).*,$(addprefix $(CUDA_12_LIB_DIR)/$(SHARED_PREFIX),$(CUDA_LIBS_SHORT))))
+NVCC_11 = $(CUDA_11)/bin/nvcc
+NVCC_12 = $(CUDA_12)/bin/nvcc
+
+CUDA_DEPS_DIR = $(DIST_BASE)cuda/
+ROCM_DEPS_DIR = $(DIST_BASE)rocm/
+
+ifneq ($(CUDA_11),)
+	CUDA_11_VARIANT= _v11
+	CUDA_11_LIB_DEPS = $(addprefix $(CUDA_DEPS_DIR),$(notdir $(CUDA_11_LIBS)))
+endif
+ifneq ($(CUDA_12),)
+	CUDA_12_VARIANT= _v12
+	CUDA_12_LIB_DEPS = $(addprefix $(CUDA_DEPS_DIR),$(notdir $(CUDA_12_LIBS)))
+endif
+ifeq ($(OLLAMA_SKIP_ROCM_GENERATE),)
+ifneq ($(HIPCC),)
+	ROCM_VERSION := $(subst $(space),.,$(wordlist 1,2,$(subst .,$(space),$(word 3,$(subst -,$(space),$(filter HIP version: %,$(shell $(HIPCC) --version)))))))
+    ifneq (,$(ROCM_VERSION))
+		ROCM_VARIANT = _v$(ROCM_VERSION)
+	endif
+	ROCM_LIBS = $(wildcard $(addsuffix .$(SHARED_EXT),$(addprefix $(HIP_LIB_DIR)/$(SHARED_PREFIX),$(ROCM_LIBS_SHORT))))
+	ROCM_LIB_DEPS = $(addprefix $(ROCM_DEPS_DIR),$(notdir $(ROCM_LIBS)))
+	ROCBLAS_DEP_MANIFEST = $(ROCM_DEPS_DIR)/rocblas/library/TensileManifest.txt
+endif
+endif
+
+COMMON_SRCS := \
+	$(wildcard *.c) \
+	$(wildcard *.cpp)
+COMMON_HDRS := \
+	$(wildcard *.h) \
+	$(wildcard *.hpp)
+
+CUDA_SRCS := \
+	ggml-cuda.cu \
+	$(wildcard ggml-cuda/*.cu) \
+	$(wildcard ggml-cuda/template-instances/fattn-wmma*.cu) \
+	$(wildcard ggml-cuda/template-instances/mmq*.cu) \
+	$(wildcard ggml-cuda/template-instances/fattn-vec*q4_0-q4_0.cu) \
+	$(wildcard ggml-cuda/template-instances/fattn-vec*q8_0-q8_0.cu) \
+	$(wildcard ggml-cuda/template-instances/fattn-vec*f16-f16.cu) \
+	ggml.c ggml-backend.c ggml-alloc.c ggml-quants.c sgemm.cpp
+CUDA_HDRS := \
+	$(wildcard ggml-cuda/*.cuh)
+
+CUDA_11_OBJS := $(CUDA_SRCS:.cu=.cuda.$(OBJ_EXT))
+CUDA_11_OBJS := $(CUDA_11_OBJS:.c=.cuda.$(OBJ_EXT))
+CUDA_11_OBJS := $(addprefix $(BUILD_DIR)/cuda_v11/,$(CUDA_11_OBJS:.cpp=.cuda.$(OBJ_EXT)))
+CUDA_12_OBJS := $(CUDA_SRCS:.cu=.cuda.$(OBJ_EXT))
+CUDA_12_OBJS := $(CUDA_12_OBJS:.c=.cuda.$(OBJ_EXT))
+CUDA_12_OBJS := $(addprefix $(BUILD_DIR)/cuda_v12/,$(CUDA_12_OBJS:.cpp=.cuda.$(OBJ_EXT)))
+
+HIP_OBJS := $(CUDA_SRCS:.cu=.hip.$(OBJ_EXT))
+HIP_OBJS := $(HIP_OBJS:.c=.hip.$(OBJ_EXT))
+HIP_OBJS := $(addprefix $(BUILD_DIR)/,$(HIP_OBJS:.cpp=.hip.$(OBJ_EXT)))
+
+CUDA_FLAGS := \
+	-t4 \
+	-DGGML_CUDA_DMMV_X=32 \
+	-DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 \
+	-DGGML_USE_CUDA=1 \
+	-DGGML_SHARED=1 \
+	-DGGML_BUILD=1 \
+	-DGGML_USE_LLAMAFILE \
+	-D_GNU_SOURCE \
+	-DCMAKE_POSITION_INDEPENDENT_CODE=on \
+	-Wno-deprecated-gpu-targets \
+	--forward-unknown-to-host-compiler \
+	-use_fast_math \
+	-link \
+	-shared \
+	-I. \
+	-O3
+
+CUDA_11_FLAGS := \
+	--generate-code=arch=compute_50,code=[compute_50,sm_50] \
+	--generate-code=arch=compute_52,code=[compute_52,sm_52] \
+	--generate-code=arch=compute_53,code=[compute_53,sm_53] \
+	--generate-code=arch=compute_60,code=[compute_60,sm_60] \
+	--generate-code=arch=compute_61,code=[compute_61,sm_61] \
+	--generate-code=arch=compute_62,code=[compute_62,sm_62] \
+	--generate-code=arch=compute_70,code=[compute_70,sm_70] \
+	--generate-code=arch=compute_72,code=[compute_72,sm_72] \
+	--generate-code=arch=compute_75,code=[compute_75,sm_75] \
+	--generate-code=arch=compute_80,code=[compute_80,sm_80] \
+	--generate-code=arch=compute_86,code=[compute_86,sm_86]
+
+CUDA_12_FLAGS := \
+	--generate-code=arch=compute_60,code=[compute_60,sm_60] \
+	--generate-code=arch=compute_61,code=[compute_61,sm_61] \
+	--generate-code=arch=compute_62,code=[compute_62,sm_62] \
+	--generate-code=arch=compute_70,code=[compute_70,sm_70] \
+	--generate-code=arch=compute_72,code=[compute_72,sm_72] \
+	--generate-code=arch=compute_75,code=[compute_75,sm_75] \
+	--generate-code=arch=compute_80,code=[compute_80,sm_80] \
+	--generate-code=arch=compute_86,code=[compute_86,sm_86] \
+	--generate-code=arch=compute_87,code=[compute_87,sm_87] \
+	--generate-code=arch=compute_89,code=[compute_89,sm_89] \
+	--generate-code=arch=compute_90,code=[compute_90,sm_90] \
+	--generate-code=arch=compute_90a,code=[compute_90a,sm_90a] \
+	-DGGML_CUDA_USE_GRAPHS=on
+
+HIP_ARCHS := gfx900 gfx940 gfx941 gfx942 gfx1010 gfx1012 gfx1030 gfx1100 gfx1101 gfx1102
+LINUX_HIP_ARCHS := gfx906:xnack- gfx908:xnack- gfx90a:xnack+ gfx90a:xnack-
+
+HIP_FLAGS := \
+	-c \
+	-O3 \
+	-DGGML_USE_CUDA \
+	-DGGML_BUILD=1 \
+	-DGGML_SHARED=1 \
+	-DGGML_CUDA_DMMV_X=32 \
+	-DGGML_CUDA_MMV_Y=1 \
+	-DGGML_SCHED_MAX_COPIES=4 \
+	-DGGML_USE_HIPBLAS \
+	-DGGML_USE_LLAMAFILE \
+	-DHIP_FAST_MATH \
+	-DNDEBUG \
+	-DK_QUANTS_PER_ITERATION=2 \
+	-D_CRT_SECURE_NO_WARNINGS \
+	-DCMAKE_POSITION_INDEPENDENT_CODE=on \
+	-D_GNU_SOURCE \
+	-Wno-expansion-to-defined \
+	-Wno-invalid-noreturn \
+	-Wno-ignored-attributes \
+	-Wno-pass-failed \
+	-Wno-deprecated-declarations \
+	-Wno-unused-result \
+	-I. \
+	$(foreach arch, $(HIP_ARCHS), --offload-arch=$(arch))
+
+ifeq ($(OS),linux)
+	HIP_FLAGS += $(foreach arch, $(LINUX_HIP_ARCHS), --offload-arch=$(arch)) -fPIC -Wno-unused-function
+	CUDA_FLAGS += -fPIC -Wno-unused-function
+	NVCC_CFLAGS = $(CFLAGS) -Xcompiler -fPIC -D_GNU_SOURCE
+	NVCC_CXXFLAGS = $(CXXFLAGS) -Xcompiler -fPIC -D_GNU_SOURCE
+	HIPCC_CFLAGS = $(CFLAGS) -fPIC -D_GNU_SOURCE
+	HIPCC_CXXFLAGS = $(CXXFLAGS) -fPIC -D_GNU_SOURCE
+else ifeq ($(OS),windows)
+	HIP_FLAGS += -Xclang --dependent-lib=msvcrt
+	CFLAGS += -D_WIN32_WINNT=0x602
+	CXXFLAGS += -D_WIN32_WINNT=0x602
+	NVCC_CFLAGS = $(CFLAGS)
+	NVCC_CXXFLAGS = $(CXXFLAGS)
+	HIPCC_CFLAGS = $(CFLAGS)
+	HIPCC_CXXFLAGS = $(CXXFLAGS)
+endif
+
+ifeq ($(OLLAMA_SKIP_CPU_GENERATE),)
+RUNNERS := $(DEFAULT_RUNNER)
+ifeq ($(ARCH),amd64)
+	RUNNERS += cpu_avx cpu_avx2
+endif
+endif
+ifeq ($(OLLAMA_SKIP_CUDA_GENERATE),)
+ifneq ($(CUDA_11),)
+	RUNNERS += cuda_v11
+endif
+ifneq ($(CUDA_12),)
+	RUNNERS += cuda_v12
+endif
+endif
+ifeq ($(OLLAMA_SKIP_ROCM_GENERATE),)
+ifneq ($(HIPCC),)
+	RUNNERS += rocm$(ROCM_VARIANT)
+endif
+endif
+
+DIST_RUNNERS = $(addprefix $(RUNNERS_DIST_DIR)/,$(addsuffix /ollama_runner$(EXE_EXT),$(RUNNERS)))
+PAYLOAD_RUNNERS = $(addprefix $(RUNNERS_PAYLOAD_DIR)/,$(addsuffix /ollama_runner$(EXE_EXT).gz,$(addsuffix /bin,$(RUNNERS))))
+BUILD_RUNNERS = $(addprefix $(RUNNERS_BUILD_DIR)/,$(addsuffix /ollama_runner$(EXE_EXT),$(RUNNERS)))
+
+all: dist payload 
+
+dist: $(DIST_RUNNERS) $(ROCBLAS_DEP_MANIFEST)
+
+ifeq ($(OS),windows)
+# Unused on windows as we don't cary the payloads in the go binary
+payload:
+else
+payload: $(PAYLOAD_RUNNERS)
+endif
+
+runners: $(BUILD_RUNNERS)
+
+$(BUILD_DIR)/cuda_v11/%.cuda.$(OBJ_EXT): %.cu
+	@-mkdir -p $(dir $@)
+	$(NVCC_11) -c $(CUDA_FLAGS) $(CUDA_11_FLAGS) -o $@ $<
+
+$(BUILD_DIR)/cuda_v11/%.cuda.$(OBJ_EXT): %.c
+	@-mkdir -p $(dir $@)
+	$(NVCC_11) -c $(NVCC_CFLAGS) -o $@ $<
+
+$(BUILD_DIR)/cuda_v11/%.cuda.$(OBJ_EXT): %.cpp
+	@-mkdir -p $(dir $@)
+	$(NVCC_11) -c $(NVCC_CXXFLAGS) -o $@ $<
+
+$(BUILD_DIR)/cuda_v12/%.cuda.$(OBJ_EXT): %.cu
+	@-mkdir -p $(dir $@)
+	$(NVCC_12) -c $(CUDA_FLAGS) $(CUDA_12_FLAGS) -o $@ $<
+
+$(BUILD_DIR)/cuda_v12/%.cuda.$(OBJ_EXT): %.c
+	@-mkdir -p $(dir $@)
+	$(NVCC_12) -c $(NVCC_CFLAGS) -o $@ $<
+
+$(BUILD_DIR)/cuda_v12/%.cuda.$(OBJ_EXT): %.cpp
+	@-mkdir -p $(dir $@)
+	$(NVCC_12) -c $(NVCC_CXXFLAGS) -o $@ $<
+
+$(RUNNERS_DIST_DIR)/%: $(RUNNERS_BUILD_DIR)/%
+	@-mkdir -p $(dir $@)
+	cp $< $@
+
+$(RUNNERS_DIST_DIR)/cuda_v11/ollama_runner$(EXE_EXT): $(RUNNERS_DIST_DIR)/cuda_v11/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT)
+$(RUNNERS_PAYLOAD_DIR)/cuda_v11/bin/ollama_runner$(EXE_EXT).gz: $(RUNNERS_PAYLOAD_DIR)/cuda_v11/bin/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT).gz
+$(RUNNERS_DIST_DIR)/cuda_v12/ollama_runner$(EXE_EXT): $(RUNNERS_DIST_DIR)/cuda_v12/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT)
+$(RUNNERS_PAYLOAD_DIR)/cuda_v12/bin/ollama_runner$(EXE_EXT).gz: $(RUNNERS_PAYLOAD_DIR)/cuda_v12/bin/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT).gz
+
+$(RUNNERS_BUILD_DIR)/cuda_v11/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT): $(CUDA_11_OBJS) $(CUDA_11_LIB_DEPS) $(COMMON_HDRS) $(CUDA_HDRS)
+	@-mkdir -p $(dir $@)
+	$(NVCC_11) --shared -lcuda -L${CUDA_DEPS_DIR} $(foreach lib, $(CUDA_LIBS_SHORT), -l$(lib)) $(CUDA_FLAGS) $(CUDA_11_FLAGS) $(CUDA_11_OBJS) -o $@
+
+$(RUNNERS_BUILD_DIR)/cuda_v12/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT): $(CUDA_12_OBJS) $(CUDA_12_LIB_DEPS) $(COMMON_HDRS) $(CUDA_HDRS)
+	@-mkdir -p $(dir $@)
+	$(NVCC_12) --shared -lcuda -L${CUDA_DEPS_DIR} $(foreach lib, $(CUDA_LIBS_SHORT), -l$(lib)) $(CUDA_FLAGS) $(CUDA_12_FLAGS) $(CUDA_12_OBJS) -o $@
+
+$(CUDA_11_LIB_DEPS): 
+	@-mkdir -p $(dir $@)
+	$(CP) $(CUDA_11_LIB_DIR)/$(notdir $@)* $(dir $@)
+
+$(CUDA_12_LIB_DEPS): 
+	@-mkdir -p $(dir $@)
+	$(CP) $(CUDA_12_LIB_DIR)/$(notdir $@)* $(dir $@)
+
+$(BUILD_DIR)/%.hip.$(OBJ_EXT): %.cu
+	@-mkdir -p $(dir $@)
+	$(HIPCC) -c $(HIP_FLAGS) -o $@ $<
+
+$(BUILD_DIR)/%.hip.$(OBJ_EXT): %.c
+	@-mkdir -p $(dir $@)
+	$(HIPCC) -c $(HIPCC_CFLAGS) -o $@ $<
+
+$(BUILD_DIR)/%.hip.$(OBJ_EXT): %.cpp
+	@-mkdir -p $(dir $@)
+	$(HIPCC) -c $(HIPCC_CXXFLAGS) -o $@ $<
+
+$(RUNNERS_DIST_DIR)/rocm$(ROCM_VARIANT)/ollama_runner$(EXE_EXT): $(RUNNERS_DIST_DIR)/rocm$(ROCM_VARIANT)/$(SHARED_PREFIX)ggml_hipblas.$(SHARED_EXT)
+$(RUNNERS_PAYLOAD_DIR)/rocm$(ROCM_VARIANT)/bin/ollama_runner$(EXE_EXT).gz: $(RUNNERS_PAYLOAD_DIR)/rocm$(ROCM_VARIANT)/bin/$(SHARED_PREFIX)ggml_hipblas.$(SHARED_EXT).gz
+
+$(RUNNERS_BUILD_DIR)/rocm$(ROCM_VARIANT)/$(SHARED_PREFIX)ggml_hipblas.$(SHARED_EXT): $(HIP_OBJS) $(ROCM_LIB_DEPS) $(COMMON_HDRS) $(CUDA_HDRS)
+	@-mkdir -p $(dir $@)
+	$(HIPCC) --shared -lamdhip64 -L${ROCM_DEPS_DIR} $(foreach lib, $(ROCM_LIBS_SHORT), -l$(lib)) $(HIP_OBJS) -o $@
+
+$(ROCM_LIB_DEPS): 
+	@-mkdir -p $(dir $@)
+	$(CP) $(HIP_LIB_DIR)/$(notdir $@)* $(dir $@)
+
+$(RUNNERS_BUILD_DIR)/$(DEFAULT_RUNNER)/ollama_runner$(EXE_EXT): *.go ./runner/*.go $(COMMON_SRCS) $(COMMON_HDRS)
+	@-mkdir -p $(dir $@)
+	CGO_ENABLED=1 GOARCH=$(ARCH) go build -ldflags "-s -w" -o $@ ./runner
+
+$(RUNNERS_BUILD_DIR)/cpu_avx/ollama_runner$(EXE_EXT): *.go ./runner/*.go $(COMMON_SRCS) $(COMMON_HDRS)
+	@-mkdir -p $(dir $@)
+	CGO_ENABLED=1 GOARCH=$(ARCH) go build -ldflags "-s -w" -tags avx -o $@ ./runner
+
+$(RUNNERS_BUILD_DIR)/cpu_avx2/ollama_runner$(EXE_EXT): *.go ./runner/*.go $(COMMON_SRCS) $(COMMON_HDRS)
+	@-mkdir -p $(dir $@)
+	CGO_ENABLED=1 GOARCH=$(ARCH) go build -ldflags "-s -w" -tags avx,avx2 -o $@ ./runner
+
+$(RUNNERS_BUILD_DIR)/cuda_v11/ollama_runner$(EXE_EXT): $(RUNNERS_BUILD_DIR)/cuda_v11/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT) *.go ./runner/*.go $(COMMON_SRCS) $(COMMON_HDRS)
+	@-mkdir -p $(dir $@)
+	CGO_ENABLED=1 GOARCH=$(ARCH) CGO_LDFLAGS=-L"$(RUNNERS_BUILD_DIR)/cuda_v11/" go build -ldflags "-s -w" -tags avx,cuda -o $@ ./runner
+
+$(RUNNERS_BUILD_DIR)/cuda_v12/ollama_runner$(EXE_EXT): $(RUNNERS_BUILD_DIR)/cuda_v12/$(SHARED_PREFIX)ggml_cuda.$(SHARED_EXT) *.go ./runner/*.go $(COMMON_SRCS) $(COMMON_HDRS)
+	@-mkdir -p $(dir $@)
+	CGO_ENABLED=1 GOARCH=$(ARCH) CGO_LDFLAGS=-L"$(RUNNERS_BUILD_DIR)/cuda_v12/" go build -ldflags "-s -w" -tags avx,cuda -o $@ ./runner
+
+$(RUNNERS_BUILD_DIR)/rocm$(ROCM_VARIANT)/ollama_runner$(EXE_EXT): $(RUNNERS_BUILD_DIR)/rocm$(ROCM_VARIANT)/$(SHARED_PREFIX)ggml_hipblas.$(SHARED_EXT) *.go ./runner/*.go $(COMMON_SRCS) $(COMMON_HDRS)
+	@-mkdir -p $(dir $@)
+	CGO_ENABLED=1 GOARCH=$(ARCH) CGO_LDFLAGS=-L"$(RUNNERS_BUILD_DIR)/rocm$(ROCM_VARIANT)/" go build -ldflags "-s -w" -tags avx,rocm -o $@ ./runner
+
+
+$(ROCBLAS_DEP_MANIFEST):
+	@-mkdir -p $(dir $@)
+	@echo "Copying rocblas library..."
+	cd $(HIP_LIB_DIR)/rocblas/library/ && tar cf - . | (cd $(dir $@) && tar xf - )
+	@echo "rocblas library copy complete"
+
+
+$(RUNNERS_PAYLOAD_DIR)/%/bin/ollama_runner.gz: $(RUNNERS_BUILD_DIR)/%/ollama_runner
+	@-mkdir -p $(dir $@)
+	gzip --best -c $< > $@
+$(RUNNERS_PAYLOAD_DIR)/cuda_v11/bin/%.gz: $(RUNNERS_BUILD_DIR)/cuda_v11/%
+	@-mkdir -p $(dir $@)
+	gzip --best -c $< > $@
+$(RUNNERS_PAYLOAD_DIR)/cuda_v12/bin/%.gz: $(RUNNERS_BUILD_DIR)/cuda_v12/%
+	@-mkdir -p $(dir $@)
+	gzip --best -c $< > $@
+$(RUNNERS_PAYLOAD_DIR)/rocm$(ROCM_VARIANT)/bin/%.gz: $(RUNNERS_BUILD_DIR)/rocm$(ROCM_VARIANT)/%
+	@-mkdir -p $(dir $@)
+	gzip --best -c $< > $@
+
+clean:
+	rm -rf $(BUILD_DIR) $(DIST_RUNNERS) $(PAYLOAD_RUNNERS)
+
+.PHONY: all dist payload runners clean $(RUNNERS)
+
+# Handy debugging for make variables
+print-%:
+	@echo '$*=$($*)'
--- a/llama/README.md
+++ b/llama/README.md
@@ -0,0 +1,102 @@
+# `llama`
+
+> Note: this package is not used in Ollama yet. For now, see the [`llm`](https://github.com/ollama/ollama/tree/main/llm) package.
+
+This package integrates the [llama.cpp](https://github.com/ggerganov/llama.cpp) library as a Go package and makes it easy to build it with tags for different CPU and GPU processors.
+
+Supported:
+
+- [x] CPU
+- [x] avx, avx2
+- [x] macOS Metal
+- [x] Windows CUDA
+- [x] Windows ROCm
+- [x] Linux CUDA
+- [x] Linux ROCm
+- [x] Llava
+
+Extra build steps are required for CUDA and ROCm on Windows since `nvcc` and `hipcc` both require using msvc as the host compiler. For these shared libraries are created:
+
+- `ggml_cuda.dll` on Windows or `ggml_cuda.so` on Linux
+- `ggml_hipblas.dll` on Windows or `ggml_hipblas.so` on Linux
+
+> Note: it's important that memory is allocated and freed by the same compiler (e.g. entirely by code compiled with msvc or mingw). Issues from this should be rare, but there are some places where pointers are returned by the CUDA or HIP runtimes and freed elsewhere, causing a a crash. In a future change the same runtime should be used in both cases to avoid crashes.
+
+## Building
+
+```
+go build .
+```
+
+### AVX
+
+```shell
+go build -tags avx .
+```
+
+### AVX2
+
+```shell
+# go doesn't recognize `-mfma` as a valid compiler flag
+# see https://github.com/golang/go/issues/17895
+go env -w "CGO_CFLAGS_ALLOW=-mfma|-mf16c"
+go env -w "CGO_CXXFLAGS_ALLOW=-mfma|-mf16c"
+go build -tags=avx,avx2 .
+```
+
+## Linux
+
+### CUDA
+
+Install the [CUDA toolkit v11.3.1](https://developer.nvidia.com/cuda-11-3-1-download-archive):
+
+```shell
+make ggml_cuda.so
+go build -tags avx,cuda .
+```
+
+### ROCm
+
+Install the [CUDA toolkit v11.3.1](https://developer.nvidia.com/cuda-11-3-1-download-archive):
+
+```shell
+make ggml_hipblas.so
+go build -tags avx,rocm .
+```
+
+## Windows
+
+Download [w64devkit](https://github.com/skeeto/w64devkit/releases/latest) for a simple MinGW development environment.
+
+### CUDA
+
+Install the [CUDA toolkit v11.3.1](https://developer.nvidia.com/cuda-11-3-1-download-archive) then build the cuda code:
+
+```shell
+make ggml_cuda.dll
+go build -tags avx,cuda .
+```
+
+### ROCm
+
+Install [ROCm 5.7.1](https://rocm.docs.amd.com/en/docs-5.7.1/).
+
+```shell
+make ggml_hipblas.dll
+go build -tags avx,rocm .
+```
+
+## Building runners
+
+```shell
+# build all runners for this platform
+make -j
+```
+
+## Syncing with llama.cpp
+
+To update this package to the latest llama.cpp code, use the `sync.sh` script:
+
+```
+./sync.sh ../../llama.cpp
+```
--- a/llama/base64.hpp
+++ b/llama/base64.hpp
@@ -0,0 +1,392 @@
+/*
+This is free and unencumbered software released into the public domain.
+
+Anyone is free to copy, modify, publish, use, compile, sell, or
+distribute this software, either in source code form or as a compiled
+binary, for any purpose, commercial or non-commercial, and by any
+means.
+
+In jurisdictions that recognize copyright laws, the author or authors
+of this software dedicate any and all copyright interest in the
+software to the public domain. We make this dedication for the benefit
+of the public at large and to the detriment of our heirs and
+successors. We intend this dedication to be an overt act of
+relinquishment in perpetuity of all present and future rights to this
+software under copyright law.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+OTHER DEALINGS IN THE SOFTWARE.
+
+For more information, please refer to <http://unlicense.org>
+*/
+
+#ifndef PUBLIC_DOMAIN_BASE64_HPP_
+#define PUBLIC_DOMAIN_BASE64_HPP_
+
+#include <cstdint>
+#include <iterator>
+#include <stdexcept>
+#include <string>
+
+class base64_error : public std::runtime_error
+{
+public:
+    using std::runtime_error::runtime_error;
+};
+
+class base64
+{
+public:
+    enum class alphabet
+    {
+        /** the alphabet is detected automatically */
+        auto_,
+        /** the standard base64 alphabet is used */
+        standard,
+        /** like `standard` except that the characters `+` and `/` are replaced by `-` and `_` respectively*/
+        url_filename_safe
+    };
+
+    enum class decoding_behavior
+    {
+        /** if the input is not padded, the remaining bits are ignored */
+        moderate,
+        /** if a padding character is encounter decoding is finished */
+        loose
+    };
+
+    /**
+     Encodes all the elements from `in_begin` to `in_end` to `out`.
+
+     @warning The source and destination cannot overlap. The destination must be able to hold at least
+     `required_encode_size(std::distance(in_begin, in_end))`, otherwise the behavior depends on the output iterator.
+
+     @tparam Input_iterator the source; the returned elements are cast to `std::uint8_t` and should not be greater than
+     8 bits
+     @tparam Output_iterator the destination; the elements written to it are from the type `char`
+     @param in_begin the beginning of the source
+     @param in_end the ending of the source
+     @param out the destination iterator
+     @param alphabet which alphabet should be used
+     @returns the iterator to the next element past the last element copied
+     @throws see `Input_iterator` and `Output_iterator`
+    */
+    template<typename Input_iterator, typename Output_iterator>
+    static Output_iterator encode(Input_iterator in_begin, Input_iterator in_end, Output_iterator out,
+                                  alphabet alphabet = alphabet::standard)
+    {
+        constexpr auto pad = '=';
+        const char* alpha  = alphabet == alphabet::url_filename_safe
+                                ? "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_"
+                                : "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
+
+        while (in_begin != in_end) {
+            std::uint8_t i0 = 0, i1 = 0, i2 = 0;
+
+            // first character
+            i0 = static_cast<std::uint8_t>(*in_begin);
+            ++in_begin;
+
+            *out = alpha[i0 >> 2 & 0x3f];
+            ++out;
+
+            // part of first character and second
+            if (in_begin != in_end) {
+                i1 = static_cast<std::uint8_t>(*in_begin);
+                ++in_begin;
+
+                *out = alpha[((i0 & 0x3) << 4) | (i1 >> 4 & 0x0f)];
+                ++out;
+            } else {
+                *out = alpha[(i0 & 0x3) << 4];
+                ++out;
+
+                // last padding
+                *out = pad;
+                ++out;
+
+                // last padding
+                *out = pad;
+                ++out;
+
+                break;
+            }
+
+            // part of second character and third
+            if (in_begin != in_end) {
+                i2 = static_cast<std::uint8_t>(*in_begin);
+                ++in_begin;
+
+                *out = alpha[((i1 & 0xf) << 2) | (i2 >> 6 & 0x03)];
+                ++out;
+            } else {
+                *out = alpha[(i1 & 0xf) << 2];
+                ++out;
+
+                // last padding
+                *out = pad;
+                ++out;
+
+                break;
+            }
+
+            // rest of third
+            *out = alpha[i2 & 0x3f];
+            ++out;
+        }
+
+        return out;
+    }
+    /**
+     Encodes a string.
+
+     @param str the string that should be encoded
+     @param alphabet which alphabet should be used
+     @returns the encoded base64 string
+     @throws see base64::encode()
+    */
+    static std::string encode(const std::string& str, alphabet alphabet = alphabet::standard)
+    {
+        std::string result;
+
+        result.reserve(required_encode_size(str.length()) + 1);
+
+        encode(str.begin(), str.end(), std::back_inserter(result), alphabet);
+
+        return result;
+    }
+    /**
+     Encodes a char array.
+
+     @param buffer the char array
+     @param size the size of the array
+     @param alphabet which alphabet should be used
+     @returns the encoded string
+    */
+    static std::string encode(const char* buffer, std::size_t size, alphabet alphabet = alphabet::standard)
+    {
+        std::string result;
+
+        result.reserve(required_encode_size(size) + 1);
+
+        encode(buffer, buffer + size, std::back_inserter(result), alphabet);
+
+        return result;
+    }
+    /**
+     Decodes all the elements from `in_begin` to `in_end` to `out`. `in_begin` may point to the same location as `out`,
+     in other words: inplace decoding is possible.
+
+     @warning The destination must be able to hold at least `required_decode_size(std::distance(in_begin, in_end))`,
+     otherwise the behavior depends on the output iterator.
+
+     @tparam Input_iterator the source; the returned elements are cast to `char`
+     @tparam Output_iterator the destination; the elements written to it are from the type `std::uint8_t`
+     @param in_begin the beginning of the source
+     @param in_end the ending of the source
+     @param out the destination iterator
+     @param alphabet which alphabet should be used
+     @param behavior the behavior when an error was detected
+     @returns the iterator to the next element past the last element copied
+     @throws base64_error depending on the set behavior
+     @throws see `Input_iterator` and `Output_iterator`
+    */
+    template<typename Input_iterator, typename Output_iterator>
+    static Output_iterator decode(Input_iterator in_begin, Input_iterator in_end, Output_iterator out,
+                                  alphabet alphabet          = alphabet::auto_,
+                                  decoding_behavior behavior = decoding_behavior::moderate)
+    {
+        //constexpr auto pad = '=';
+        std::uint8_t last  = 0;
+        auto bits          = 0;
+
+        while (in_begin != in_end) {
+            auto c = *in_begin;
+            ++in_begin;
+
+            if (c == '=') {
+                break;
+            }
+
+            auto part = _base64_value(alphabet, c);
+
+            // enough bits for one byte
+            if (bits + 6 >= 8) {
+                *out = (last << (8 - bits)) | (part >> (bits - 2));
+                ++out;
+
+                bits -= 2;
+            } else {
+                bits += 6;
+            }
+
+            last = part;
+        }
+
+        // check padding
+        if (behavior != decoding_behavior::loose) {
+            while (in_begin != in_end) {
+                auto c = *in_begin;
+                ++in_begin;
+
+                if (c != '=') {
+                    throw base64_error("invalid base64 character.");
+                }
+            }
+        }
+
+        return out;
+    }
+    /**
+     Decodes a string.
+
+     @param str the base64 encoded string
+     @param alphabet which alphabet should be used
+     @param behavior the behavior when an error was detected
+     @returns the decoded string
+     @throws see base64::decode()
+    */
+    static std::string decode(const std::string& str, alphabet alphabet = alphabet::auto_,
+                              decoding_behavior behavior = decoding_behavior::moderate)
+    {
+        std::string result;
+
+        result.reserve(max_decode_size(str.length()));
+
+        decode(str.begin(), str.end(), std::back_inserter(result), alphabet, behavior);
+
+        return result;
+    }
+    /**
+     Decodes a string.
+
+     @param buffer the base64 encoded buffer
+     @param size the size of the buffer
+     @param alphabet which alphabet should be used
+     @param behavior the behavior when an error was detected
+     @returns the decoded string
+     @throws see base64::decode()
+    */
+    static std::string decode(const char* buffer, std::size_t size, alphabet alphabet = alphabet::auto_,
+                              decoding_behavior behavior = decoding_behavior::moderate)
+    {
+        std::string result;
+
+        result.reserve(max_decode_size(size));
+
+        decode(buffer, buffer + size, std::back_inserter(result), alphabet, behavior);
+
+        return result;
+    }
+    /**
+     Decodes a string inplace.
+
+     @param[in,out] str the base64 encoded string
+     @param alphabet which alphabet should be used
+     @param behavior the behavior when an error was detected
+     @throws base64::decode_inplace()
+    */
+    static void decode_inplace(std::string& str, alphabet alphabet = alphabet::auto_,
+                               decoding_behavior behavior = decoding_behavior::moderate)
+    {
+        str.resize(decode(str.begin(), str.end(), str.begin(), alphabet, behavior) - str.begin());
+    }
+    /**
+     Decodes a char array inplace.
+
+     @param[in,out] str the string array
+     @param size the length of the array
+     @param alphabet which alphabet should be used
+     @param behavior the behavior when an error was detected
+     @returns the pointer to the next element past the last element decoded
+     @throws base64::decode_inplace()
+    */
+    static char* decode_inplace(char* str, std::size_t size, alphabet alphabet = alphabet::auto_,
+                                decoding_behavior behavior = decoding_behavior::moderate)
+    {
+        return decode(str, str + size, str, alphabet, behavior);
+    }
+    /**
+     Returns the required decoding size for a given size. The value is calculated with the following formula:
+
+     $$
+     \lceil \frac{size}{4} \rceil \cdot 3
+     $$
+
+     @param size the size of the encoded input
+     @returns the size of the resulting decoded buffer; this the absolute maximum
+    */
+    static std::size_t max_decode_size(std::size_t size) noexcept
+    {
+        return (size / 4 + (size % 4 ? 1 : 0)) * 3;
+    }
+    /**
+     Returns the required encoding size for a given size. The value is calculated with the following formula:
+
+     $$
+     \lceil \frac{size}{3} \rceil \cdot 4
+     $$
+
+     @param size the size of the decoded input
+     @returns the size of the resulting encoded buffer
+    */
+    static std::size_t required_encode_size(std::size_t size) noexcept
+    {
+        return (size / 3 + (size % 3 ? 1 : 0)) * 4;
+    }
+
+private:
+    static std::uint8_t _base64_value(alphabet& alphabet, char c)
+    {
+        if (c >= 'A' && c <= 'Z') {
+            return c - 'A';
+        } else if (c >= 'a' && c <= 'z') {
+            return c - 'a' + 26;
+        } else if (c >= '0' && c <= '9') {
+            return c - '0' + 52;
+        }
+
+        // comes down to alphabet
+        if (alphabet == alphabet::standard) {
+            if (c == '+') {
+                return 62;
+            } else if (c == '/') {
+                return 63;
+            }
+        } else if (alphabet == alphabet::url_filename_safe) {
+            if (c == '-') {
+                return 62;
+            } else if (c == '_') {
+                return 63;
+            }
+        } // auto detect
+        else {
+            if (c == '+') {
+                alphabet = alphabet::standard;
+
+                return 62;
+            } else if (c == '/') {
+                alphabet = alphabet::standard;
+
+                return 63;
+            } else if (c == '-') {
+                alphabet = alphabet::url_filename_safe;
+
+                return 62;
+            } else if (c == '_') {
+                alphabet = alphabet::url_filename_safe;
+
+                return 63;
+            }
+        }
+
+        throw base64_error("invalid base64 character.");
+    }
+};
+
+#endif // !PUBLIC_DOMAIN_BASE64_HPP_
--- a/llama/build-info.cpp
+++ b/llama/build-info.cpp
@@ -0,0 +1,30 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+int LLAMA_BUILD_NUMBER = 0;
+char const *LLAMA_COMMIT = "";
+char const *LLAMA_COMPILER = "";
+char const *LLAMA_BUILD_TARGET = "";
--- a/llama/clip.cpp
+++ b/llama/clip.cpp
--- a/llama/clip.h
+++ b/llama/clip.h
@@ -0,0 +1,120 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef CLIP_H
+#define CLIP_H
+
+#include <stddef.h>
+#include <stdint.h>
+
+#ifdef LLAMA_SHARED
+#    if defined(_WIN32) && !defined(__MINGW32__)
+#        ifdef LLAMA_BUILD
+#            define CLIP_API __declspec(dllexport)
+#        else
+#            define CLIP_API __declspec(dllimport)
+#        endif
+#    else
+#        define CLIP_API __attribute__ ((visibility ("default")))
+#    endif
+#else
+#    define CLIP_API
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct clip_ctx;
+
+struct clip_image_size {
+    int width;
+    int height;
+};
+
+struct clip_image_u8_batch {
+    struct clip_image_u8 * data;
+    size_t size;
+};
+
+struct clip_image_f32_batch {
+    struct clip_image_f32 * data;
+    size_t size;
+};
+
+CLIP_API struct clip_ctx * clip_model_load    (const char * fname, int verbosity);
+CLIP_API struct clip_ctx * clip_model_load_cpu(const char * fname, int verbosity);
+
+CLIP_API void clip_free(struct clip_ctx * ctx);
+
+CLIP_API size_t clip_embd_nbytes(const struct clip_ctx * ctx);
+
+CLIP_API int32_t clip_image_size (const struct clip_ctx * ctx);
+CLIP_API int32_t clip_patch_size (const struct clip_ctx * ctx);
+CLIP_API int32_t clip_hidden_size(const struct clip_ctx * ctx);
+
+// TODO: should be enum, not string
+CLIP_API const char * clip_patch_merge_type(const struct clip_ctx * ctx);
+
+CLIP_API const int32_t * clip_image_grid(const struct clip_ctx * ctx);
+
+CLIP_API int clip_n_patches    (const struct clip_ctx * ctx);
+CLIP_API int clip_n_mmproj_embd(const struct clip_ctx * ctx);
+
+CLIP_API int clip_uhd_num_image_embeds_col(struct clip_ctx * ctx_clip);
+CLIP_API void clip_add_load_image_size(struct clip_ctx * ctx_clip, struct clip_image_size * load_image_size);
+
+CLIP_API struct clip_image_size * clip_image_size_init();
+CLIP_API struct clip_image_u8  * clip_image_u8_init ();
+CLIP_API struct clip_image_f32 * clip_image_f32_init();
+
+CLIP_API void clip_image_u8_free (struct clip_image_u8  * img);
+CLIP_API void clip_image_f32_free(struct clip_image_f32 * img);
+CLIP_API void clip_image_u8_batch_free (struct clip_image_u8_batch  * batch);
+CLIP_API void clip_image_f32_batch_free(struct clip_image_f32_batch * batch);
+
+CLIP_API bool clip_image_load_from_file(const char * fname, struct clip_image_u8 * img);
+
+/** interpret bytes as an image file with length bytes_length, and use the result to populate img */
+CLIP_API bool clip_image_load_from_bytes(const unsigned char * bytes, size_t bytes_length, struct clip_image_u8 * img);
+
+/** preprocess img and store the result in res_imgs, pad_to_square may be overridden to false depending on model configuration */
+CLIP_API bool clip_image_preprocess(struct clip_ctx * ctx, const struct clip_image_u8 * img, struct clip_image_f32_batch * res_imgs );
+
+CLIP_API struct ggml_tensor * clip_get_newline_tensor(const struct clip_ctx * ctx);
+
+CLIP_API bool clip_image_encode      (struct clip_ctx * ctx, int n_threads, struct clip_image_f32 * img, float * vec);
+CLIP_API bool clip_image_batch_encode(struct clip_ctx * ctx, int n_threads, const struct clip_image_f32_batch * imgs, float * vec);
+
+CLIP_API bool clip_model_quantize(const char * fname_inp, const char * fname_out, int itype);
+
+CLIP_API int clip_is_minicpmv(const struct clip_ctx * ctx);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif // CLIP_H
--- a/llama/common.cpp
+++ b/llama/common.cpp
--- a/llama/common.h
+++ b/llama/common.h
@@ -0,0 +1,514 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+// Various helper functions and utilities
+
+#pragma once
+
+#include "llama.h"
+
+#include "sampling.h"
+
+#define LOG_NO_FILE_LINE_FUNCTION
+#include "log.h"
+
+#include <cmath>
+#include <string>
+#include <vector>
+#include <random>
+#include <thread>
+#include <unordered_map>
+#include <tuple>
+
+#ifdef _WIN32
+#define DIRECTORY_SEPARATOR '\\'
+#else
+#define DIRECTORY_SEPARATOR '/'
+#endif // _WIN32
+
+#define die(msg)          do { fputs("error: " msg "\n", stderr);                exit(1); } while (0)
+#define die_fmt(fmt, ...) do { fprintf(stderr, "error: " fmt "\n", __VA_ARGS__); exit(1); } while (0)
+
+#define print_build_info() do {                                                                     \
+    fprintf(stderr, "%s: build = %d (%s)\n",      __func__, LLAMA_BUILD_NUMBER, LLAMA_COMMIT);      \
+    fprintf(stderr, "%s: built with %s for %s\n", __func__, LLAMA_COMPILER, LLAMA_BUILD_TARGET);    \
+} while(0)
+
+#define DEFAULT_MODEL_PATH "models/7B/ggml-model-f16.gguf"
+
+struct llama_lora_adapter_info {
+    std::string path;
+    float scale;
+};
+
+struct llama_lora_adapter_container : llama_lora_adapter_info {
+    struct llama_lora_adapter * adapter;
+};
+
+// build info
+extern int LLAMA_BUILD_NUMBER;
+extern char const * LLAMA_COMMIT;
+extern char const * LLAMA_COMPILER;
+extern char const * LLAMA_BUILD_TARGET;
+
+struct llama_control_vector_load_info;
+
+//
+// CPU utils
+//
+
+int32_t cpu_get_num_physical_cores();
+int32_t cpu_get_num_math();
+
+//
+// CLI argument parsing
+//
+
+// dimensionality reduction methods, used by cvector-generator
+enum dimre_method {
+    DIMRE_METHOD_PCA,
+    DIMRE_METHOD_MEAN,
+};
+
+struct cpu_params {
+    int      n_threads                   = -1;
+    bool     cpumask[GGML_MAX_N_THREADS] = {false}; // CPU affinity mask.
+    bool     mask_valid                  = false;   // Default: any CPU
+    enum ggml_sched_priority  priority   = GGML_SCHED_PRIO_NORMAL;  // Scheduling prio : (0 - normal, 1 - medium, 2 - high, 3 - realtime)
+    bool     strict_cpu                  = false;   // Use strict CPU placement
+    uint32_t poll                        = 50;      // Polling (busywait) level (0 - no polling, 100 - mostly polling)
+};
+
+struct gpt_params {
+    uint32_t seed                 = LLAMA_DEFAULT_SEED; // RNG seed
+
+    int32_t n_predict             =    -1; // new tokens to predict
+    int32_t n_ctx                 =     0; // context size
+    int32_t n_batch               =  2048; // logical batch size for prompt processing (must be >=32 to use BLAS)
+    int32_t n_ubatch              =   512; // physical batch size for prompt processing (must be >=32 to use BLAS)
+    int32_t n_keep                =     0; // number of tokens to keep from initial prompt
+    int32_t n_draft               =     5; // number of tokens to draft during speculative decoding
+    int32_t n_chunks              =    -1; // max number of chunks to process (-1 = unlimited)
+    int32_t n_parallel            =     1; // number of parallel sequences to decode
+    int32_t n_sequences           =     1; // number of sequences to decode
+    float   p_split               =  0.1f; // speculative decoding split probability
+    int32_t n_gpu_layers          =    -1; // number of layers to store in VRAM (-1 - use default)
+    int32_t n_gpu_layers_draft    =    -1; // number of layers to store in VRAM for the draft model (-1 - use default)
+    int32_t main_gpu              =     0; // the GPU that is used for scratch and small tensors
+    float   tensor_split[128]     =   {0}; // how split tensors should be distributed across GPUs
+    int32_t grp_attn_n            =     1; // group-attention factor
+    int32_t grp_attn_w            =   512; // group-attention width
+    int32_t n_print               =    -1; // print token count every n tokens (-1 = disabled)
+    float   rope_freq_base        =  0.0f; // RoPE base frequency
+    float   rope_freq_scale       =  0.0f; // RoPE frequency scaling factor
+    float   yarn_ext_factor       = -1.0f; // YaRN extrapolation mix factor
+    float   yarn_attn_factor      =  1.0f; // YaRN magnitude scaling factor
+    float   yarn_beta_fast        = 32.0f; // YaRN low correction dim
+    float   yarn_beta_slow        =  1.0f; // YaRN high correction dim
+    int32_t yarn_orig_ctx         =     0; // YaRN original context length
+    float   defrag_thold          = -1.0f; // KV cache defragmentation threshold
+
+    struct cpu_params cpuparams;
+    struct cpu_params cpuparams_batch;
+    struct cpu_params draft_cpuparams;
+    struct cpu_params draft_cpuparams_batch;
+
+    ggml_backend_sched_eval_callback cb_eval = nullptr;
+    void * cb_eval_user_data                 = nullptr;
+
+    ggml_numa_strategy numa = GGML_NUMA_STRATEGY_DISABLED;
+
+    enum llama_split_mode        split_mode        = LLAMA_SPLIT_MODE_LAYER; // how to split the model across GPUs
+    enum llama_rope_scaling_type rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED;
+    enum llama_pooling_type      pooling_type      = LLAMA_POOLING_TYPE_UNSPECIFIED; // pooling type for embeddings
+    enum llama_attention_type    attention_type    = LLAMA_ATTENTION_TYPE_UNSPECIFIED; // attention type for embeddings
+
+    // // sampling parameters
+    struct llama_sampling_params sparams;
+
+    std::string model                = ""; // model path
+    std::string model_draft          = ""; // draft model for speculative decoding
+    std::string model_alias          = "unknown"; // model alias
+    std::string model_url            = ""; // model url to download
+    std::string hf_token             = ""; // HF token
+    std::string hf_repo              = ""; // HF repo
+    std::string hf_file              = ""; // HF file
+    std::string prompt               = "";
+    std::string prompt_file          = ""; // store the external prompt file name
+    std::string path_prompt_cache    = ""; // path to file for saving/loading prompt eval state
+    std::string input_prefix         = ""; // string to prefix user inputs with
+    std::string input_suffix         = ""; // string to suffix user inputs with
+    std::string logdir               = ""; // directory in which to save YAML log files
+    std::string lookup_cache_static  = ""; // path of static ngram cache file for lookup decoding
+    std::string lookup_cache_dynamic = ""; // path of dynamic ngram cache file for lookup decoding
+    std::string logits_file          = ""; // file for saving *all* logits
+    std::string rpc_servers          = ""; // comma separated list of RPC servers
+
+    std::vector<std::string> in_files;   // all input files
+    std::vector<std::string> antiprompt; // strings upon which more user input is prompted (a.k.a. reverse prompts)
+    std::vector<llama_model_kv_override> kv_overrides;
+
+    bool lora_init_without_apply = false; // only load lora to memory, but do not apply it to ctx (user can manually apply lora later using llama_lora_adapter_apply)
+    std::vector<llama_lora_adapter_info> lora_adapters; // lora adapter path with user defined scale
+
+    std::vector<llama_control_vector_load_info> control_vectors; // control vector with user defined scale
+
+    int32_t verbosity                  = 0;
+    int32_t control_vector_layer_start = -1; // layer range for control vector
+    int32_t control_vector_layer_end   = -1; // layer range for control vector
+
+    int32_t ppl_stride      = 0;     // stride for perplexity calculations. If left at 0, the pre-existing approach will be used.
+    int32_t ppl_output_type = 0;     // = 0 -> ppl output is as usual, = 1 -> ppl output is num_tokens, ppl, one per line
+                                     //                                       (which is more convenient to use for plotting)
+                                     //
+    bool   hellaswag        = false; // compute HellaSwag score over random tasks from datafile supplied in prompt
+    size_t hellaswag_tasks  = 400;   // number of tasks to use when computing the HellaSwag score
+
+    bool   winogrande       = false; // compute Winogrande score over random tasks from datafile supplied in prompt
+    size_t winogrande_tasks = 0;     // number of tasks to use when computing the Winogrande score. If 0, all tasks will be computed
+
+    bool   multiple_choice  = false;  // compute TruthfulQA score over random tasks from datafile supplied in prompt
+    size_t multiple_choice_tasks = 0; // number of tasks to use when computing the TruthfulQA score. If 0, all tasks will be computed
+
+    bool   kl_divergence    = false; // compute KL divergence
+
+    bool usage             = false; // print usage
+    bool use_color         = false; // use color to distinguish generations and inputs
+    bool special           = false; // enable special token output
+    bool interactive       = false; // interactive mode
+    bool interactive_first = false; // wait for user input immediately
+    bool conversation      = false; // conversation mode (does not print special tokens and suffix/prefix)
+    bool prompt_cache_all  = false; // save user input and generations to prompt cache
+    bool prompt_cache_ro   = false; // open the prompt cache read-only and do not update it
+
+    bool escape            = true;  // escape "\n", "\r", "\t", "\'", "\"", and "\\"
+    bool multiline_input   = false; // reverse the usage of `\`
+    bool simple_io         = false; // improves compatibility with subprocesses and limited consoles
+    bool cont_batching     = true;  // insert new sequences for decoding on-the-fly
+    bool flash_attn        = false; // flash attention
+
+    bool input_prefix_bos  = false; // prefix BOS to user inputs, preceding input_prefix
+    bool ignore_eos        = false; // ignore generated EOS tokens
+    bool logits_all        = false; // return logits for all tokens in the batch
+    bool use_mmap          = true;  // use mmap for faster loads
+    bool use_mlock         = false; // use mlock to keep model in memory
+    bool verbose_prompt    = false; // print prompt tokens before generation
+    bool display_prompt    = true;  // print prompt before generation
+    bool infill            = false; // use infill mode
+    bool dump_kv_cache     = false; // dump the KV cache contents for debugging purposes
+    bool no_kv_offload     = false; // disable KV offloading
+    bool warmup            = true;  // warmup run
+    bool check_tensors     = false; // validate tensor data
+
+    std::string cache_type_k = "f16"; // KV cache data type for the K
+    std::string cache_type_v = "f16"; // KV cache data type for the V
+
+    // multimodal models (see examples/llava)
+    std::string mmproj = "";        // path to multimodal projector
+    std::vector<std::string> image; // path to image file(s)
+
+    // embedding
+    bool embedding         = false; // get only sentence embedding
+    int32_t embd_normalize = 2;     // normalisation for embendings (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)
+    std::string embd_out   = "";    // empty = default, "array" = [[],[]...], "json" = openai style, "json+" = same "json" + cosine similarity matrix
+    std::string embd_sep   = "\n";  // separator of embendings
+
+    // server params
+    int32_t port           = 8080;         // server listens on this network port
+    int32_t timeout_read   = 600;          // http read timeout in seconds
+    int32_t timeout_write  = timeout_read; // http write timeout in seconds
+    int     n_threads_http = -1;           // number of threads to process HTTP requests (TODO: support threadpool)
+
+    std::string hostname      = "127.0.0.1";
+    std::string public_path   = "";
+    std::string chat_template = "";
+    std::string system_prompt = "";
+    bool enable_chat_template = true;
+
+    std::vector<std::string> api_keys;
+
+    std::string ssl_file_key  = "";
+    std::string ssl_file_cert = "";
+
+    bool endpoint_slots   = true;
+    bool endpoint_metrics = false;
+
+    bool log_json = false;
+
+    std::string slot_save_path;
+
+    float slot_prompt_similarity = 0.5f;
+
+    // batched-bench params
+    bool is_pp_shared = false;
+
+    std::vector<int32_t> n_pp;
+    std::vector<int32_t> n_tg;
+    std::vector<int32_t> n_pl;
+
+    // retrieval params
+    std::vector<std::string> context_files; // context files to embed
+
+    int32_t chunk_size = 64; // chunk size for context embedding
+
+    std::string chunk_separator = "\n"; // chunk separator for context embedding
+
+    // passkey params
+    int32_t n_junk = 250; // number of times to repeat the junk text
+    int32_t i_pos  = -1;  // position of the passkey in the junk text
+
+    // imatrix params
+    std::string out_file = "imatrix.dat"; // save the resulting imatrix to this file
+
+    int32_t n_out_freq  = 10; // output the imatrix every n_out_freq iterations
+    int32_t n_save_freq =  0; // save the imatrix every n_save_freq iterations
+    int32_t i_chunk     =  0; // start processing from this chunk
+
+    bool process_output = false; // collect data for the output tensor
+    bool compute_ppl    = true;  // whether to compute perplexity
+
+    // cvector-generator params
+    int n_pca_batch = 100;
+    int n_pca_iterations = 1000;
+    dimre_method cvector_dimre_method = DIMRE_METHOD_PCA;
+    std::string cvector_outfile       = "control_vector.gguf";
+    std::string cvector_positive_file = "examples/cvector-generator/positive.txt";
+    std::string cvector_negative_file = "examples/cvector-generator/negative.txt";
+
+    bool spm_infill = false; // suffix/prefix/middle pattern for infill
+
+    std::string lora_outfile = "ggml-lora-merged-f16.gguf";
+};
+
+void gpt_params_parse_from_env(gpt_params & params);
+void gpt_params_handle_model_default(gpt_params & params);
+
+bool gpt_params_parse_ex   (int argc, char ** argv, gpt_params & params);
+bool gpt_params_parse      (int argc, char ** argv, gpt_params & params);
+bool gpt_params_find_arg   (int argc, char ** argv, const std::string & arg, gpt_params & params, int & i, bool & invalid_param);
+void gpt_params_print_usage(int argc, char ** argv, const gpt_params & params);
+
+std::string gpt_params_get_system_info(const gpt_params & params);
+
+bool parse_cpu_range(const std::string& range, bool(&boolmask)[GGML_MAX_N_THREADS]);
+bool parse_cpu_mask(const std::string& mask, bool(&boolmask)[GGML_MAX_N_THREADS]);
+void postprocess_cpu_params(cpu_params& cpuparams, const cpu_params* role_model = nullptr);
+bool set_process_priority(enum ggml_sched_priority prio);
+
+//
+// String utils
+//
+
+std::vector<std::string> string_split(std::string input, char separator);
+
+std::string string_strip(const std::string & str);
+std::string string_get_sortable_timestamp();
+
+void string_replace_all(std::string & s, const std::string & search, const std::string & replace);
+
+template<class T>
+static std::vector<T> string_split(const std::string & str, char delim) {
+    std::vector<T> values;
+    std::istringstream str_stream(str);
+    std::string token;
+    while (std::getline(str_stream, token, delim)) {
+        T value;
+        std::istringstream token_stream(token);
+        token_stream >> value;
+        values.push_back(value);
+    }
+    return values;
+}
+
+bool string_parse_kv_override(const char * data, std::vector<llama_model_kv_override> & overrides);
+void string_process_escapes(std::string & input);
+
+//
+// Filesystem utils
+//
+
+bool fs_validate_filename(const std::string & filename);
+bool fs_create_directory_with_parents(const std::string & path);
+
+std::string fs_get_cache_directory();
+std::string fs_get_cache_file(const std::string & filename);
+
+//
+// Model utils
+//
+
+struct llama_init_result {
+    struct llama_model   * model   = nullptr;
+    struct llama_context * context = nullptr;
+    std::vector<llama_lora_adapter_container> lora_adapters;
+};
+
+struct llama_init_result    llama_init_from_gpt_params(gpt_params & params);
+
+struct llama_model_params     llama_model_params_from_gpt_params    (const gpt_params & params);
+struct llama_context_params   llama_context_params_from_gpt_params  (const gpt_params & params);
+struct ggml_threadpool_params ggml_threadpool_params_from_cpu_params(const cpu_params & params);
+
+struct llama_model * llama_load_model_from_url(const char * model_url, const char * path_model, const char * hf_token, const struct llama_model_params & params);
+struct llama_model * llama_load_model_from_hf(const char * repo, const char * file, const char * path_model, const char * hf_token, const struct llama_model_params & params);
+
+// clear LoRA adapters from context, then apply new list of adapters
+void llama_lora_adapters_apply(struct llama_context * ctx, std::vector<llama_lora_adapter_container> & lora_adapters);
+
+// Batch utils
+
+void llama_batch_clear(struct llama_batch & batch);
+
+void llama_batch_add(
+                 struct llama_batch & batch,
+                        llama_token   id,
+                          llama_pos   pos,
+    const std::vector<llama_seq_id> & seq_ids,
+                               bool   logits);
+
+//
+// Vocab utils
+//
+
+// tokenizes a string into a vector of tokens
+// should work similar to Python's `tokenizer.encode`
+std::vector<llama_token> llama_tokenize(
+  const struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_special,
+                        bool   parse_special = false);
+
+std::vector<llama_token> llama_tokenize(
+    const struct llama_model * model,
+           const std::string & text,
+                        bool   add_special,
+                        bool   parse_special = false);
+
+// tokenizes a token into a piece, optionally renders special/control tokens
+// should work similar to Python's `tokenizer.id_to_piece`
+std::string llama_token_to_piece(
+        const struct llama_context * ctx,
+                       llama_token   token,
+                       bool          special = true);
+
+// detokenizes a vector of tokens into a string
+// should work similar to Python's `tokenizer.decode`
+// optionally renders special/control tokens
+std::string llama_detokenize(
+                         llama_context * ctx,
+        const std::vector<llama_token> & tokens,
+                                  bool   special = true);
+
+//
+// Chat template utils
+//
+
+// same with llama_chat_message, but uses std::string
+struct llama_chat_msg {
+    std::string role;
+    std::string content;
+};
+
+// Check if the template supplied via "--chat-template" is supported or not. Returns true if it's valid
+bool llama_chat_verify_template(const std::string & tmpl);
+
+// CPP wrapper for llama_chat_apply_template
+// If the built-in template is not supported, we default to chatml
+// If the custom "tmpl" is not supported, we throw an error
+std::string llama_chat_apply_template(const struct llama_model * model,
+        const std::string & tmpl,
+        const std::vector<llama_chat_msg> & chat,
+        bool add_ass);
+
+// Format single message, while taking into account the position of that message in chat history
+std::string llama_chat_format_single(const struct llama_model * model,
+        const std::string & tmpl,
+        const std::vector<llama_chat_msg> & past_msg,
+        const llama_chat_msg & new_msg,
+        bool add_ass);
+
+// Returns an example of formatted chat
+std::string llama_chat_format_example(const struct llama_model * model,
+        const std::string & tmpl);
+
+//
+// KV cache utils
+//
+
+// Dump the KV cache view with the number of sequences per cell.
+void llama_kv_cache_dump_view(const llama_kv_cache_view & view, int row_size = 80);
+
+// Dump the KV cache view showing individual sequences in each cell (long output).
+void llama_kv_cache_dump_view_seqs(const llama_kv_cache_view & view, int row_size = 40);
+
+//
+// Embedding utils
+//
+
+void llama_embd_normalize(const float * inp, float * out, int n, int embd_norm = 2);
+
+float llama_embd_similarity_cos(const float * embd1, const float * embd2, int n);
+
+//
+// Control vector utils
+//
+
+struct llama_control_vector_data {
+    int n_embd;
+
+    // stores data for layers [1, n_layer] where n_layer = data.size() / n_embd
+    std::vector<float> data;
+};
+
+struct llama_control_vector_load_info {
+    float strength;
+
+    std::string fname;
+};
+
+// Load control vectors, scale each by strength, and add them together.
+// On error, returns {-1, empty}
+llama_control_vector_data llama_control_vector_load(const std::vector<llama_control_vector_load_info> & load_infos);
+
+//
+// Split utils
+//
+
+static const char * const LLM_KV_SPLIT_NO            = "split.no";
+static const char * const LLM_KV_SPLIT_COUNT         = "split.count";
+static const char * const LLM_KV_SPLIT_TENSORS_COUNT = "split.tensors.count";
+
+//
+// YAML utils
+//
+
+void yaml_dump_vector_float    (FILE * stream, const char * prop_name, const std::vector<float> & data);
+void yaml_dump_vector_int      (FILE * stream, const char * prop_name, const std::vector<int> & data);
+void yaml_dump_string_multiline(FILE * stream, const char * prop_name, const char * data);
+
+void yaml_dump_non_result_info(
+    FILE * stream, const gpt_params & params, const llama_context * lctx,
+    const std::string & timestamp, const std::vector<int> & prompt_tokens, const char * model_desc);
--- a/llama/example/README.md
+++ b/llama/example/README.md
@@ -0,0 +1,31 @@
+# `example`
+
+Demo app for the `llama` package
+
+Pull a model:
+
+```
+ollama pull mistral:7b-instruct-v0.3-q4_0
+```
+
+Then run it:
+
+```
+go run -x . \
+    -model ~/.ollama/models/blobs/sha256-ff82381e2bea77d91c1b824c7afb83f6fb73e9f7de9dda631bcdbca564aa5435 \
+    -prompt "[ISNT] Why is the sky blue? [/INST]"
+```
+
+## Vision
+
+```
+ollama pull llava:7b-v1.6-mistral-q4_0
+```
+
+```
+go run -x . \
+    -model ~/.ollama/models/blobs/sha256-170370233dd5c5415250a2ecd5c71586352850729062ccef1496385647293868 \
+    -projector ~/.ollama/models/blobs/sha256-72d6f08a42f656d36b356dbe0920675899a99ce21192fd66266fb7d82ed07539 \
+    -image ./alonso.jpg \
+    -prompt "[ISNT] What is in this image? <image> [/INST]"
+```
--- a/llama/example/alonso.jpg
+++ b/llama/example/alonso.jpg
--- a/llama/example/main.go
+++ b/llama/example/main.go
@@ -0,0 +1,136 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"io"
+	"log"
+	"os"
+	"runtime"
+	"strings"
+
+	"github.com/ollama/ollama/llama"
+)
+
+func main() {
+	mpath := flag.String("model", "", "Path to model binary file")
+	ppath := flag.String("projector", "", "Path to projector binary file")
+	image := flag.String("image", "", "Path to image file")
+	prompt := flag.String("prompt", "", "Prompt including <image> tag")
+	flag.Parse()
+
+	if *mpath == "" {
+		panic("model path is required")
+	}
+
+	if *prompt == "" {
+		panic("prompt is required")
+	}
+
+	// load the model
+	llama.BackendInit()
+	params := llama.ModelParams{
+		NumGpuLayers: 999,
+		MainGpu:      0,
+		UseMmap:      true,
+		Progress: func(p float32) {
+			fmt.Printf("loading... %f\n", p)
+		},
+	}
+	model := llama.LoadModelFromFile(*mpath, params)
+	ctxParams := llama.NewContextParams(2048, runtime.NumCPU(), false)
+
+	// language model context
+	lc := llama.NewContextWithModel(model, ctxParams)
+
+	// eval before
+	batch := llama.NewBatch(512, 0, 1)
+	var nPast int
+
+	// clip context
+	var clipCtx *llama.ClipContext
+
+	// multi-modal
+	if *ppath != "" {
+		clipCtx = llama.NewClipContext(*ppath)
+
+		// open image file
+		file, err := os.Open(*image)
+		if err != nil {
+			panic(err)
+		}
+		defer file.Close()
+
+		data, err := io.ReadAll(file)
+		if err != nil {
+			log.Fatal(err)
+		}
+
+		embedding := llama.NewLlavaImageEmbed(clipCtx, data)
+
+		parts := strings.Split(*prompt, "<image>")
+		if len(parts) != 2 {
+			panic("prompt must contain exactly one <image>")
+		}
+
+		beforeTokens, err := lc.Model().Tokenize(parts[0], true, true)
+		if err != nil {
+			panic(err)
+		}
+
+		for _, t := range beforeTokens {
+			batch.Add(t, nPast, []int{0}, true)
+			nPast++
+		}
+
+		err = lc.Decode(batch)
+		if err != nil {
+			panic(err)
+		}
+
+		llama.LlavaEvalImageEmbed(lc, embedding, 512, &nPast)
+
+		afterTokens, err := lc.Model().Tokenize(parts[1], true, true)
+		if err != nil {
+			panic(err)
+		}
+
+		for _, t := range afterTokens {
+			batch.Add(t, nPast, []int{0}, true)
+			nPast++
+		}
+	} else {
+		tokens, err := lc.Model().Tokenize(*prompt, true, true)
+		if err != nil {
+			panic(err)
+		}
+
+		for _, t := range tokens {
+			batch.Add(t, nPast, []int{0}, true)
+			nPast++
+		}
+	}
+
+	// main loop
+	for n := nPast; n < 4096; n++ {
+		err := lc.Decode(batch)
+		if err != nil {
+			panic(err)
+		}
+
+		// sample a token
+		logits := lc.GetLogitsIth(batch.NumTokens() - 1)
+		token := lc.SampleTokenGreedy(logits)
+
+		// if it's an end of sequence token, break
+		if lc.Model().TokenIsEog(token) {
+			break
+		}
+
+		// print the token
+		str := lc.Model().TokenToPiece(token)
+		fmt.Print(str)
+		batch.Clear()
+		batch.Add(token, n, []int{0}, true)
+	}
+}
--- a/llama/ggml-aarch64.c
+++ b/llama/ggml-aarch64.c
--- a/llama/ggml-aarch64.h
+++ b/llama/ggml-aarch64.h
@@ -0,0 +1,65 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+// SPDX-FileCopyrightText: Copyright 2024 Arm Ltd.
+#pragma once
+
+#define GGML_COMMON_DECL_C
+#include "ggml-common.h"
+
+#include "ggml.h"
+
+// GGML internal header
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Quantization
+void quantize_q8_0_4x4(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
+void quantize_q8_0_4x8(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k);
+
+void quantize_mat_q8_0(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t nrows, int64_t n_per_row, int64_t blck_size_interleave);
+
+// Quantization utilizing an importance matrix (a.k.a. "Activation aWare Quantization")
+size_t quantize_q4_0_4x4(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
+size_t quantize_q4_0_4x8(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
+size_t quantize_q4_0_8x8(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix);
+
+// GEMV
+void ggml_gemv_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);
+void ggml_gemv_q4_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_gemv_q4_0_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);
+
+// GEMM
+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);
+void ggml_gemm_q4_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_q4_0_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);
+
+#ifdef __cplusplus
+}
+#endif
+
--- a/llama/ggml-alloc.c
+++ b/llama/ggml-alloc.c
--- a/llama/ggml-alloc.h
+++ b/llama/ggml-alloc.h
@@ -0,0 +1,102 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+typedef struct ggml_backend_buffer_type * ggml_backend_buffer_type_t;
+typedef struct      ggml_backend_buffer * ggml_backend_buffer_t;
+typedef struct             ggml_backend * ggml_backend_t;
+
+// Tensor allocator
+struct ggml_tallocr {
+    ggml_backend_buffer_t buffer;
+    void * base;
+    size_t alignment;
+    size_t offset;
+};
+
+GGML_API struct ggml_tallocr ggml_tallocr_new(ggml_backend_buffer_t buffer);
+GGML_API void                ggml_tallocr_alloc(struct ggml_tallocr * talloc, struct ggml_tensor * tensor);
+
+// Graph allocator
+/*
+  Example usage:
+    ggml_gallocr_t galloc = ggml_gallocr_new(ggml_bacckend_cpu_buffer_type());
+
+    // optional: create a worst-case graph and reserve the buffers to avoid reallocations
+    ggml_gallocr_reserve(galloc, build_graph(max_batch));
+
+    // allocate the graph
+    struct ggml_cgraph * graph = build_graph(batch);
+    ggml_gallocr_alloc_graph(galloc, graph);
+
+    printf("compute buffer size: %zu bytes\n", ggml_gallocr_get_buffer_size(galloc, 0));
+
+    // evaluate the graph
+    ggml_backend_graph_compute(backend, graph);
+*/
+
+// special tensor flags for use with the graph allocator:
+//   ggml_set_input(): all input tensors are allocated at the beginning of the graph in non-overlapping addresses
+//   ggml_set_output(): output tensors are never freed and never overwritten
+
+typedef struct ggml_gallocr * ggml_gallocr_t;
+
+GGML_API ggml_gallocr_t ggml_gallocr_new(ggml_backend_buffer_type_t buft);
+GGML_API ggml_gallocr_t ggml_gallocr_new_n(ggml_backend_buffer_type_t * bufts, int n_bufs);
+GGML_API void           ggml_gallocr_free(ggml_gallocr_t galloc);
+
+// pre-allocate buffers from a measure graph - does not allocate or modify the graph
+// call with a worst-case graph to avoid buffer reallocations
+// not strictly required for single buffer usage: ggml_gallocr_alloc_graph will reallocate the buffers automatically if needed
+// returns false if the buffer allocation failed
+GGML_API bool ggml_gallocr_reserve(ggml_gallocr_t galloc, struct ggml_cgraph * graph);
+GGML_API bool ggml_gallocr_reserve_n(
+    ggml_gallocr_t galloc,
+    struct ggml_cgraph * graph,
+    const int * node_buffer_ids,
+    const int * leaf_buffer_ids);
+
+// automatic reallocation if the topology changes when using a single buffer
+// returns false if using multiple buffers and a re-allocation is needed (call ggml_gallocr_reserve_n first to set the node buffers)
+GGML_API bool ggml_gallocr_alloc_graph(ggml_gallocr_t galloc, struct ggml_cgraph * graph);
+
+GGML_API size_t ggml_gallocr_get_buffer_size(ggml_gallocr_t galloc, int buffer_id);
+
+// Utils
+// Create a buffer and allocate all the tensors in a ggml_context
+GGML_API struct ggml_backend_buffer * ggml_backend_alloc_ctx_tensors_from_buft(struct ggml_context * ctx, ggml_backend_buffer_type_t buft);
+GGML_API struct ggml_backend_buffer * ggml_backend_alloc_ctx_tensors(struct ggml_context * ctx, ggml_backend_t backend);
+
+#ifdef  __cplusplus
+}
+#endif
--- a/llama/ggml-backend-impl.h
+++ b/llama/ggml-backend-impl.h
@@ -0,0 +1,179 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+// ggml-backend internal header
+
+#include "ggml-backend.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+    //
+    // Backend buffer
+    //
+
+    // buffer type
+    typedef void * ggml_backend_buffer_type_context_t;
+
+    struct ggml_backend_buffer_type_i {
+        const char *          (*GGML_CALL get_name)        (ggml_backend_buffer_type_t buft);
+        // allocate a buffer of this type
+        ggml_backend_buffer_t (*GGML_CALL alloc_buffer)    (ggml_backend_buffer_type_t buft, size_t size);
+        // tensor alignment
+        size_t                (*GGML_CALL get_alignment)   (ggml_backend_buffer_type_t buft);
+        // max buffer size that can be allocated
+        size_t                (*GGML_CALL get_max_size)    (ggml_backend_buffer_type_t buft);
+        // data size needed to allocate the tensor, including padding
+        size_t                (*GGML_CALL get_alloc_size)  (ggml_backend_buffer_type_t buft, const struct ggml_tensor * tensor);
+        // check if tensor data is in host memory
+        bool                  (*GGML_CALL is_host)         (ggml_backend_buffer_type_t buft);
+    };
+
+    struct ggml_backend_buffer_type {
+        struct ggml_backend_buffer_type_i  iface;
+        ggml_backend_buffer_type_context_t context;
+    };
+
+    // buffer
+    typedef void * ggml_backend_buffer_context_t;
+
+    struct ggml_backend_buffer_i {
+        const char * (*GGML_CALL get_name)   (ggml_backend_buffer_t buffer);
+        void         (*GGML_CALL free_buffer)(ggml_backend_buffer_t buffer);
+        void *       (*GGML_CALL get_base)   (ggml_backend_buffer_t buffer);
+        void         (*GGML_CALL init_tensor)(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor);
+        void         (*GGML_CALL set_tensor) (ggml_backend_buffer_t buffer,       struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+        void         (*GGML_CALL get_tensor) (ggml_backend_buffer_t buffer, const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+        bool         (*GGML_CALL cpy_tensor) (ggml_backend_buffer_t buffer, const struct ggml_tensor * src, struct ggml_tensor * dst); // dst is in the buffer, src may be in any buffer
+        void         (*GGML_CALL clear)      (ggml_backend_buffer_t buffer, uint8_t value);
+        void         (*GGML_CALL reset)      (ggml_backend_buffer_t buffer); // reset any internal state due to tensor initialization, such as tensor extras
+    };
+
+    struct ggml_backend_buffer {
+        struct ggml_backend_buffer_i  iface;
+        ggml_backend_buffer_type_t    buft;
+        ggml_backend_buffer_context_t context;
+        size_t size;
+        enum ggml_backend_buffer_usage usage;
+    };
+
+    GGML_CALL ggml_backend_buffer_t ggml_backend_buffer_init(
+                   ggml_backend_buffer_type_t      buft,
+            struct ggml_backend_buffer_i           iface,
+                   ggml_backend_buffer_context_t   context,
+                   size_t                          size);
+
+    // do not use directly, use ggml_backend_tensor_copy instead
+    bool ggml_backend_buffer_copy_tensor(const struct ggml_tensor * src, struct ggml_tensor * dst);
+
+    // buffer that contains a collection of buffers
+    GGML_CALL ggml_backend_buffer_t ggml_backend_multi_buffer_alloc_buffer(ggml_backend_buffer_t * buffers, size_t n_buffers);
+    GGML_CALL bool                  ggml_backend_buffer_is_multi_buffer(ggml_backend_buffer_t buffer);
+    GGML_CALL void                  ggml_backend_multi_buffer_set_usage(ggml_backend_buffer_t buffer, enum ggml_backend_buffer_usage usage);
+
+    //
+    // Backend
+    //
+
+    typedef void * ggml_backend_context_t;
+
+    struct ggml_backend_i {
+        const char * (*GGML_CALL get_name)(ggml_backend_t backend);
+
+        void (*GGML_CALL free)(ggml_backend_t backend);
+
+        // buffer allocation
+        ggml_backend_buffer_type_t (*GGML_CALL get_default_buffer_type)(ggml_backend_t backend);
+
+        // (optional) asynchronous tensor data access
+        void (*GGML_CALL set_tensor_async)(ggml_backend_t backend,       struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+        void (*GGML_CALL get_tensor_async)(ggml_backend_t backend, const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+        bool (*GGML_CALL cpy_tensor_async)(ggml_backend_t backend_src, ggml_backend_t backend_dst, const struct ggml_tensor * src, struct ggml_tensor * dst);
+
+        // (optional) complete all pending operations
+        void (*GGML_CALL synchronize)(ggml_backend_t backend);
+
+        // compute graph with a plan (not used currently)
+        // create a new plan for a graph
+        ggml_backend_graph_plan_t (*GGML_CALL graph_plan_create) (ggml_backend_t backend, const struct ggml_cgraph * cgraph);
+        void                      (*GGML_CALL graph_plan_free)   (ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+        // update the plan with a new graph - this should be faster than creating a new plan when the graph has the same topology
+        void                      (*GGML_CALL graph_plan_update) (ggml_backend_t backend, ggml_backend_graph_plan_t plan, const struct ggml_cgraph * cgraph);
+        // compute the graph with the plan
+        enum ggml_status          (*GGML_CALL graph_plan_compute)(ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+
+        // compute graph without a plan (async)
+        enum ggml_status (*GGML_CALL graph_compute)     (ggml_backend_t backend, struct ggml_cgraph * cgraph);
+
+        // check if the backend can compute an operation
+        bool (*GGML_CALL supports_op)(ggml_backend_t backend, const struct ggml_tensor * op);
+
+        // check if the backend can use tensors allocated in a buffer type
+        bool (*GGML_CALL supports_buft)(ggml_backend_t backend, ggml_backend_buffer_type_t buft);
+
+        // check if the backend wants to run an operation, even if the weights are allocated in a CPU buffer
+        // these should be expensive operations with large batch sizes that may benefit from running on this backend
+        // even if the weight has to be copied from the CPU temporarily
+        bool (*GGML_CALL offload_op)(ggml_backend_t backend, const struct ggml_tensor * op);
+
+        // (optional) event synchronization
+        // create a new event that can record events on this backend instance
+        ggml_backend_event_t (*GGML_CALL event_new)         (ggml_backend_t backend);
+        void                 (*GGML_CALL event_free)        (ggml_backend_event_t event);
+        // record an event on the backend instance that created it
+        void                 (*GGML_CALL event_record)      (ggml_backend_event_t event);
+        // wait for an event on on a different backend instance
+        void                 (*GGML_CALL event_wait)        (ggml_backend_t backend, ggml_backend_event_t event);
+        // block until an event is recorded
+        void                 (*GGML_CALL event_synchronize) (ggml_backend_event_t event);
+    };
+
+    struct ggml_backend {
+        ggml_guid_t guid;
+
+        struct ggml_backend_i iface;
+        ggml_backend_context_t context;
+    };
+
+    struct ggml_backend_event {
+        ggml_backend_t backend;
+        void * context;
+    };
+
+    //
+    // Backend registry
+    //
+
+    typedef ggml_backend_t (*GGML_CALL ggml_backend_init_fn)(const char * params, void * user_data);
+
+    GGML_CALL void ggml_backend_register(const char * name, ggml_backend_init_fn init_fn, ggml_backend_buffer_type_t default_buffer_type, void * user_data);
+
+#ifdef  __cplusplus
+}
+#endif
--- a/llama/ggml-backend.c
+++ b/llama/ggml-backend.c
--- a/llama/ggml-backend.h
+++ b/llama/ggml-backend.h
@@ -0,0 +1,266 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+#include "ggml-alloc.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+    typedef struct ggml_backend_buffer_type * ggml_backend_buffer_type_t;
+    typedef struct ggml_backend_buffer * ggml_backend_buffer_t;
+    typedef struct ggml_backend_event * ggml_backend_event_t;
+    typedef struct ggml_backend * ggml_backend_t;
+    typedef void * ggml_backend_graph_plan_t;
+
+    //
+    // Backend buffer
+    //
+
+    // buffer type
+    GGML_API           const char *          ggml_backend_buft_name            (ggml_backend_buffer_type_t buft);
+    GGML_API GGML_CALL ggml_backend_buffer_t ggml_backend_buft_alloc_buffer    (ggml_backend_buffer_type_t buft, size_t size);
+    GGML_API           size_t                ggml_backend_buft_get_alignment   (ggml_backend_buffer_type_t buft);
+    GGML_API           size_t                ggml_backend_buft_get_max_size    (ggml_backend_buffer_type_t buft);
+    GGML_API GGML_CALL size_t                ggml_backend_buft_get_alloc_size  (ggml_backend_buffer_type_t buft, struct ggml_tensor * tensor);
+    GGML_API           bool                  ggml_backend_buft_is_host         (ggml_backend_buffer_type_t buft);
+
+    // buffer
+    enum ggml_backend_buffer_usage {
+        GGML_BACKEND_BUFFER_USAGE_ANY = 0,
+        GGML_BACKEND_BUFFER_USAGE_WEIGHTS = 1,
+        GGML_BACKEND_BUFFER_USAGE_COMPUTE = 2,
+    };
+
+    GGML_API           const char *                   ggml_backend_buffer_name          (ggml_backend_buffer_t buffer);
+    GGML_API           void                           ggml_backend_buffer_free          (ggml_backend_buffer_t buffer);
+    GGML_API           void *                         ggml_backend_buffer_get_base      (ggml_backend_buffer_t buffer);
+    GGML_API           size_t                         ggml_backend_buffer_get_size      (ggml_backend_buffer_t buffer);
+    GGML_API GGML_CALL void                           ggml_backend_buffer_init_tensor   (ggml_backend_buffer_t buffer, struct ggml_tensor * tensor);
+    GGML_API           size_t                         ggml_backend_buffer_get_alignment (ggml_backend_buffer_t buffer);
+    GGML_API           size_t                         ggml_backend_buffer_get_max_size  (ggml_backend_buffer_t buffer);
+    GGML_API           size_t                         ggml_backend_buffer_get_alloc_size(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor);
+    GGML_API           void                           ggml_backend_buffer_clear         (ggml_backend_buffer_t buffer, uint8_t value);
+    GGML_API           bool                           ggml_backend_buffer_is_host       (ggml_backend_buffer_t buffer);
+    GGML_API           void                           ggml_backend_buffer_set_usage     (ggml_backend_buffer_t buffer, enum ggml_backend_buffer_usage usage);
+    GGML_API           enum ggml_backend_buffer_usage ggml_backend_buffer_get_usage     (ggml_backend_buffer_t buffer);
+    GGML_API           ggml_backend_buffer_type_t     ggml_backend_buffer_get_type      (ggml_backend_buffer_t buffer);
+    GGML_API           void                           ggml_backend_buffer_reset         (ggml_backend_buffer_t buffer);
+
+    //
+    // Backend
+    //
+
+    GGML_API ggml_guid_t  ggml_backend_guid(ggml_backend_t backend);
+    GGML_API const char * ggml_backend_name(ggml_backend_t backend);
+    GGML_API void         ggml_backend_free(ggml_backend_t backend);
+
+    GGML_API ggml_backend_buffer_type_t ggml_backend_get_default_buffer_type(ggml_backend_t backend);
+    GGML_API ggml_backend_buffer_t      ggml_backend_alloc_buffer(ggml_backend_t backend, size_t size);
+    GGML_API size_t                     ggml_backend_get_alignment(ggml_backend_t backend);
+    GGML_API size_t                     ggml_backend_get_max_size(ggml_backend_t backend);
+
+    GGML_API void ggml_backend_tensor_set_async(ggml_backend_t backend,       struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+    GGML_API void ggml_backend_tensor_get_async(ggml_backend_t backend, const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+
+    // "offset" refers to the offset of the tensor data for setting/getting data
+    GGML_API GGML_CALL void ggml_backend_tensor_set(      struct ggml_tensor * tensor, const void * data, size_t offset, size_t size);
+    GGML_API GGML_CALL void ggml_backend_tensor_get(const struct ggml_tensor * tensor,       void * data, size_t offset, size_t size);
+
+    GGML_API void ggml_backend_synchronize(ggml_backend_t backend);
+
+    GGML_API ggml_backend_graph_plan_t ggml_backend_graph_plan_create(ggml_backend_t backend, struct ggml_cgraph * cgraph);
+    GGML_API void                      ggml_backend_graph_plan_free  (ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+
+    GGML_API enum ggml_status ggml_backend_graph_plan_compute (ggml_backend_t backend, ggml_backend_graph_plan_t plan);
+    GGML_API enum ggml_status ggml_backend_graph_compute      (ggml_backend_t backend, struct ggml_cgraph * cgraph);
+    GGML_API enum ggml_status ggml_backend_graph_compute_async(ggml_backend_t backend, struct ggml_cgraph * cgraph);
+    GGML_API bool ggml_backend_supports_op(ggml_backend_t backend, const struct ggml_tensor * op);
+    GGML_API bool ggml_backend_supports_buft(ggml_backend_t backend, ggml_backend_buffer_type_t buft);
+    GGML_API bool ggml_backend_offload_op(ggml_backend_t backend, const struct ggml_tensor * op);
+
+    // tensor copy between different backends
+    GGML_API void ggml_backend_tensor_copy(struct ggml_tensor * src, struct ggml_tensor * dst);
+
+    // asynchronous copy
+    // the copy is performed after all the currently queued operations in backend_src
+    // backend_dst will wait for the copy to complete before performing other operations
+    // automatic fallback to sync copy if async is not supported
+    GGML_API void ggml_backend_tensor_copy_async(ggml_backend_t backend_src, ggml_backend_t backend_dst, struct ggml_tensor * src, struct ggml_tensor * dst);
+
+    // events
+    GGML_API ggml_backend_event_t   ggml_backend_event_new        (ggml_backend_t backend);
+    GGML_API void                   ggml_backend_event_free       (ggml_backend_event_t event);
+    GGML_API void                   ggml_backend_event_record     (ggml_backend_event_t event);
+    GGML_API void                   ggml_backend_event_synchronize(ggml_backend_event_t event);
+    GGML_API void                   ggml_backend_event_wait       (ggml_backend_t backend, ggml_backend_event_t event);
+
+    //
+    // CPU backend
+    //
+
+    GGML_API ggml_backend_t ggml_backend_cpu_init(void);
+
+    GGML_API GGML_CALL bool ggml_backend_is_cpu                (ggml_backend_t backend);
+    GGML_API           void ggml_backend_cpu_set_n_threads     (ggml_backend_t backend_cpu, int n_threads);
+    GGML_API           void ggml_backend_cpu_set_threadpool    (ggml_backend_t backend_cpu, ggml_threadpool_t threadpool);
+    GGML_API           void ggml_backend_cpu_set_abort_callback(ggml_backend_t backend_cpu, ggml_abort_callback abort_callback, void * abort_callback_data);
+
+    // Create a backend buffer from an existing pointer
+    GGML_API GGML_CALL ggml_backend_buffer_t ggml_backend_cpu_buffer_from_ptr(void * ptr, size_t size);
+
+    GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cpu_buffer_type(void);
+
+#ifdef GGML_USE_CPU_HBM
+    GGML_API ggml_backend_buffer_type_t ggml_backend_cpu_hbm_buffer_type(void);
+#endif
+
+    //
+    // Backend registry
+    //
+
+    // The backend registry is a registry of all the available backends, and allows initializing backends in a generic way
+
+    GGML_API size_t                     ggml_backend_reg_get_count(void);
+    GGML_API size_t                     ggml_backend_reg_find_by_name(const char * name);
+    GGML_API ggml_backend_t             ggml_backend_reg_init_backend_from_str(const char * backend_str); // str is backend_name:params (params is optional)
+    GGML_API const char *               ggml_backend_reg_get_name(size_t i);
+    GGML_API ggml_backend_t             ggml_backend_reg_init_backend(size_t i, const char * params); // params is backend-specific
+    GGML_API ggml_backend_buffer_type_t ggml_backend_reg_get_default_buffer_type(size_t i);
+    GGML_API ggml_backend_buffer_t      ggml_backend_reg_alloc_buffer(size_t i, size_t size);
+
+    //
+    // Backend scheduler
+    //
+
+    // The backend scheduler allows for multiple backends to be used together
+    // Handles compute buffer allocation, assignment of tensors to backends, and copying of tensors between backends
+    // The backends are selected based on:
+    // - the backend that supports the operation
+    // - the location of the pre-allocated tensors (e.g. the weights)
+    /*
+      Example usage:
+
+        // operations that use tensors allocated in a buffer with USAGE_WEIGHTS will be assigned
+        // preferrably to run on the same backend as the buffer
+        ggml_backend_buffer_set_usage(buf_weights, GGML_BACKEND_BUFFER_USAGE_WEIGHTS);
+
+        sched = ggml_backend_sched_new({backend_gpu, backend_gpu2, backend_cpu}, NULL, num_backends, GGML_DEFAULT_GRAPH_SIZE, false);
+
+        // initialize buffers from a max size graph (optional)
+        reserve_graph = build_graph(sched, max_batch_size);
+
+        // manually assign nodes to a backend (optional, should not be needed in most cases)
+        struct ggml_tensor * node = ggml_mul_mat(ctx, ...);
+        ggml_backend_sched_set_tensor_backend(sched, node, backend_gpu);
+
+        ggml_backend_sched_reserve(sched, reserve_graph);
+
+        // compute
+        graph = build_graph(sched);
+        ggml_backend_sched_graph_compute(sched, graph);
+
+        // if there are graph inputs:
+        ggml_backend_sched_reset(sched);
+        ggml_backend_sched_alloc_graph(sched, graph);
+        ggml_backend_tensor_set(input_tensor, ...);
+        ggml_backend_sched_graph_compute(sched, graph);
+    }
+    */
+
+    struct ggml_backend_sched;
+    typedef struct ggml_backend_sched * ggml_backend_sched_t;
+
+    // when ask == true, the scheduler wants to know if the user wants to observe this node
+    // this allows the scheduler to batch nodes together in order to evaluate them in a single call
+    //
+    // when ask == false, the scheduler is passing the node tensor to the user for observation
+    // if the user returns false, the scheduler will cancel the graph compute
+    //
+    typedef bool (*ggml_backend_sched_eval_callback)(struct ggml_tensor * t, bool ask, void * user_data);
+
+    // Initialize a backend scheduler
+    GGML_API ggml_backend_sched_t ggml_backend_sched_new(ggml_backend_t * backends, ggml_backend_buffer_type_t * bufts, int n_backends, size_t graph_size, bool parallel);
+    GGML_API void                 ggml_backend_sched_free(ggml_backend_sched_t sched);
+
+    // Initialize backend buffers from a measure graph
+    GGML_API bool                 ggml_backend_sched_reserve(ggml_backend_sched_t sched, struct ggml_cgraph * measure_graph);
+
+    GGML_API int                  ggml_backend_sched_get_n_backends(ggml_backend_sched_t sched);
+    GGML_API ggml_backend_t       ggml_backend_sched_get_backend(ggml_backend_sched_t sched, int i);
+
+    // Get the number of splits of the last graph
+    GGML_API int                  ggml_backend_sched_get_n_splits(ggml_backend_sched_t sched);
+    GGML_API int                  ggml_backend_sched_get_n_copies(ggml_backend_sched_t sched);
+
+    GGML_API size_t               ggml_backend_sched_get_buffer_size(ggml_backend_sched_t sched, ggml_backend_t backend);
+
+    GGML_API void                 ggml_backend_sched_set_tensor_backend(ggml_backend_sched_t sched, struct ggml_tensor * node, ggml_backend_t backend);
+    GGML_API ggml_backend_t       ggml_backend_sched_get_tensor_backend(ggml_backend_sched_t sched, struct ggml_tensor * node);
+
+    // Allocate and compute graph on the backend scheduler
+    GGML_API bool                 ggml_backend_sched_alloc_graph(ggml_backend_sched_t sched, struct ggml_cgraph * graph);
+    GGML_API enum ggml_status     ggml_backend_sched_graph_compute(ggml_backend_sched_t sched, struct ggml_cgraph * graph);
+    GGML_API enum ggml_status     ggml_backend_sched_graph_compute_async(ggml_backend_sched_t sched, struct ggml_cgraph * graph);
+    GGML_API void                 ggml_backend_sched_synchronize(ggml_backend_sched_t sched);
+
+    // Reset all assignments and allocators - must be called before changing the node backends
+    GGML_API void                 ggml_backend_sched_reset(ggml_backend_sched_t sched);
+
+    // Set a callback to be called for each resulting node during graph compute
+    GGML_API void                 ggml_backend_sched_set_eval_callback(ggml_backend_sched_t sched, ggml_backend_sched_eval_callback callback, void * user_data);
+
+    //
+    // Utils
+    //
+
+    struct ggml_backend_graph_copy {
+        ggml_backend_buffer_t buffer;
+        struct ggml_context * ctx_allocated;
+        struct ggml_context * ctx_unallocated;
+        struct ggml_cgraph * graph;
+    };
+
+    // Copy a graph to a different backend
+    GGML_API struct ggml_backend_graph_copy ggml_backend_graph_copy(ggml_backend_t backend, struct ggml_cgraph * graph);
+    GGML_API void                           ggml_backend_graph_copy_free(struct ggml_backend_graph_copy copy);
+
+    typedef bool (*GGML_CALL ggml_backend_eval_callback)(int node_index, struct ggml_tensor * t1, struct ggml_tensor * t2, void * user_data);
+
+    // Compare the output of two backends
+    GGML_API bool ggml_backend_compare_graph_backend(ggml_backend_t backend1, ggml_backend_t backend2, struct ggml_cgraph * graph, ggml_backend_eval_callback callback, void * user_data);
+
+    // Tensor initialization
+    GGML_API void ggml_backend_tensor_alloc(ggml_backend_buffer_t buffer, struct ggml_tensor * tensor, void * addr);
+    GGML_API void ggml_backend_view_init(struct ggml_tensor * tensor);
+
+
+#ifdef  __cplusplus
+}
+#endif
--- a/llama/ggml-common.h
+++ b/llama/ggml-common.h
--- a/llama/ggml-cuda.cu
+++ b/llama/ggml-cuda.cu
--- a/llama/ggml-cuda.h
+++ b/llama/ggml-cuda.h
@@ -0,0 +1,75 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+#include "ggml-backend.h"
+
+#ifdef GGML_USE_HIPBLAS
+#define GGML_CUDA_NAME "ROCm"
+#define GGML_CUBLAS_NAME "hipBLAS"
+#elif defined(GGML_USE_MUSA)
+#define GGML_CUDA_NAME "MUSA"
+#define GGML_CUBLAS_NAME "muBLAS"
+#else
+#define GGML_CUDA_NAME "CUDA"
+#define GGML_CUBLAS_NAME "cuBLAS"
+#endif
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+#define GGML_CUDA_MAX_DEVICES       16
+
+// backend API
+GGML_API GGML_CALL ggml_backend_t ggml_backend_cuda_init(int device);
+
+GGML_API GGML_CALL bool ggml_backend_is_cuda(ggml_backend_t backend);
+
+// device buffer
+GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_buffer_type(int device);
+
+// split tensor buffer that splits matrices by rows across multiple devices
+GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_split_buffer_type(const float * tensor_split);
+
+// pinned host buffer for use with the CPU backend for faster copies between CPU and GPU
+GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_host_buffer_type(void);
+
+GGML_API GGML_CALL int ggml_backend_cuda_reg_devices();
+
+GGML_API GGML_CALL int  ggml_backend_cuda_get_device_count(void);
+GGML_API GGML_CALL void ggml_backend_cuda_get_device_description(int device, char * description, size_t description_size);
+GGML_API GGML_CALL void ggml_backend_cuda_get_device_memory(int device, size_t * free, size_t * total);
+
+GGML_API GGML_CALL bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size);
+GGML_API GGML_CALL void ggml_backend_cuda_unregister_host_buffer(void * buffer);
+
+GGML_API void ggml_backend_cuda_log_set_callback(ggml_log_callback log_callback, void * user_data);
+#ifdef  __cplusplus
+}
+#endif
--- a/llama/ggml-cuda/acc.cu
+++ b/llama/ggml-cuda/acc.cu
@@ -0,0 +1,73 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "acc.cuh"
+
+static __global__ void acc_f32(const float * x, const float * y, float * dst, const int ne,
+    const int ne10, const int ne11, const int ne12,
+    const int nb1, const int nb2, int offset) {
+    const int i = blockDim.x * blockIdx.x + threadIdx.x;
+    if (i >= ne) {
+        return;
+    }
+    int src1_idx = i - offset;
+    int oz = src1_idx / nb2;
+    int oy = (src1_idx - (oz * nb2)) / nb1;
+    int ox = src1_idx % nb1;
+    if (src1_idx >= 0 && ox < ne10 && oy < ne11 && oz < ne12) {
+        dst[i] = x[i] + y[ox + oy * ne10 + oz * ne10 * ne11];
+    } else {
+        dst[i] = x[i];
+    }
+}
+
+static void acc_f32_cuda(const float * x, const float * y, float * dst, const int n_elements,
+    const int ne10, const int ne11, const int ne12,
+    const int nb1, const int nb2, const int offset, cudaStream_t stream) {
+    int num_blocks = (n_elements + CUDA_ACC_BLOCK_SIZE - 1) / CUDA_ACC_BLOCK_SIZE;
+    acc_f32<<<num_blocks, CUDA_ACC_BLOCK_SIZE, 0, stream>>>(x, y, dst, n_elements, ne10, ne11, ne12, nb1, nb2, offset);
+}
+
+void ggml_cuda_op_acc(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const ggml_tensor * src1 = dst->src[1];
+    const float * src0_d = (const float *)src0->data;
+    const float * src1_d = (const float *)src1->data;
+    float * dst_d = (float *)dst->data;
+    cudaStream_t stream = ctx.stream();
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT(src1->type == GGML_TYPE_F32);
+    GGML_ASSERT( dst->type == GGML_TYPE_F32);
+    GGML_ASSERT(dst->ne[3] == 1); // just 3D tensors supported
+
+    int nb1 = dst->op_params[0] / 4; // 4 bytes of float32
+    int nb2 = dst->op_params[1] / 4; // 4 bytes of float32
+    // int nb3 = dst->op_params[2] / 4; // 4 bytes of float32 - unused
+    int offset = dst->op_params[3] / 4; // offset in bytes
+
+    acc_f32_cuda(src0_d, src1_d, dst_d, ggml_nelements(dst), src1->ne[0], src1->ne[1], src1->ne[2], nb1, nb2, offset, stream);
+}
--- a/llama/ggml-cuda/acc.cuh
+++ b/llama/ggml-cuda/acc.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_ACC_BLOCK_SIZE 256
+
+void ggml_cuda_op_acc(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/arange.cu
+++ b/llama/ggml-cuda/arange.cu
@@ -0,0 +1,60 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "arange.cuh"
+
+static __global__ void arange_f32(float * dst, const int ne0, const float start, const float step) {
+    // blockIDx.x: idx of ne0 / BLOCK_SIZE
+    int nidx = threadIdx.x + blockIdx.x * blockDim.x;
+    if (nidx >= ne0) {
+        return;
+    }
+    dst[nidx] = start + step * nidx;
+}
+
+static void arange_f32_cuda(float * dst, const int ne0, const float start, const float step, cudaStream_t stream) {
+    int num_blocks = (ne0 + CUDA_ARANGE_BLOCK_SIZE - 1) / CUDA_ARANGE_BLOCK_SIZE;
+    arange_f32<<<num_blocks, CUDA_ARANGE_BLOCK_SIZE, 0, stream>>>(dst, ne0, start,  step);
+}
+
+void ggml_cuda_op_arange(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    float * dst_d = (float *)dst->data;
+    cudaStream_t stream = ctx.stream();
+
+    GGML_ASSERT(dst->type == GGML_TYPE_F32);
+
+    float start;
+    float stop;
+    float step;
+    memcpy(&start, (float *)dst->op_params + 0, sizeof(float));
+    memcpy(&stop,  (float *)dst->op_params + 1, sizeof(float));
+    memcpy(&step,  (float *)dst->op_params + 2, sizeof(float));
+
+    int64_t steps = (int64_t)ceil((stop - start) / step);
+    GGML_ASSERT(ggml_nelements(dst) == steps);
+
+    arange_f32_cuda(dst_d, dst->ne[0], start, step, stream);
+}
--- a/llama/ggml-cuda/arange.cuh
+++ b/llama/ggml-cuda/arange.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_ARANGE_BLOCK_SIZE 256
+
+void ggml_cuda_op_arange(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/argsort.cu
+++ b/llama/ggml-cuda/argsort.cu
@@ -0,0 +1,130 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "argsort.cuh"
+
+template<typename T>
+static inline __device__ void ggml_cuda_swap(T & a, T & b) {
+    T tmp = a;
+    a = b;
+    b = tmp;
+}
+
+template<ggml_sort_order order>
+static __global__ void k_argsort_f32_i32(const float * x, int * dst, const int ncols, int ncols_pad) {
+    // bitonic sort
+    int col = threadIdx.x;
+    int row = blockIdx.y;
+
+    if (col >= ncols_pad) {
+        return;
+    }
+
+    const float * x_row = x + row * ncols;
+    extern __shared__ int dst_row[];
+
+    // initialize indices
+    dst_row[col] = col;
+
+    __syncthreads();
+
+    for (int k = 2; k <= ncols_pad; k *= 2) {
+        for (int j = k / 2; j > 0; j /= 2) {
+            int ixj = col ^ j;
+            if (ixj > col) {
+                if ((col & k) == 0) {
+                    if (dst_row[col] >= ncols ||
+                        (dst_row[ixj] < ncols && (order == GGML_SORT_ORDER_ASC ?
+                            x_row[dst_row[col]] > x_row[dst_row[ixj]] :
+                            x_row[dst_row[col]] < x_row[dst_row[ixj]]))
+                    ) {
+                        ggml_cuda_swap(dst_row[col], dst_row[ixj]);
+                    }
+                } else {
+                    if (dst_row[ixj] >= ncols ||
+                        (dst_row[col] < ncols && (order == GGML_SORT_ORDER_ASC ?
+                            x_row[dst_row[col]] < x_row[dst_row[ixj]] :
+                            x_row[dst_row[col]] > x_row[dst_row[ixj]]))
+                    ) {
+                        ggml_cuda_swap(dst_row[col], dst_row[ixj]);
+                    }
+                }
+            }
+            __syncthreads();
+        }
+    }
+
+    // copy the result to dst without the padding
+    if (col < ncols) {
+        dst[row * ncols + col] = dst_row[col];
+    }
+}
+
+static int next_power_of_2(int x) {
+    int n = 1;
+    while (n < x) {
+        n *= 2;
+    }
+    return n;
+}
+
+static void argsort_f32_i32_cuda(const float * x, int * dst, const int ncols, const int nrows, ggml_sort_order order, cudaStream_t stream) {
+    // bitonic sort requires ncols to be power of 2
+    const int ncols_pad = next_power_of_2(ncols);
+
+    const dim3 block_dims(ncols_pad, 1, 1);
+    const dim3 block_nums(1, nrows, 1);
+    const size_t shared_mem = ncols_pad * sizeof(int);
+
+    // FIXME: this limit could be raised by ~2-4x on Ampere or newer
+    GGML_ASSERT(shared_mem <= ggml_cuda_info().devices[ggml_cuda_get_device()].smpb);
+
+    if (order == GGML_SORT_ORDER_ASC) {
+        k_argsort_f32_i32<GGML_SORT_ORDER_ASC><<<block_nums, block_dims, shared_mem, stream>>>(x, dst, ncols, ncols_pad);
+    } else if (order == GGML_SORT_ORDER_DESC) {
+        k_argsort_f32_i32<GGML_SORT_ORDER_DESC><<<block_nums, block_dims, shared_mem, stream>>>(x, dst, ncols, ncols_pad);
+    } else {
+        GGML_ABORT("fatal error");
+    }
+}
+
+void ggml_cuda_op_argsort(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const float * src0_d = (const float *)src0->data;
+    float * dst_d = (float *)dst->data;
+    cudaStream_t stream = ctx.stream();
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT( dst->type == GGML_TYPE_I32);
+    GGML_ASSERT(ggml_is_contiguous(src0));
+
+    const int64_t ncols = src0->ne[0];
+    const int64_t nrows = ggml_nrows(src0);
+
+    enum ggml_sort_order order = (enum ggml_sort_order) dst->op_params[0];
+
+    argsort_f32_i32_cuda(src0_d, (int *)dst_d, ncols, nrows, order, stream);
+}
--- a/llama/ggml-cuda/argsort.cuh
+++ b/llama/ggml-cuda/argsort.cuh
@@ -0,0 +1,29 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+void ggml_cuda_op_argsort(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/binbcast.cu
+++ b/llama/ggml-cuda/binbcast.cu
@@ -0,0 +1,314 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "binbcast.cuh"
+
+static __device__ __forceinline__ float op_repeat(const float a, const float b) {
+    return b;
+    GGML_UNUSED(a);
+}
+
+static __device__ __forceinline__ float op_add(const float a, const float b) {
+    return a + b;
+}
+
+static __device__ __forceinline__ float op_sub(const float a, const float b) {
+    return a - b;
+}
+
+static __device__ __forceinline__ float op_mul(const float a, const float b) {
+    return a * b;
+}
+
+static __device__ __forceinline__ float op_div(const float a, const float b) {
+    return a / b;
+}
+
+template<float (*bin_op)(const float, const float), typename src0_t, typename src1_t, typename dst_t>
+static __global__ void k_bin_bcast(const src0_t * src0, const src1_t * src1, dst_t * dst,
+        int ne0, int ne1, int ne2, int ne3,
+        int ne10, int ne11, int ne12, int ne13,
+        /*int s0, */ int s1,  int s2,  int s3,
+        /*int s00,*/ int s01, int s02, int s03,
+        /*int s10,*/ int s11, int s12, int s13) {
+    const int i0s = blockDim.x*blockIdx.x + threadIdx.x;
+    const int i1 = (blockDim.y*blockIdx.y + threadIdx.y);
+    const int i2 = (blockDim.z*blockIdx.z + threadIdx.z) / ne3;
+    const int i3 = (blockDim.z*blockIdx.z + threadIdx.z) % ne3;
+
+    if (i0s >= ne0 || i1 >= ne1 || i2 >= ne2 || i3 >= ne3) {
+        return;
+    }
+
+    const int i11 = i1 % ne11;
+    const int i12 = i2 % ne12;
+    const int i13 = i3 % ne13;
+
+    const size_t i_src0 =  i3*s03 +  i2*s02 +  i1*s01;
+    const size_t i_src1 = i13*s13 + i12*s12 + i11*s11;
+    const size_t i_dst  =  i3*s3  +  i2*s2  +  i1*s1;
+
+    const src0_t * src0_row = src0 + i_src0;
+    const src1_t * src1_row = src1 + i_src1;
+    dst_t * dst_row = dst + i_dst;
+
+    for (int i0 = i0s; i0 < ne0; i0 += blockDim.x*gridDim.x) {
+        const int i10 = i0 % ne10;
+        dst_row[i0] = (dst_t)bin_op(src0 ? (float)src0_row[i0] : 0.0f, (float)src1_row[i10]);
+    }
+}
+
+template<float (*bin_op)(const float, const float), typename src0_t, typename src1_t, typename dst_t>
+static __global__ void k_bin_bcast_unravel(const src0_t * src0, const src1_t * src1, dst_t * dst,
+        int ne0, int ne1, int ne2, int ne3,
+        int ne10, int ne11, int ne12, int ne13,
+        /*int s0, */ int s1,  int s2,  int s3,
+        /*int s00,*/ int s01, int s02, int s03,
+        /*int s10,*/ int s11, int s12, int s13) {
+
+    const int i = blockDim.x*blockIdx.x + threadIdx.x;
+
+    const int i3 = i/(ne2*ne1*ne0);
+    const int i2 = (i/(ne1*ne0)) % ne2;
+    const int i1 = (i/ne0) % ne1;
+    const int i0 = i % ne0;
+
+    if (i0 >= ne0 || i1 >= ne1 || i2 >= ne2 || i3 >= ne3) {
+        return;
+    }
+
+    const int i11 = i1 % ne11;
+    const int i12 = i2 % ne12;
+    const int i13 = i3 % ne13;
+
+    const size_t i_src0 =  i3*s03 +  i2*s02 +  i1*s01;
+    const size_t i_src1 = i13*s13 + i12*s12 + i11*s11;
+    const size_t i_dst  =  i3*s3  +  i2*s2  +  i1*s1;
+
+    const src0_t * src0_row = src0 + i_src0;
+    const src1_t * src1_row = src1 + i_src1;
+    dst_t * dst_row = dst + i_dst;
+
+    const int i10 = i0 % ne10;
+    dst_row[i0] = (dst_t)bin_op(src0 ? (float)src0_row[i0] : 0.0f, (float)src1_row[i10]);
+}
+
+template<float (*bin_op)(const float, const float)>
+struct bin_bcast_cuda {
+    template<typename src0_t, typename src1_t, typename dst_t>
+    void operator()(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst,
+            const src0_t * src0_dd, const src1_t * src1_dd, dst_t * dst_dd,
+            cudaStream_t stream) {
+
+        GGML_TENSOR_BINARY_OP_LOCALS
+
+        int nr0 = ne10/ne0;
+        int nr1 = ne11/ne1;
+        int nr2 = ne12/ne2;
+        int nr3 = ne13/ne3;
+
+        int nr[4] = { nr0, nr1, nr2, nr3 };
+
+        // collapse dimensions until first broadcast dimension
+        int64_t cne[] = {ne0, ne1, ne2, ne3};
+        int64_t cne0[] = {ne00, ne01, ne02, ne03};
+        int64_t cne1[] = {ne10, ne11, ne12, ne13};
+
+        size_t cnb[] = {nb0, nb1, nb2, nb3};
+        size_t cnb0[] = {nb00, nb01, nb02, nb03};
+        size_t cnb1[] = {nb10, nb11, nb12, nb13};
+
+        auto collapse = [](int64_t cne[]) {
+            cne[0] *= cne[1];
+            cne[1] = cne[2];
+            cne[2] = cne[3];
+            cne[3] = 1;
+        };
+
+        auto collapse_nb = [](size_t cnb[], const int64_t cne[]) {
+            cnb[1] *= cne[1];
+            cnb[2] *= cne[2];
+            cnb[3] *= cne[3];
+        };
+
+        if (ggml_is_contiguous(src0) && ggml_is_contiguous(src1) && ggml_is_contiguous(dst)) {
+            for (int i = 0; i < 4; i++) {
+                if (nr[i] != 1) {
+                    break;
+                }
+                if (i > 0) {
+                    collapse_nb(cnb, cne);
+                    collapse_nb(cnb0, cne0);
+                    collapse_nb(cnb1, cne1);
+                    collapse(cne);
+                    collapse(cne0);
+                    collapse(cne1);
+                }
+            }
+        }
+
+        {
+            int64_t ne0 = cne[0];
+            int64_t ne1 = cne[1];
+            int64_t ne2 = cne[2];
+            int64_t ne3 = cne[3];
+
+            //int64_t ne00 = cne0[0]; GGML_UNUSED(ne00);
+            //int64_t ne01 = cne0[1]; GGML_UNUSED(ne01);
+            //int64_t ne02 = cne0[2]; GGML_UNUSED(ne02);
+            //int64_t ne03 = cne0[3]; GGML_UNUSED(ne03);
+
+            int64_t ne10 = cne1[0];
+            int64_t ne11 = cne1[1];
+            int64_t ne12 = cne1[2];
+            int64_t ne13 = cne1[3];
+
+            size_t nb0 = cnb[0];
+            size_t nb1 = cnb[1];
+            size_t nb2 = cnb[2];
+            size_t nb3 = cnb[3];
+
+            size_t nb00 = cnb0[0];
+            size_t nb01 = cnb0[1];
+            size_t nb02 = cnb0[2];
+            size_t nb03 = cnb0[3];
+
+            size_t nb10 = cnb1[0];
+            size_t nb11 = cnb1[1];
+            size_t nb12 = cnb1[2];
+            size_t nb13 = cnb1[3];
+
+            size_t s0 = nb0 / sizeof(dst_t);
+            size_t s1 = nb1 / sizeof(dst_t);
+            size_t s2 = nb2 / sizeof(dst_t);
+            size_t s3 = nb3 / sizeof(dst_t);
+
+            size_t s10 = nb10 / sizeof(src1_t);
+            size_t s11 = nb11 / sizeof(src1_t);
+            size_t s12 = nb12 / sizeof(src1_t);
+            size_t s13 = nb13 / sizeof(src1_t);
+
+            size_t s00 = nb00 / sizeof(src0_t);
+            size_t s01 = nb01 / sizeof(src0_t);
+            size_t s02 = nb02 / sizeof(src0_t);
+            size_t s03 = nb03 / sizeof(src0_t);
+
+            GGML_ASSERT(nb0 % sizeof(dst_t) == 0);
+            GGML_ASSERT(nb1 % sizeof(dst_t) == 0);
+            GGML_ASSERT(nb2 % sizeof(dst_t) == 0);
+            GGML_ASSERT(nb3 % sizeof(dst_t) == 0);
+
+            GGML_ASSERT(nb00 % sizeof(src0_t) == 0);
+            GGML_ASSERT(nb01 % sizeof(src0_t) == 0);
+            GGML_ASSERT(nb02 % sizeof(src0_t) == 0);
+            GGML_ASSERT(nb03 % sizeof(src0_t) == 0);
+
+            GGML_ASSERT(nb10 % sizeof(src1_t) == 0);
+            GGML_ASSERT(nb11 % sizeof(src1_t) == 0);
+            GGML_ASSERT(nb12 % sizeof(src1_t) == 0);
+            GGML_ASSERT(nb13 % sizeof(src1_t) == 0);
+
+            GGML_ASSERT(s0 == 1);
+            GGML_ASSERT(s00 == 1);
+            GGML_ASSERT(s10 == 1);
+
+            const int block_size = 128;
+
+            int64_t hne0 = std::max(ne0/2LL, 1LL);
+
+            dim3 block_dims;
+            block_dims.x = std::min<unsigned int>(hne0, block_size);
+            block_dims.y = std::min<unsigned int>(ne1, block_size / block_dims.x);
+            block_dims.z = std::min(std::min<unsigned int>(ne2*ne3, block_size / block_dims.x / block_dims.y), 64U);
+
+            dim3 block_nums(
+                (hne0 + block_dims.x - 1) / block_dims.x,
+                (ne1 + block_dims.y - 1) / block_dims.y,
+                (ne2*ne3 + block_dims.z - 1) / block_dims.z
+            );
+
+            if (block_nums.z > 65535) {
+                // this is the maximum number of blocks in z dimension, fallback to 1D grid kernel
+                int block_num = (ne0*ne1*ne2*ne3 + block_size - 1) / block_size;
+                k_bin_bcast_unravel<bin_op><<<block_num, block_size, 0, stream>>>(
+                    src0_dd, src1_dd, dst_dd,
+                    ne0, ne1, ne2, ne3,
+                    ne10, ne11, ne12, ne13,
+                    /* s0, */ s1, s2, s3,
+                    /* s00, */ s01, s02, s03,
+                    /* s10, */ s11, s12, s13);
+            } else {
+                k_bin_bcast<bin_op><<<block_nums, block_dims, 0, stream>>>(
+                    src0_dd, src1_dd, dst_dd,
+                    ne0, ne1, ne2, ne3,
+                    ne10, ne11, ne12, ne13,
+                    /* s0, */ s1, s2, s3,
+                    /* s00, */ s01, s02, s03,
+                    /* s10, */ s11, s12, s13);
+            }
+        }
+    }
+};
+
+template<class op>
+static void ggml_cuda_op_bin_bcast(
+    const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst,
+    const void * src0_dd, const void * src1_dd, void * dst_dd, cudaStream_t stream) {
+
+    GGML_ASSERT(src1->type == GGML_TYPE_F32);
+
+    if (src0->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32) {
+        op()(src0, src1, dst, (const float *)src0_dd, (const float *)src1_dd, (float *)dst_dd, stream);
+    } else if (src0->type == GGML_TYPE_F16 && dst->type == GGML_TYPE_F16) {
+        op()(src0, src1, dst, (const half *) src0_dd, (const float *)src1_dd, (half *) dst_dd, stream);
+    } else if (src0->type == GGML_TYPE_F16 && dst->type == GGML_TYPE_F32) {
+        op()(src0, src1, dst, (const half *) src0_dd, (const float *)src1_dd, (float *)dst_dd, stream);
+    } else {
+        fprintf(stderr, "%s: unsupported types: dst: %s, src0: %s, src1: %s\n", __func__,
+            ggml_type_name(dst->type), ggml_type_name(src0->type), ggml_type_name(src1->type));
+        GGML_ABORT("fatal error");
+    }
+}
+
+void ggml_cuda_op_repeat(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    ggml_cuda_op_bin_bcast<bin_bcast_cuda<op_repeat>>(dst, dst->src[0], dst, nullptr, dst->src[0]->data, dst->data, ctx.stream());
+}
+
+void ggml_cuda_op_add(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    ggml_cuda_op_bin_bcast<bin_bcast_cuda<op_add>>(dst->src[0], dst->src[1], dst, dst->src[0]->data, dst->src[1]->data, dst->data, ctx.stream());
+}
+
+void ggml_cuda_op_sub(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    ggml_cuda_op_bin_bcast<bin_bcast_cuda<op_sub>>(dst->src[0], dst->src[1], dst, dst->src[0]->data, dst->src[1]->data, dst->data, ctx.stream());
+}
+
+void ggml_cuda_op_mul(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    ggml_cuda_op_bin_bcast<bin_bcast_cuda<op_mul>>(dst->src[0], dst->src[1], dst, dst->src[0]->data, dst->src[1]->data, dst->data, ctx.stream());
+}
+
+void ggml_cuda_op_div(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    ggml_cuda_op_bin_bcast<bin_bcast_cuda<op_div>>(dst->src[0], dst->src[1], dst, dst->src[0]->data, dst->src[1]->data, dst->data, ctx.stream());
+}
--- a/llama/ggml-cuda/binbcast.cuh
+++ b/llama/ggml-cuda/binbcast.cuh
@@ -0,0 +1,33 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+void ggml_cuda_op_repeat(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
+void ggml_cuda_op_add(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
+void ggml_cuda_op_sub(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
+void ggml_cuda_op_mul(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
+void ggml_cuda_op_div(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/clamp.cu
+++ b/llama/ggml-cuda/clamp.cu
@@ -0,0 +1,60 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "clamp.cuh"
+
+static __global__ void clamp_f32(const float * x, float * dst, const float min, const float max, const int k) {
+    const int i = blockDim.x*blockIdx.x + threadIdx.x;
+
+    if (i >= k) {
+        return;
+    }
+
+    dst[i] = x[i] < min ? min : (x[i] > max ? max : x[i]);
+}
+
+static void clamp_f32_cuda(const float * x, float * dst, const float min, const float max, const int k, cudaStream_t stream) {
+    const int num_blocks = (k + CUDA_CLAMP_BLOCK_SIZE - 1) / CUDA_CLAMP_BLOCK_SIZE;
+    clamp_f32<<<num_blocks, CUDA_CLAMP_BLOCK_SIZE, 0, stream>>>(x, dst, min, max, k);
+}
+
+
+void ggml_cuda_op_clamp(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const float * src0_d = (const float *)src0->data;
+    float * dst_d = (float *)dst->data;
+    cudaStream_t stream = ctx.stream();
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT( dst->type == GGML_TYPE_F32);
+
+    float min;
+    float max;
+    memcpy(&min, dst->op_params, sizeof(float));
+    memcpy(&max, (float *) dst->op_params + 1, sizeof(float));
+
+    clamp_f32_cuda(src0_d, dst_d, min, max, ggml_nelements(src0), stream);
+}
--- a/llama/ggml-cuda/clamp.cuh
+++ b/llama/ggml-cuda/clamp.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_CLAMP_BLOCK_SIZE 256
+
+void ggml_cuda_op_clamp(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/common.cuh
+++ b/llama/ggml-cuda/common.cuh
@@ -0,0 +1,695 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#pragma once
+
+#include "ggml.h"
+#include "ggml-cuda.h"
+
+#include <cstdint>
+#include <memory>
+
+#if defined(GGML_USE_HIPBLAS)
+#define GGML_COMMON_DECL_HIP
+#define GGML_COMMON_IMPL_HIP
+#else
+#define GGML_COMMON_DECL_CUDA
+#define GGML_COMMON_IMPL_CUDA
+#if defined(GGML_USE_MUSA)
+#define GGML_COMMON_DECL_MUSA
+#define GGML_COMMON_IMPL_MUSA
+#endif
+#endif
+#include "ggml-common.h"
+
+#include <cstdio>
+#include <array>
+#include <cassert>
+#include <cfloat>
+#include <string>
+#include <vector>
+
+#if defined(GGML_USE_HIPBLAS)
+#include "vendors/hip.h"
+#elif defined(GGML_USE_MUSA)
+#include "vendors/musa.h"
+#else
+#include "vendors/cuda.h"
+#endif // defined(GGML_USE_HIPBLAS)
+
+#define STRINGIZE_IMPL(...) #__VA_ARGS__
+#define STRINGIZE(...) STRINGIZE_IMPL(__VA_ARGS__)
+
+#define WARP_SIZE 32
+#define CUDART_HMAX   11070 // CUDA 11.7, min. ver. for which __hmax and __hmax2 are known to work (may be higher than needed)
+#define CUDART_HMASK  12000 // CUDA 12.0, min. ver. for half2 -> uint mask comparisons
+
+#define CC_PASCAL     600
+#define MIN_CC_DP4A   610 // minimum compute capability for __dp4a, an intrinsic for byte-wise dot products
+#define CC_VOLTA      700
+#define CC_TURING     750
+#define CC_AMPERE     800
+#define CC_OFFSET_AMD 1000000
+#define CC_RDNA1      (CC_OFFSET_AMD + 1010)
+#define CC_RDNA2      (CC_OFFSET_AMD + 1030)
+#define CC_RDNA3      (CC_OFFSET_AMD + 1100)
+
+#define MATRIX_ROW_PADDING 512 // last row of quant. matrices is a multiple of this to avoid out-of-bounds memory accesses
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244 4267) // possible loss of data
+#endif
+
+#define GGML_CUDA_MAX_STREAMS 8
+
+[[noreturn]]
+void ggml_cuda_error(const char * stmt, const char * func, const char * file, int line, const char * msg);
+
+#define CUDA_CHECK_GEN(err, success, error_fn)                                      \
+     do {                                                                           \
+        auto err_ = (err);                                                          \
+        if (err_ != (success)) {                                                    \
+            ggml_cuda_error(#err, __func__, __FILE__, __LINE__, error_fn(err_));    \
+        }                                                                           \
+    } while (0)
+
+#define CUDA_CHECK(err) CUDA_CHECK_GEN(err, cudaSuccess, cudaGetErrorString)
+
+#if CUDART_VERSION >= 12000 || defined(GGML_USE_MUSA)
+    static const char * cublas_get_error_str(const cublasStatus_t err) {
+        return cublasGetStatusString(err);
+    }
+#else
+    static const char * cublas_get_error_str(const cublasStatus_t err) {
+        switch (err) {
+            case CUBLAS_STATUS_SUCCESS: return "CUBLAS_STATUS_SUCCESS";
+            case CUBLAS_STATUS_NOT_INITIALIZED: return "CUBLAS_STATUS_NOT_INITIALIZED";
+            case CUBLAS_STATUS_ALLOC_FAILED: return "CUBLAS_STATUS_ALLOC_FAILED";
+            case CUBLAS_STATUS_INVALID_VALUE: return "CUBLAS_STATUS_INVALID_VALUE";
+            case CUBLAS_STATUS_ARCH_MISMATCH: return "CUBLAS_STATUS_ARCH_MISMATCH";
+            case CUBLAS_STATUS_MAPPING_ERROR: return "CUBLAS_STATUS_MAPPING_ERROR";
+            case CUBLAS_STATUS_EXECUTION_FAILED: return "CUBLAS_STATUS_EXECUTION_FAILED";
+            case CUBLAS_STATUS_INTERNAL_ERROR: return "CUBLAS_STATUS_INTERNAL_ERROR";
+            case CUBLAS_STATUS_NOT_SUPPORTED: return "CUBLAS_STATUS_NOT_SUPPORTED";
+            default: return "unknown error";
+        }
+    }
+#endif // CUDART_VERSION >= 12000
+
+#define CUBLAS_CHECK(err) CUDA_CHECK_GEN(err, CUBLAS_STATUS_SUCCESS, cublas_get_error_str)
+
+#if !defined(GGML_USE_HIPBLAS)
+static const char * cu_get_error_str(CUresult err) {
+    const char * err_str;
+    cuGetErrorString(err, &err_str);
+    return err_str;
+}
+#define CU_CHECK(err) CUDA_CHECK_GEN(err, CUDA_SUCCESS, cu_get_error_str)
+#endif
+
+#if CUDART_VERSION >= 11100 || defined(GGML_USE_MUSA)
+#define GGML_CUDA_ASSUME(x) __builtin_assume(x)
+#else
+#define GGML_CUDA_ASSUME(x)
+#endif // CUDART_VERSION >= 11100
+
+#ifdef GGML_CUDA_F16
+typedef half dfloat; // dequantize float
+typedef half2 dfloat2;
+#else
+typedef float dfloat; // dequantize float
+typedef float2 dfloat2;
+#endif // GGML_CUDA_F16
+
+#if (defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) || __CUDA_ARCH__ >= CC_PASCAL
+#define FP16_AVAILABLE
+#endif // (defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) || __CUDA_ARCH__ >= CC_PASCAL
+
+#if defined(FP16_AVAILABLE) && __CUDA_ARCH__ != 610
+#define FAST_FP16_AVAILABLE
+#endif // defined(FP16_AVAILABLE) && __CUDA_ARCH__ != 610
+
+#if !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && __CUDA_ARCH__ >= CC_VOLTA
+#define FP16_MMA_AVAILABLE
+#endif // !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && __CUDA_ARCH__ >= CC_VOLTA
+
+#if !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && __CUDA_ARCH__ >= CC_TURING
+#define INT8_MMA_AVAILABLE
+#endif // !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && __CUDA_ARCH__ >= CC_TURING
+
+static constexpr bool fast_fp16_available(const int cc) {
+    return cc >= CC_PASCAL && cc != 610;
+}
+
+static constexpr bool fp16_mma_available(const int cc) {
+    return cc < CC_OFFSET_AMD && cc >= CC_VOLTA;
+}
+
+static constexpr bool int8_mma_available(const int cc) {
+    return cc < CC_OFFSET_AMD && cc >= CC_TURING;
+}
+
+[[noreturn]]
+static __device__ void no_device_code(
+    const char * file_name, const int line, const char * function_name, const int arch, const char * arch_list) {
+
+#if defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+    printf("%s:%d: ERROR: HIP kernel %s has no device code compatible with HIP arch %d.\n",
+           file_name, line, function_name, arch);
+    GGML_UNUSED(arch_list);
+#else
+    printf("%s:%d: ERROR: CUDA kernel %s has no device code compatible with CUDA arch %d. ggml-cuda.cu was compiled for: %s\n",
+           file_name, line, function_name, arch, arch_list);
+#endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+    __trap();
+
+    GGML_UNUSED(no_device_code); // suppress unused function warning
+}
+
+#ifdef __CUDA_ARCH__
+#define NO_DEVICE_CODE no_device_code(__FILE__, __LINE__, __FUNCTION__, __CUDA_ARCH__, STRINGIZE(__CUDA_ARCH_LIST__))
+#else
+#define NO_DEVICE_CODE //GGML_ABORT("NO_DEVICE_CODE not valid in host code.")
+#endif // __CUDA_ARCH__
+
+static __device__ __forceinline__ float warp_reduce_sum(float x) {
+#pragma unroll
+    for (int mask = 16; mask > 0; mask >>= 1) {
+        x += __shfl_xor_sync(0xffffffff, x, mask, 32);
+    }
+    return x;
+}
+
+static __device__ __forceinline__ float2 warp_reduce_sum(float2 a) {
+#pragma unroll
+    for (int mask = 16; mask > 0; mask >>= 1) {
+        a.x += __shfl_xor_sync(0xffffffff, a.x, mask, 32);
+        a.y += __shfl_xor_sync(0xffffffff, a.y, mask, 32);
+    }
+    return a;
+}
+
+static __device__ __forceinline__ half2 warp_reduce_sum(half2 a) {
+#ifdef FP16_AVAILABLE
+
+#if defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+#pragma unroll
+    for (int mask = 16; mask > 0; mask >>= 1) {
+        const half2 a_other = __shfl_xor_sync(0xffffffff, a, mask, 32);
+        reinterpret_cast<half&>(a.x) +=  __low2half(a_other);
+        reinterpret_cast<half&>(a.y) += __high2half(a_other);
+    }
+    return a;
+#else
+#pragma unroll
+    for (int mask = 16; mask > 0; mask >>= 1) {
+        a = __hadd2(a, __shfl_xor_sync(0xffffffff, a, mask, 32));
+    }
+    return a;
+#endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+
+#else
+    NO_DEVICE_CODE;
+    return a;
+#endif // FP16_AVAILABLE
+}
+
+static __device__ __forceinline__ float warp_reduce_max(float x) {
+#pragma unroll
+    for (int mask = 16; mask > 0; mask >>= 1) {
+        x = fmaxf(x, __shfl_xor_sync(0xffffffff, x, mask, 32));
+    }
+    return x;
+}
+
+static __device__ __forceinline__ half ggml_cuda_hmax(const half a, const half b) {
+#ifdef FP16_AVAILABLE
+
+#if !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && CUDART_VERSION < CUDART_HMAX
+    return __float2half(fmaxf(__half2float(a), __half2float(b)));
+#else
+    return __hmax(a, b);
+#endif // !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && CUDART_VERSION < CUDART_HMAX
+
+#else
+   NO_DEVICE_CODE;
+   GGML_UNUSED(b);
+   return a;
+#endif // FP16_AVAILABLE
+}
+
+static __device__ __forceinline__ half2 ggml_cuda_hmax2(const half2 a, const half2 b) {
+#if !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__))
+
+#if CUDART_VERSION >= CUDART_HMAX
+    return __hmax2(a, b);
+#else
+    half2 ret;
+    reinterpret_cast<half&>(ret.x) = __float2half(fmaxf( __low2float(a),  __low2float(b)));
+    reinterpret_cast<half&>(ret.y) = __float2half(fmaxf(__high2float(a), __high2float(b)));
+    return ret;
+#endif // CUDART_VERSION >= CUDART_HMAX
+
+#else
+    GGML_UNUSED(a);
+    GGML_UNUSED(b);
+    NO_DEVICE_CODE;
+#endif // !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__))
+}
+
+static __device__ __forceinline__ half2 warp_reduce_max(half2 x) {
+#if !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && __CUDA_ARCH__ >= CC_PASCAL
+#pragma unroll
+   for (int mask = 16; mask > 0; mask >>= 1) {
+       x = ggml_cuda_hmax2(x, __shfl_xor_sync(0xffffffff, x, mask, 32));
+   }
+   return x;
+#else
+   GGML_UNUSED(x);
+   NO_DEVICE_CODE;
+#endif // !(defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)) && __CUDA_ARCH__ >= CC_PASCAL
+}
+
+#if CUDART_VERSION < CUDART_HMASK
+static __device__ __forceinline__ uint32_t __hgt2_mask(const half2 a, const half2 b) {
+    const uint32_t mask_low  = 0x0000FFFF * (float( __low2half(a)) > float( __low2half(b)));
+    const uint32_t mask_high = 0xFFFF0000 * (float(__high2half(a)) > float(__high2half(b)));
+    return mask_low | mask_high;
+}
+#endif // CUDART_VERSION < CUDART_HMASK
+
+static __device__ __forceinline__ int ggml_cuda_dp4a(const int a, const int b, int c) {
+#if defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+#if defined(__gfx906__) || defined(__gfx908__) || defined(__gfx90a__) || defined(RDNA2)
+    c = __builtin_amdgcn_sdot4(a, b, c, false);
+#elif defined(RDNA3)
+    c = __builtin_amdgcn_sudot4( true, a, true, b, c, false);
+#elif defined(__gfx1010__) || defined(__gfx900__)
+    int tmp1;
+    int tmp2;
+    asm("\n \
+        v_mul_i32_i24 %1, sext(%3), sext(%4) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0 \n \
+        v_mul_i32_i24 %2, sext(%3), sext(%4) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1 \n \
+        v_add3_u32 %0, %1, %2, %0 \n \
+        v_mul_i32_i24 %1, sext(%3), sext(%4) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2 \n \
+        v_mul_i32_i24 %2, sext(%3), sext(%4) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3 \n \
+        v_add3_u32 %0, %1, %2, %0 \n \
+        "
+        : "+v"(c), "=&v"(tmp1), "=&v"(tmp2)
+        : "v"(a), "v"(b)
+    );
+#else
+    const int8x4_t va = reinterpret_cast<const int8x4_t&>(a);
+    const int8x4_t vb = reinterpret_cast<const int8x4_t&>(b);
+    c += va[0] * vb[0] + va[1] * vb[1] + va[2] * vb[2] + va[3] * vb[3];
+#endif
+    return c;
+
+#else // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+
+#if __CUDA_ARCH__ >= MIN_CC_DP4A
+    return __dp4a(a, b, c);
+#else // __CUDA_ARCH__ >= MIN_CC_DP4A
+    const int8_t * a8 = (const int8_t *) &a;
+    const int8_t * b8 = (const int8_t *) &b;
+    return c + a8[0]*b8[0] + a8[1]*b8[1] + a8[2]*b8[2] + a8[3]*b8[3];
+#endif // __CUDA_ARCH__ >= MIN_CC_DP4A
+
+#endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
+}
+
+// TODO: move to ggml-common.h
+static constexpr __device__ int8_t kvalues_iq4nl[16] = {-127, -104, -83, -65, -49, -35, -22, -10, 1, 13, 25, 38, 53, 69, 89, 113};
+
+typedef void (*dequantize_kernel_t)(const void * vx, const int64_t ib, const int iqs, dfloat2 & v);
+
+static __device__ __forceinline__ float get_alibi_slope(
+    const float max_bias, const uint32_t h, const uint32_t n_head_log2, const float m0, const float m1
+) {
+    if (max_bias <= 0.0f) {
+        return 1.0f;
+    }
+    const float base = h < n_head_log2 ? m0 : m1;
+    const int   exph = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1;
+
+    return powf(base, exph);
+}
+
+template <ggml_type type>
+struct ggml_cuda_type_traits;
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_F16> {
+    static constexpr int qk = 1;
+    static constexpr int qr = 1;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q4_0> {
+    static constexpr int qk = QK4_0;
+    static constexpr int qr = QR4_0;
+    static constexpr int qi = QI4_0;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q4_1> {
+    static constexpr int qk = QK4_1;
+    static constexpr int qr = QR4_1;
+    static constexpr int qi = QI4_1;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q5_0> {
+    static constexpr int qk = QK5_0;
+    static constexpr int qr = QR5_0;
+    static constexpr int qi = QI5_0;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q5_1> {
+    static constexpr int qk = QK5_1;
+    static constexpr int qr = QR5_1;
+    static constexpr int qi = QI5_1;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q8_0> {
+    static constexpr int qk = QK8_0;
+    static constexpr int qr = QR8_0;
+    static constexpr int qi = QI8_0;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q2_K> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR2_K;
+    static constexpr int qi = QI2_K;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q3_K> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR3_K;
+    static constexpr int qi = QI3_K;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q4_K> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR4_K;
+    static constexpr int qi = QI4_K;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q5_K> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR5_K;
+    static constexpr int qi = QI5_K;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_Q6_K> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR6_K;
+    static constexpr int qi = QI6_K;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ2_XXS> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR2_XXS;
+    static constexpr int qi = QI2_XXS;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ2_XS> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR2_XS;
+    static constexpr int qi = QI2_XS;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ2_S> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR2_S;
+    static constexpr int qi = QI2_S;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ3_XXS> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR3_XXS;
+    static constexpr int qi = QI3_XXS;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ1_S> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR1_S;
+    static constexpr int qi = QI1_S;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ1_M> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR1_M;
+    static constexpr int qi = QI1_M;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ4_NL> {
+    static constexpr int qk = QK4_NL;
+    static constexpr int qr = QR4_NL;
+    static constexpr int qi = QI4_NL;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ4_XS> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR4_XS;
+    static constexpr int qi = QI4_XS;
+};
+
+template<>
+struct ggml_cuda_type_traits<GGML_TYPE_IQ3_S> {
+    static constexpr int qk = QK_K;
+    static constexpr int qr = QR3_S;
+    static constexpr int qi = QI3_S;
+};
+
+//////////////////////
+
+struct ggml_cuda_device_info {
+    int device_count;
+
+    struct cuda_device_info {
+        int     cc;                 // compute capability
+        int     nsm;                // number of streaming multiprocessors
+        size_t  smpb;               // max. shared memory per block
+        size_t  smpbo;              // max. shared memory per block (with opt-in)
+        bool    vmm;                // virtual memory support
+        size_t  vmm_granularity;    // granularity of virtual memory
+        size_t  total_vram;
+    };
+
+    cuda_device_info devices[GGML_CUDA_MAX_DEVICES] = {};
+
+    std::array<float, GGML_CUDA_MAX_DEVICES> default_tensor_split = {};
+};
+
+const ggml_cuda_device_info & ggml_cuda_info();
+
+void ggml_cuda_set_device(int device);
+int ggml_cuda_get_device();
+
+struct ggml_cuda_pool {
+    virtual ~ggml_cuda_pool() = default;
+
+    virtual void * alloc(size_t size, size_t * actual_size) = 0;
+    virtual void free(void * ptr, size_t size) = 0;
+};
+
+template<typename T>
+struct ggml_cuda_pool_alloc {
+    ggml_cuda_pool * pool = nullptr;
+    T * ptr = nullptr;
+    size_t actual_size = 0;
+
+    ggml_cuda_pool_alloc() = default;
+
+    explicit ggml_cuda_pool_alloc(ggml_cuda_pool & pool) : pool(&pool) {
+    }
+
+    ggml_cuda_pool_alloc(ggml_cuda_pool & pool, size_t size) : pool(&pool) {
+        alloc(size);
+    }
+
+    ~ggml_cuda_pool_alloc() {
+        if (ptr != nullptr) {
+            pool->free(ptr, actual_size);
+        }
+    }
+
+    // size is in number of elements
+    T * alloc(size_t size) {
+        GGML_ASSERT(pool != nullptr);
+        GGML_ASSERT(ptr == nullptr);
+        ptr = (T *) pool->alloc(size * sizeof(T), &this->actual_size);
+        return ptr;
+    }
+
+    T * alloc(ggml_cuda_pool & pool, size_t size) {
+        this->pool = &pool;
+        return alloc(size);
+    }
+
+    T * get() {
+        return ptr;
+    }
+
+    ggml_cuda_pool_alloc(const ggml_cuda_pool_alloc &) = delete;
+    ggml_cuda_pool_alloc(ggml_cuda_pool_alloc &&) = delete;
+    ggml_cuda_pool_alloc& operator=(const ggml_cuda_pool_alloc &) = delete;
+    ggml_cuda_pool_alloc& operator=(ggml_cuda_pool_alloc &&) = delete;
+};
+
+
+// backend interface
+
+struct ggml_tensor_extra_gpu {
+    void * data_device[GGML_CUDA_MAX_DEVICES]; // 1 pointer for each device for split tensors
+    cudaEvent_t events[GGML_CUDA_MAX_DEVICES][GGML_CUDA_MAX_STREAMS]; // events for synchronizing multiple GPUs
+};
+
+
+#if (CUDART_VERSION >= 12000) && defined(GGML_CUDA_USE_GRAPHS)
+#define USE_CUDA_GRAPH
+#endif
+
+struct ggml_graph_node_properties {
+    void * node_address;
+    ggml_op node_op;
+    int64_t ne[GGML_MAX_DIMS];
+    size_t nb[GGML_MAX_DIMS];
+    void * src_address[GGML_MAX_SRC];
+};
+
+struct ggml_cuda_graph {
+#ifdef USE_CUDA_GRAPH
+    ~ggml_cuda_graph() {
+        if (instance != nullptr) {
+            CUDA_CHECK(cudaGraphExecDestroy(instance));
+        }
+        if (graph != nullptr) {
+            CUDA_CHECK(cudaGraphDestroy(graph));
+        }
+    }
+    cudaGraph_t graph = nullptr;
+    cudaGraphExec_t instance = nullptr;
+    size_t num_nodes = 0;
+    std::vector<cudaGraphNode_t> nodes;
+    std::vector<cudaKernelNodeParams> params;
+    bool disable_due_to_gpu_arch = false;
+    bool disable_due_to_too_many_updates = false;
+    bool disable_due_to_failed_graph_capture = false;
+    int number_consecutive_updates = 0;
+    std::vector<ggml_graph_node_properties> ggml_graph_properties;
+    std::vector<char **> updated_kernel_arg;
+#endif
+};
+
+struct ggml_backend_cuda_context {
+    int device;
+    std::string name;
+    cudaEvent_t copy_event = nullptr;
+
+    cudaStream_t streams[GGML_CUDA_MAX_DEVICES][GGML_CUDA_MAX_STREAMS] = { { nullptr } };
+    cublasHandle_t cublas_handles[GGML_CUDA_MAX_DEVICES] = {nullptr};
+
+    std::unique_ptr<ggml_cuda_graph> cuda_graph;
+
+    explicit ggml_backend_cuda_context(int device) :
+        device(device),
+        name(GGML_CUDA_NAME + std::to_string(device)) {
+    }
+
+    ~ggml_backend_cuda_context() {
+        if (copy_event != nullptr) {
+            CUDA_CHECK(cudaEventDestroy(copy_event));
+        }
+        for (int i = 0; i < GGML_CUDA_MAX_DEVICES; ++i) {
+            for (int j = 0; j < GGML_CUDA_MAX_STREAMS; ++j) {
+                if (streams[i][j] != nullptr) {
+                    CUDA_CHECK(cudaStreamDestroy(streams[i][j]));
+                }
+            }
+            if (cublas_handles[i] != nullptr) {
+                CUBLAS_CHECK(cublasDestroy(cublas_handles[i]));
+            }
+        }
+    }
+
+    cudaStream_t stream(int device, int stream) {
+        if (streams[device][stream] == nullptr) {
+            ggml_cuda_set_device(device);
+            CUDA_CHECK(cudaStreamCreateWithFlags(&streams[device][stream], cudaStreamNonBlocking));
+        }
+        return streams[device][stream];
+    }
+
+    cudaStream_t stream() {
+        return stream(device, 0);
+    }
+
+    cublasHandle_t cublas_handle(int device) {
+        if (cublas_handles[device] == nullptr) {
+            ggml_cuda_set_device(device);
+            CUBLAS_CHECK(cublasCreate(&cublas_handles[device]));
+            CUBLAS_CHECK(cublasSetMathMode(cublas_handles[device], CUBLAS_TF32_TENSOR_OP_MATH));
+        }
+        return cublas_handles[device];
+    }
+
+    cublasHandle_t cublas_handle() {
+        return cublas_handle(device);
+    }
+
+    // pool
+    std::unique_ptr<ggml_cuda_pool> pools[GGML_CUDA_MAX_DEVICES];
+
+    static std::unique_ptr<ggml_cuda_pool> new_pool_for_device(int device);
+
+    ggml_cuda_pool & pool(int device) {
+        if (pools[device] == nullptr) {
+            pools[device] = new_pool_for_device(device);
+        }
+        return *pools[device];
+    }
+
+    ggml_cuda_pool & pool() {
+        return pool(device);
+    }
+};
--- a/llama/ggml-cuda/concat.cu
+++ b/llama/ggml-cuda/concat.cu
@@ -0,0 +1,222 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "concat.cuh"
+
+// contiguous kernels
+static __global__ void concat_f32_dim0(const float * x, const float * y, float * dst, const int ne0, const int ne00) {
+    int nidx = threadIdx.x + blockIdx.x * blockDim.x;
+    if (nidx >= ne0) {
+        return;
+    }
+
+    int offset_dst =
+        nidx +
+        blockIdx.y * ne0 +
+        blockIdx.z * ne0 * gridDim.y;
+
+    if (nidx < ne00) { // src0
+        int offset_src =
+            nidx +
+            blockIdx.y * ne00 +
+            blockIdx.z * ne00 * gridDim.y;
+        dst[offset_dst] = x[offset_src];
+    } else {
+        int offset_src =
+            (nidx - ne00) +
+            blockIdx.y * (ne0 - ne00) +
+            blockIdx.z * (ne0 - ne00) * gridDim.y;
+        dst[offset_dst] = y[offset_src];
+    }
+}
+
+static __global__ void concat_f32_dim1(const float * x, const float * y, float * dst, const int ne0, const int ne01) {
+    int nidx = threadIdx.x + blockIdx.x * blockDim.x;
+    if (nidx >= ne0) {
+        return;
+    }
+
+    int offset_dst =
+        nidx +
+        blockIdx.y * ne0 +
+        blockIdx.z * ne0 * gridDim.y;
+
+    if (blockIdx.y < ne01) { // src0
+        int offset_src =
+            nidx +
+            blockIdx.y * ne0 +
+            blockIdx.z * ne0 * ne01;
+        dst[offset_dst] = x[offset_src];
+    } else {
+        int offset_src =
+            nidx +
+            (blockIdx.y - ne01) * ne0 +
+            blockIdx.z * ne0 * (gridDim.y - ne01);
+        dst[offset_dst] = y[offset_src];
+    }
+}
+
+static __global__ void concat_f32_dim2(const float * x, const float * y, float * dst, const int ne0, const int ne02) {
+    int nidx = threadIdx.x + blockIdx.x * blockDim.x;
+    if (nidx >= ne0) {
+        return;
+    }
+
+    int offset_dst =
+        nidx +
+        blockIdx.y * ne0 +
+        blockIdx.z * ne0 * gridDim.y;
+
+    if (blockIdx.z < ne02) { // src0
+        int offset_src =
+            nidx +
+            blockIdx.y * ne0 +
+            blockIdx.z * ne0 * gridDim.y;
+        dst[offset_dst] = x[offset_src];
+    } else {
+        int offset_src =
+            nidx +
+            blockIdx.y * ne0 +
+            (blockIdx.z - ne02) * ne0 *  gridDim.y;
+        dst[offset_dst] = y[offset_src];
+    }
+}
+
+static void concat_f32_cuda(const float * x, const float * y, float * dst, int ne00, int ne01, int ne02, int ne0, int ne1, int ne2, int dim, cudaStream_t stream) {
+    int num_blocks = (ne0 + CUDA_CONCAT_BLOCK_SIZE - 1) / CUDA_CONCAT_BLOCK_SIZE;
+    dim3 gridDim(num_blocks, ne1, ne2);
+    if (dim == 0) {
+        concat_f32_dim0<<<gridDim, CUDA_CONCAT_BLOCK_SIZE, 0, stream>>>(x, y, dst, ne0, ne00);
+        return;
+    }
+    if (dim == 1) {
+        concat_f32_dim1<<<gridDim, CUDA_CONCAT_BLOCK_SIZE, 0, stream>>>(x, y, dst, ne0, ne01);
+        return;
+    }
+    concat_f32_dim2<<<gridDim, CUDA_CONCAT_BLOCK_SIZE, 0, stream>>>(x, y, dst, ne0, ne02);
+}
+
+// non-contiguous kernel (slow)
+static __global__ void concat_f32_non_cont(
+        const char * src0,
+        const char * src1,
+              char * dst,
+           int64_t   ne00,
+           int64_t   ne01,
+           int64_t   ne02,
+           int64_t   ne03,
+          uint64_t   nb00,
+          uint64_t   nb01,
+          uint64_t   nb02,
+          uint64_t   nb03,
+           int64_t /*ne10*/,
+           int64_t /*ne11*/,
+           int64_t /*ne12*/,
+           int64_t /*ne13*/,
+          uint64_t   nb10,
+          uint64_t   nb11,
+          uint64_t   nb12,
+          uint64_t   nb13,
+           int64_t   ne0,
+           int64_t /*ne1*/,
+           int64_t /*ne2*/,
+           int64_t /*ne3*/,
+          uint64_t   nb0,
+          uint64_t   nb1,
+          uint64_t   nb2,
+          uint64_t   nb3,
+          int32_t   dim) {
+    const int64_t i3 = blockIdx.z;
+    const int64_t i2 = blockIdx.y;
+    const int64_t i1 = blockIdx.x;
+
+    int64_t o[4] = {0, 0, 0, 0};
+    o[dim] = dim == 0 ? ne00 : (dim == 1 ? ne01 : (dim == 2 ? ne02 : ne03));
+
+    const float * x;
+
+    for (int i0 = threadIdx.x; i0 < ne0; i0 += blockDim.x) {
+        if (i0 < ne00 && i1 < ne01 && i2 < ne02 && i3 < ne03) {
+            x = (const float *)(src0 + (i3       )*nb03 + (i2       )*nb02 + (i1       )*nb01 + (i0       )*nb00);
+        } else {
+            x = (const float *)(src1 + (i3 - o[3])*nb13 + (i2 - o[2])*nb12 + (i1 - o[1])*nb11 + (i0 - o[0])*nb10);
+        }
+
+        float * y = (float *)(dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
+
+        *y = *x;
+    }
+}
+
+
+void ggml_cuda_op_concat(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const ggml_tensor * src1 = dst->src[1];
+
+    cudaStream_t stream = ctx.stream();
+
+    const int32_t dim = ((int32_t *) dst->op_params)[0];
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT(src1->type == GGML_TYPE_F32);
+    GGML_ASSERT(dst->type  == GGML_TYPE_F32);
+
+    if (ggml_is_contiguous(src0) && ggml_is_contiguous(src1)) {
+        const float * src0_d = (const float *)src0->data;
+        const float * src1_d = (const float *)src1->data;
+
+        float * dst_d = (float *)dst->data;
+
+        if (dim != 3) {
+            for (int i3 = 0; i3 < dst->ne[3]; i3++) {
+                concat_f32_cuda(
+                        src0_d + i3 * (src0->nb[3] / 4),
+                        src1_d + i3 * (src1->nb[3] / 4),
+                        dst_d + i3 * ( dst->nb[3] / 4),
+                        src0->ne[0], src0->ne[1], src0->ne[2],
+                        dst->ne[0],  dst->ne[1],  dst->ne[2], dim, stream);
+            }
+        } else {
+            const size_t size0 = ggml_nbytes(src0);
+            const size_t size1 = ggml_nbytes(src1);
+
+            CUDA_CHECK(cudaMemcpyAsync(dst_d,           src0_d, size0, cudaMemcpyDeviceToDevice, stream));
+            CUDA_CHECK(cudaMemcpyAsync(dst_d + size0/4, src1_d, size1, cudaMemcpyDeviceToDevice, stream));
+        }
+    } else {
+        dim3 grid_dim(dst->ne[1], dst->ne[2], dst->ne[3]);
+        concat_f32_non_cont<<<grid_dim, CUDA_CONCAT_BLOCK_SIZE, 0, stream>>>(
+                (const char *)src0->data,
+                (const char *)src1->data,
+                (      char *)dst->data,
+                src0->ne[0], src0->ne[1], src0->ne[2], src0->ne[3],
+                src0->nb[0], src0->nb[1], src0->nb[2], src0->nb[3],
+                src1->ne[0], src1->ne[1], src1->ne[2], src1->ne[3],
+                src1->nb[0], src1->nb[1], src1->nb[2], src1->nb[3],
+                dst->ne[0],  dst->ne[1],  dst->ne[2],  dst->ne[3],
+                dst->nb[0],  dst->nb[1],  dst->nb[2],  dst->nb[3], dim);
+    }
+}
--- a/llama/ggml-cuda/concat.cuh
+++ b/llama/ggml-cuda/concat.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_CONCAT_BLOCK_SIZE 256
+
+void ggml_cuda_op_concat(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/conv-transpose-1d.cu
+++ b/llama/ggml-cuda/conv-transpose-1d.cu
@@ -0,0 +1,113 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "conv-transpose-1d.cuh"
+
+static  __global__ void conv_transpose_1d_kernel(
+        const int s0, const int p0, const int d0, const int output_size,
+        const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
+        const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
+        const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
+        const float * src0, const float * src1,  float * dst) {
+    int global_index = threadIdx.x + blockIdx.x * blockDim.x;
+    if (global_index >= output_size) {
+        return;
+    }
+
+    int out_index = global_index / dst_ne0;
+
+    float accumulator = 0;
+
+    for (int c = 0; c < src0_ne2; c++) {
+        int idx = global_index % dst_ne0;
+
+        int kernel_offset = (src0_ne0 * src0_ne1 * c) + (out_index * src0_ne0);
+        int input_offset = src1_ne0 * c;
+
+        for (int i = 0; i < src1_ne0; i++) {
+            if (!(idx >= i*s0 && idx < i*s0 + src0_ne0)) {
+                continue;
+            }
+            int weight_idx = idx - i*s0;
+
+            float kernel_weight = src0[kernel_offset + weight_idx];
+            float input_value =  src1[input_offset+i];
+
+            accumulator += kernel_weight * input_value;
+        }
+    }
+    dst[global_index] = accumulator;
+}
+
+static void conv_transpose_1d_f32_f32_cuda(
+        const int s0, const int p0, const int d0, const int output_size,
+        const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
+        const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
+        const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
+        const float * src0, const float * src1,  float * dst,
+        cudaStream_t stream) {
+
+    const int num_blocks = (output_size + CUDA_CONV_TRANPOSE_1D_BLOCK_SIZE - 1) / CUDA_CONV_TRANPOSE_1D_BLOCK_SIZE;
+    conv_transpose_1d_kernel<<<num_blocks,CUDA_CONV_TRANPOSE_1D_BLOCK_SIZE, 0, stream>>>(
+        s0,p0,d0,output_size,
+        src0_ne0, src0_ne1,  src0_ne2, src0_ne3,
+        src1_ne0, src1_ne1,  src1_ne2, src1_ne3,
+        dst_ne0,  dst_ne1,   dst_ne2,  dst_ne3,
+        src0,src1, dst);
+}
+
+void ggml_cuda_op_conv_transpose_1d(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const float * src0_d = (const float *)src0->data;
+
+    const ggml_tensor * src1 = dst->src[1];
+    const float * src1_d = (const float *)src1->data;
+
+    float * dst_d = (float *)dst->data;
+    cudaStream_t stream = ctx.stream();
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT( dst->type == GGML_TYPE_F32);
+
+    GGML_ASSERT(ggml_is_contiguous(src0));
+    GGML_ASSERT(ggml_is_contiguous(src1));
+
+    const int32_t * opts = (const int32_t *)dst->op_params;
+
+    const int s0 = opts[0];
+    const int p0 = 0;//opts[3];
+    const int d0 = 1;//opts[4];
+
+    const int64_t kernel_size = ggml_nelements(src0);
+    const int64_t input_size = ggml_nelements(src1);
+    const int64_t output_size = ggml_nelements(dst);
+
+    conv_transpose_1d_f32_f32_cuda(s0, p0, d0, output_size,
+        src0->ne[0], src0->ne[1], src0->ne[2], src0->ne[3],
+        src1->ne[0], src1->ne[1], src1->ne[2], src1->ne[3],
+        dst->ne[0], dst->ne[1], dst->ne[2], dst->ne[3],
+        src0_d, src1_d, dst_d, stream);
+}
--- a/llama/ggml-cuda/conv-transpose-1d.cuh
+++ b/llama/ggml-cuda/conv-transpose-1d.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_CONV_TRANPOSE_1D_BLOCK_SIZE 256
+
+void ggml_cuda_op_conv_transpose_1d(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/convert.cu
+++ b/llama/ggml-cuda/convert.cu
@@ -0,0 +1,712 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "convert.cuh"
+#include "dequantize.cuh"
+
+#define CUDA_Q8_0_NE_ALIGN 2048
+
+template <int qk, int qr, dequantize_kernel_t dequantize_kernel, typename dst_t>
+static __global__ void dequantize_block(const void * __restrict__ vx, dst_t * __restrict__ y, const int64_t k) {
+    const int64_t i = (int64_t)2*(blockDim.x*blockIdx.x + threadIdx.x);
+
+    if (i >= k) {
+        return;
+    }
+
+    const int64_t ib = i/qk; // block index
+    const int64_t iqs = (i%qk)/qr; // quant index
+    const int64_t iybs = i - i%qk; // y block start index
+    const int64_t y_offset = qr == 1 ? 1 : qk/2;
+
+    // dequantize
+    dfloat2 v;
+    dequantize_kernel(vx, ib, iqs, v);
+
+    y[iybs + iqs + 0]        = v.x;
+    y[iybs + iqs + y_offset] = v.y;
+}
+
+template <bool need_check>
+static __global__ void dequantize_block_q8_0_f16(const void * __restrict__ vx, half * __restrict__ y, const int64_t k) {
+#if __CUDA_ARCH__ >= CC_PASCAL
+    constexpr int nint = CUDA_Q8_0_NE_ALIGN/sizeof(int) + WARP_SIZE;
+
+    const int64_t   i0 = CUDA_Q8_0_NE_ALIGN*blockIdx.x;
+    const int * x0 = ((int *) vx) + blockIdx.x * nint;
+    half2 * y2 = (half2 *) (y + i0);
+
+    __shared__ int vals[nint];
+
+#pragma unroll
+    for (int ix0 = 0; ix0 < nint; ix0 += WARP_SIZE) {
+        if (need_check && i0*sizeof(block_q8_0)/QK8_0 + sizeof(int)*(ix0 + threadIdx.x) >= k*sizeof(block_q8_0)/QK8_0) {
+            break;
+        }
+
+        const int ix = ix0 + threadIdx.x;
+        vals[ix] = x0[ix];
+    }
+
+    __syncthreads();
+
+#pragma unroll
+    for (int iy = 0; iy < CUDA_Q8_0_NE_ALIGN; iy += 2*WARP_SIZE) {
+        if (need_check && i0 + iy + 2*threadIdx.x >= k) {
+            return;
+        }
+
+        const half * b0 = ((const half  *) vals) + (sizeof(block_q8_0)/sizeof(half)) * ((iy + 2*threadIdx.x)/QK8_0);
+        const half    d = *b0;
+        const char2  qs = ((const char2 *) (b0 + 1))[threadIdx.x % (QK8_0/2)];
+
+        y2[iy/2 + threadIdx.x] = __hmul2(make_half2(qs.x, qs.y), __half2half2(d));
+    }
+#else
+    GGML_UNUSED(vx);
+    GGML_UNUSED(y);
+    GGML_UNUSED(k);
+    NO_DEVICE_CODE;
+#endif // __CUDA_ARCH__ >= CC_PASCAL
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_q4_0(const void * __restrict__ vx, dst_t * __restrict__ yy, int nb32) {
+
+    const int64_t i = blockIdx.x;
+
+    // assume 32 threads
+    const int64_t tid = threadIdx.x;
+    const int64_t il  = tid/8;
+    const int64_t ir  = tid%8;
+    const int64_t ib = 8*i + ir;
+    if (ib >= nb32) {
+        return;
+    }
+
+    dst_t * y = yy + 256*i + 32*ir + 4*il;
+
+    const block_q4_0 * x = (const block_q4_0 *)vx + ib;
+    const float d = __half2float(x->d);
+    const float dm = -8*d;
+
+    const uint8_t * q = x->qs + 4*il;
+
+    for (int l = 0; l < 4; ++l) {
+        y[l+ 0] = d * (q[l] & 0xF) + dm;
+        y[l+16] = d * (q[l] >>  4) + dm;
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_q4_1(const void * __restrict__ vx, dst_t * __restrict__ yy, int nb32) {
+
+    const int64_t i = blockIdx.x;
+
+    // assume 32 threads
+    const int64_t tid = threadIdx.x;
+    const int64_t il  = tid/8;
+    const int64_t ir  = tid%8;
+    const int64_t ib = 8*i + ir;
+    if (ib >= nb32) {
+        return;
+    }
+
+    dst_t * y = yy + 256*i + 32*ir + 4*il;
+
+    const block_q4_1 * x = (const block_q4_1 *)vx + ib;
+    const float2 d = __half22float2(x->dm);
+
+    const uint8_t * q = x->qs + 4*il;
+
+    for (int l = 0; l < 4; ++l) {
+        y[l+ 0] = d.x * (q[l] & 0xF) + d.y;
+        y[l+16] = d.x * (q[l] >>  4) + d.y;
+    }
+}
+
+//================================== k-quants
+
+template<typename dst_t>
+static __global__ void dequantize_block_q2_K(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_q2_K * x = (const block_q2_K *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t n   = tid/32;
+    const int64_t l   = tid - 32*n;
+    const int64_t is  = 8*n + l/16;
+
+    const uint8_t q = x[i].qs[32*n + l];
+    dst_t * y = yy + i*QK_K + 128*n;
+
+    float dall = __low2half(x[i].dm);
+    float dmin = __high2half(x[i].dm);
+    y[l+ 0] = dall * (x[i].scales[is+0] & 0xF) * ((q >> 0) & 3) - dmin * (x[i].scales[is+0] >> 4);
+    y[l+32] = dall * (x[i].scales[is+2] & 0xF) * ((q >> 2) & 3) - dmin * (x[i].scales[is+2] >> 4);
+    y[l+64] = dall * (x[i].scales[is+4] & 0xF) * ((q >> 4) & 3) - dmin * (x[i].scales[is+4] >> 4);
+    y[l+96] = dall * (x[i].scales[is+6] & 0xF) * ((q >> 6) & 3) - dmin * (x[i].scales[is+6] >> 4);
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_q3_K(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i = blockIdx.x;
+    const block_q3_K * x = (const block_q3_K *) vx;
+
+    const int64_t r = threadIdx.x/4;
+    const int64_t tid = r/2;
+    const int64_t is0 = r%2;
+    const int64_t l0 = 16*is0 + 4*(threadIdx.x%4);
+    const int64_t n = tid / 4;
+    const int64_t j = tid - 4*n;
+
+    uint8_t m = 1 << (4*n + j);
+    int64_t is = 8*n + 2*j + is0;
+    int shift = 2*j;
+
+    int8_t us = is <  4 ? (x[i].scales[is-0] & 0xF) | (((x[i].scales[is+8] >> 0) & 3) << 4) :
+                is <  8 ? (x[i].scales[is-0] & 0xF) | (((x[i].scales[is+4] >> 2) & 3) << 4) :
+                is < 12 ? (x[i].scales[is-8] >>  4) | (((x[i].scales[is+0] >> 4) & 3) << 4) :
+                          (x[i].scales[is-8] >>  4) | (((x[i].scales[is-4] >> 6) & 3) << 4);
+    float d_all = x[i].d;
+    float dl = d_all * (us - 32);
+
+    dst_t * y = yy + i*QK_K + 128*n + 32*j;
+    const uint8_t * q = x[i].qs + 32*n;
+    const uint8_t * hm = x[i].hmask;
+
+    for (int l = l0; l < l0+4; ++l) y[l] = dl * ((int8_t)((q[l] >> shift) & 3) - ((hm[l] & m) ? 0 : 4));
+}
+
+static inline __device__ void get_scale_min_k4(int j, const uint8_t * q, uint8_t & d, uint8_t & m) {
+    if (j < 4) {
+        d = q[j] & 63; m = q[j + 4] & 63;
+    } else {
+        d = (q[j+4] & 0xF) | ((q[j-4] >> 6) << 4);
+        m = (q[j+4] >>  4) | ((q[j-0] >> 6) << 4);
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_q4_K(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+    const block_q4_K * x = (const block_q4_K *) vx;
+
+    const int64_t i = blockIdx.x;
+
+    // assume 32 threads
+    const int64_t tid = threadIdx.x;
+    const int64_t il  = tid/8;
+    const int64_t ir  = tid%8;
+    const int64_t is  = 2*il;
+    const int64_t n   = 4;
+
+    dst_t * y = yy + i*QK_K + 64*il + n*ir;
+
+    const float dall = __low2half(x[i].dm);
+    const float dmin = __high2half(x[i].dm);
+
+    const uint8_t * q = x[i].qs + 32*il + n*ir;
+
+    uint8_t sc, m;
+    get_scale_min_k4(is + 0, x[i].scales, sc, m);
+    const float d1 = dall * sc; const float m1 = dmin * m;
+    get_scale_min_k4(is + 1, x[i].scales, sc, m);
+    const float d2 = dall * sc; const float m2 = dmin * m;
+    for (int l = 0; l < n; ++l) {
+        y[l + 0] = d1 * (q[l] & 0xF) - m1;
+        y[l +32] = d2 * (q[l] >>  4) - m2;
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_q5_K(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+    const block_q5_K * x = (const block_q5_K *) vx;
+
+    const int64_t i = blockIdx.x;
+
+    // assume 64 threads - this is very slightly better than the one below
+    const int64_t tid = threadIdx.x;
+    const int64_t il  = tid/16;   // il is in 0...3
+    const int64_t ir  = tid%16;   // ir is in 0...15
+    const int64_t is  = 2*il;     // is is in 0...6
+
+    dst_t * y = yy + i*QK_K + 64*il + 2*ir;
+
+    const float dall = __low2half(x[i].dm);
+    const float dmin = __high2half(x[i].dm);
+
+    const uint8_t * ql = x[i].qs + 32*il + 2*ir;
+    const uint8_t * qh = x[i].qh + 2*ir;
+
+    uint8_t sc, m;
+    get_scale_min_k4(is + 0, x[i].scales, sc, m);
+    const float d1 = dall * sc; const float m1 = dmin * m;
+    get_scale_min_k4(is + 1, x[i].scales, sc, m);
+    const float d2 = dall * sc; const float m2 = dmin * m;
+
+    uint8_t   hm  = 1 << (2*il);
+    y[ 0] = d1 * ((ql[ 0] & 0xF) + (qh[ 0] & hm ? 16 : 0)) - m1;
+    y[ 1] = d1 * ((ql[ 1] & 0xF) + (qh[ 1] & hm ? 16 : 0)) - m1;
+    hm <<= 1;
+    y[32] = d2 * ((ql[ 0] >>  4) + (qh[ 0] & hm ? 16 : 0)) - m2;
+    y[33] = d2 * ((ql[ 1] >>  4) + (qh[ 1] & hm ? 16 : 0)) - m2;
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_q6_K(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+    const block_q6_K * x = (const block_q6_K *) vx;
+
+    const int64_t i = blockIdx.x;
+
+    // assume 64 threads - this is very slightly better than the one below
+    const int64_t tid = threadIdx.x;
+    const int64_t ip  = tid/32;   // ip is 0 or 1
+    const int64_t il  = tid - 32*ip; // 0...32
+    const int64_t is  = 8*ip + il/16;
+
+    dst_t * y = yy + i*QK_K + 128*ip + il;
+
+    const float d = x[i].d;
+
+    const uint8_t * ql = x[i].ql + 64*ip + il;
+    const uint8_t   qh = x[i].qh[32*ip + il];
+    const int8_t  * sc = x[i].scales + is;
+
+    y[ 0] = d * sc[0] * ((int8_t)((ql[ 0] & 0xF) | (((qh >> 0) & 3) << 4)) - 32);
+    y[32] = d * sc[2] * ((int8_t)((ql[32] & 0xF) | (((qh >> 2) & 3) << 4)) - 32);
+    y[64] = d * sc[4] * ((int8_t)((ql[ 0]  >> 4) | (((qh >> 4) & 3) << 4)) - 32);
+    y[96] = d * sc[6] * ((int8_t)((ql[32]  >> 4) | (((qh >> 6) & 3) << 4)) - 32);
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq2_xxs(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq2_xxs * x = (const block_iq2_xxs  *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const uint16_t * q2 = x[i].qs + 4*ib;
+    const uint8_t  * aux8 = (const uint8_t *)q2;
+    const uint8_t  * grid = (const uint8_t *)(iq2xxs_grid + aux8[il]);
+    const uint32_t aux32 = q2[2] | (q2[3] << 16);
+    const float d = (float)x[i].d * (0.5f + (aux32 >> 28)) * 0.25f;
+    const uint8_t signs = ksigns_iq2xs[(aux32 >> 7*il) & 127];
+    for (int j = 0; j < 8; ++j) y[j] = d * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f);
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq2_xs(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq2_xs * x = (const block_iq2_xs *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const uint16_t * q2 = x[i].qs + 4*ib;
+    const uint8_t  * grid = (const uint8_t *)(iq2xs_grid + (q2[il] & 511));
+    const float d = (float)x[i].d * (0.5f + ((x[i].scales[ib] >> 4*(il/2)) & 0xf)) * 0.25f;
+    const uint8_t signs = ksigns_iq2xs[q2[il] >> 9];
+    for (int j = 0; j < 8; ++j) y[j] = d * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f);
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq2_s(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq2_s * x = (const block_iq2_s *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const uint8_t * grid = (const uint8_t *)(iq2s_grid + (x[i].qs[4*ib+il] | ((x[i].qh[ib] << (8-2*il)) & 0x300)));
+    const float d = (float)x[i].d * (0.5f + ((x[i].scales[ib] >> 4*(il/2)) & 0xf)) * 0.25f;
+    const uint8_t signs = x[i].qs[QK_K/8+4*ib+il];
+    for (int j = 0; j < 8; ++j) y[j] = d * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f);
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq3_xxs(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq3_xxs * x = (const block_iq3_xxs  *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const uint8_t  * q3 = x[i].qs + 8*ib;
+    const uint16_t * gas = (const uint16_t *)(x[i].qs + QK_K/4) + 2*ib;
+    const uint8_t  * grid1 = (const uint8_t *)(iq3xxs_grid + q3[2*il+0]);
+    const uint8_t  * grid2 = (const uint8_t *)(iq3xxs_grid + q3[2*il+1]);
+    const uint32_t aux32 = gas[0] | (gas[1] << 16);
+    const float d = (float)x[i].d * (0.5f + (aux32 >> 28)) * 0.5f;
+    const uint8_t signs = ksigns_iq2xs[(aux32 >> 7*il) & 127];
+    for (int j = 0; j < 4; ++j) {
+        y[j+0] = d * grid1[j] * (signs & kmask_iq2xs[j+0] ? -1.f : 1.f);
+        y[j+4] = d * grid2[j] * (signs & kmask_iq2xs[j+4] ? -1.f : 1.f);
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq3_s(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq3_s * x = (const block_iq3_s *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const uint8_t * qs = x[i].qs + 8*ib;
+    const uint8_t * grid1 = (const uint8_t *)(iq3s_grid + (qs[2*il+0] | ((x[i].qh[ib] << (8-2*il)) & 256)));
+    const uint8_t * grid2 = (const uint8_t *)(iq3s_grid + (qs[2*il+1] | ((x[i].qh[ib] << (7-2*il)) & 256)));
+    const float d = (float)x[i].d * (1 + 2*((x[i].scales[ib/2] >> 4*(ib%2)) & 0xf));
+    const uint8_t signs = x[i].signs[4*ib + il];
+    for (int j = 0; j < 4; ++j) {
+        y[j+0] = d * grid1[j] * (signs & kmask_iq2xs[j+0] ? -1.f : 1.f);
+        y[j+4] = d * grid2[j] * (signs & kmask_iq2xs[j+4] ? -1.f : 1.f);
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq1_s(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq1_s * x = (const block_iq1_s  *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const float delta = x[i].qh[ib] & 0x8000 ? -1 - IQ1S_DELTA : -1 + IQ1S_DELTA;
+    const float d = (float)x[i].d * (2*((x[i].qh[ib] >> 12) & 7) + 1);
+    uint32_t grid32[2]; const int8_t * q = (const int8_t *)grid32;
+    grid32[0] = iq1s_grid_gpu[x[i].qs[4*ib+il] | (((x[i].qh[ib] >> 3*il) & 7) << 8)];
+    grid32[1] = (grid32[0] >> 4) & 0x0f0f0f0f;
+    grid32[0] &= 0x0f0f0f0f;
+    for (int j = 0; j < 8; ++j) {
+        y[j] = d * (q[j] + delta);
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq1_m(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq1_m * x = (const block_iq1_m  *) vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 8*il;
+    const uint16_t * sc = (const uint16_t *)x[i].scales;
+    iq1m_scale_t scale;
+    scale.u16 = (sc[0] >> 12) | ((sc[1] >> 8) & 0x00f0) | ((sc[2] >> 4) & 0x0f00) | (sc[3] & 0xf000);
+    const int64_t ib16 = 2*ib + il/2; // sc[ib16/4] >> 3*(ib16%4) -> sc[ib/2] >> 3*((2*ib+il/2)%4);
+    const float d = (float)scale.f16 * (2*((sc[ib16/4] >> 3*(ib16%4)) & 0x7) + 1);
+    const float delta = x[i].qh[2*ib+il/2] & (0x08 << 4*(il%2)) ? -1 - IQ1M_DELTA : -1 + IQ1M_DELTA;
+    uint32_t grid32[2]; const int8_t * q = (const int8_t *)grid32;
+    grid32[0] = iq1s_grid_gpu[x[i].qs[4*ib+il] | (((x[i].qh[2*ib+il/2] >> 4*(il%2)) & 7) << 8)];
+    grid32[1] = (grid32[0] >> 4) & 0x0f0f0f0f;
+    grid32[0] &= 0x0f0f0f0f;
+    for (int j = 0; j < 8; ++j) {
+        y[j] = d * (q[j] + delta);
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq4_nl(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+
+    const int64_t i   = blockIdx.x;
+    const block_iq4_nl * x = (const block_iq4_nl *) vx + i*(QK_K/QK4_NL);
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 4*il;
+    const uint8_t  * q4 = x[ib].qs + 4*il;
+    const float d = (float)x[ib].d;
+    for (int j = 0; j < 4; ++j) {
+        y[j+ 0] = d * kvalues_iq4nl[q4[j] & 0xf];
+        y[j+16] = d * kvalues_iq4nl[q4[j] >>  4];
+    }
+}
+
+template<typename dst_t>
+static __global__ void dequantize_block_iq4_xs(const void * __restrict__ vx, dst_t * __restrict__ yy) {
+    const int64_t i   = blockIdx.x;
+    const block_iq4_xs * x = (const block_iq4_xs *)vx;
+
+    const int64_t tid = threadIdx.x;
+    const int64_t il = tid/8; // 0...3
+    const int64_t ib = tid%8; // 0...7
+    dst_t * y = yy + i*QK_K + 32*ib + 4*il;
+    const uint8_t  * q4 = x[i].qs + 16*ib + 4*il;
+    const float d = (float)x[i].d * ((((x[i].scales_l[ib/2] >> 4*(ib%2)) & 0xf) | (((x[i].scales_h >> 2*ib) & 3) << 4)) - 32);
+    for (int j = 0; j < 4; ++j) {
+        y[j+ 0] = d * kvalues_iq4nl[q4[j] & 0xf];
+        y[j+16] = d * kvalues_iq4nl[q4[j] >>  4];
+    }
+}
+
+template <int qk, int qr, dequantize_kernel_t dequantize_kernel, typename dst_t>
+static void dequantize_block_cuda(const void * __restrict__ vx, dst_t * __restrict__ y, const int64_t k, cudaStream_t stream) {
+    const int num_blocks = (k + 2*CUDA_DEQUANTIZE_BLOCK_SIZE - 1) / (2*CUDA_DEQUANTIZE_BLOCK_SIZE);
+    dequantize_block<qk, qr, dequantize_kernel><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
+}
+
+static void dequantize_block_q8_0_f16_cuda(const void * __restrict__ vx, half * __restrict__ y, const int64_t k, cudaStream_t stream) {
+    const int num_blocks = (k + CUDA_Q8_0_NE_ALIGN - 1) / CUDA_Q8_0_NE_ALIGN;
+    if (k % CUDA_Q8_0_NE_ALIGN == 0) {
+        const bool need_check = false;
+        dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
+    } else {
+        const bool need_check = true;
+        dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
+    }
+}
+
+template<typename dst_t>
+static void dequantize_row_q2_K_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_q2_K<<<nb, 64, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_q3_K_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_q3_K<<<nb, 64, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_q4_0_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb32 = k / 32;
+    const int nb = (k + 255) / 256;
+    dequantize_block_q4_0<<<nb, 32, 0, stream>>>(vx, y, nb32);
+}
+
+template<typename dst_t>
+static void dequantize_row_q4_1_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb32 = k / 32;
+    const int nb = (k + 255) / 256;
+    dequantize_block_q4_1<<<nb, 32, 0, stream>>>(vx, y, nb32);
+}
+
+template<typename dst_t>
+static void dequantize_row_q4_K_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_q4_K<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_q5_K_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_q5_K<<<nb, 64, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_q6_K_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_q6_K<<<nb, 64, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq2_xxs_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq2_xxs<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq2_xs_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq2_xs<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq2_s_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq2_s<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq3_xxs_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq3_xxs<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq3_s_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq3_s<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq1_s_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq1_s<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq4_nl_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = (k + QK_K - 1) / QK_K;
+    dequantize_block_iq4_nl<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq1_m_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = k / QK_K;
+    dequantize_block_iq1_m<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template<typename dst_t>
+static void dequantize_row_iq4_xs_cuda(const void * vx, dst_t * y, const int64_t k, cudaStream_t stream) {
+    const int nb = (k + QK_K - 1) / QK_K;
+    dequantize_block_iq4_xs<<<nb, 32, 0, stream>>>(vx, y);
+}
+
+template <typename src_t, typename dst_t>
+static __global__ void convert_unary(const void * __restrict__ vx, dst_t * __restrict__ y, const int64_t k) {
+    const int64_t i = (int64_t)blockDim.x*blockIdx.x + threadIdx.x;
+
+    if (i >= k) {
+        return;
+    }
+
+    const src_t * x = (src_t *) vx;
+
+    y[i] = x[i];
+}
+
+template <typename src_t, typename dst_t>
+static void convert_unary_cuda(const void * __restrict__ vx, dst_t * __restrict__ y, const int64_t k, cudaStream_t stream) {
+    const int num_blocks = (k + CUDA_DEQUANTIZE_BLOCK_SIZE - 1) / CUDA_DEQUANTIZE_BLOCK_SIZE;
+    convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
+}
+
+to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_Q4_0:
+            return dequantize_row_q4_0_cuda;
+        case GGML_TYPE_Q4_1:
+            return dequantize_row_q4_1_cuda;
+        case GGML_TYPE_Q5_0:
+            return dequantize_block_cuda<QK5_0, QR5_0, dequantize_q5_0>;
+        case GGML_TYPE_Q5_1:
+            return dequantize_block_cuda<QK5_1, QR5_1, dequantize_q5_1>;
+        case GGML_TYPE_Q8_0:
+            if (ggml_cuda_info().devices[ggml_cuda_get_device()].cc >= CC_PASCAL) {
+                return dequantize_block_q8_0_f16_cuda;
+            }
+            return dequantize_block_cuda<QK8_0, QR8_0, dequantize_q8_0>;
+        case GGML_TYPE_Q2_K:
+            return dequantize_row_q2_K_cuda;
+        case GGML_TYPE_Q3_K:
+            return dequantize_row_q3_K_cuda;
+        case GGML_TYPE_Q4_K:
+            return dequantize_row_q4_K_cuda;
+        case GGML_TYPE_Q5_K:
+            return dequantize_row_q5_K_cuda;
+        case GGML_TYPE_Q6_K:
+            return dequantize_row_q6_K_cuda;
+        case GGML_TYPE_IQ2_XXS:
+            return dequantize_row_iq2_xxs_cuda;
+        case GGML_TYPE_IQ2_XS:
+            return dequantize_row_iq2_xs_cuda;
+        case GGML_TYPE_IQ2_S:
+            return dequantize_row_iq2_s_cuda;
+        case GGML_TYPE_IQ3_XXS:
+            return dequantize_row_iq3_xxs_cuda;
+        case GGML_TYPE_IQ1_S:
+            return dequantize_row_iq1_s_cuda;
+        case GGML_TYPE_IQ1_M:
+            return dequantize_row_iq1_m_cuda;
+        case GGML_TYPE_IQ4_NL:
+            return dequantize_row_iq4_nl_cuda;
+        case GGML_TYPE_IQ4_XS:
+            return dequantize_row_iq4_xs_cuda;
+        case GGML_TYPE_IQ3_S:
+            return dequantize_row_iq3_s_cuda;
+        case GGML_TYPE_F32:
+            return convert_unary_cuda<float>;
+        default:
+            return nullptr;
+    }
+}
+
+to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_Q4_0:
+            return dequantize_row_q4_0_cuda;
+        case GGML_TYPE_Q4_1:
+            return dequantize_row_q4_1_cuda;
+        case GGML_TYPE_Q5_0:
+            return dequantize_block_cuda<QK5_0, QR5_0, dequantize_q5_0>;
+        case GGML_TYPE_Q5_1:
+            return dequantize_block_cuda<QK5_1, QR5_1, dequantize_q5_1>;
+        case GGML_TYPE_Q8_0:
+            return dequantize_block_cuda<QK8_0, QR8_0, dequantize_q8_0>;
+        case GGML_TYPE_Q2_K:
+            return dequantize_row_q2_K_cuda;
+        case GGML_TYPE_Q3_K:
+            return dequantize_row_q3_K_cuda;
+        case GGML_TYPE_Q4_K:
+            return dequantize_row_q4_K_cuda;
+        case GGML_TYPE_Q5_K:
+            return dequantize_row_q5_K_cuda;
+        case GGML_TYPE_Q6_K:
+            return dequantize_row_q6_K_cuda;
+        case GGML_TYPE_IQ2_XXS:
+            return dequantize_row_iq2_xxs_cuda;
+        case GGML_TYPE_IQ2_XS:
+            return dequantize_row_iq2_xs_cuda;
+        case GGML_TYPE_IQ2_S:
+            return dequantize_row_iq2_s_cuda;
+        case GGML_TYPE_IQ3_XXS:
+            return dequantize_row_iq3_xxs_cuda;
+        case GGML_TYPE_IQ1_S:
+            return dequantize_row_iq1_s_cuda;
+        case GGML_TYPE_IQ1_M:
+            return dequantize_row_iq1_m_cuda;
+        case GGML_TYPE_IQ4_NL:
+            return dequantize_row_iq4_nl_cuda;
+        case GGML_TYPE_IQ4_XS:
+            return dequantize_row_iq4_xs_cuda;
+        case GGML_TYPE_IQ3_S:
+            return dequantize_row_iq3_s_cuda;
+        case GGML_TYPE_F16:
+            return convert_unary_cuda<half>;
+        default:
+            return nullptr;
+    }
+}
--- a/llama/ggml-cuda/convert.cuh
+++ b/llama/ggml-cuda/convert.cuh
@@ -0,0 +1,39 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_DEQUANTIZE_BLOCK_SIZE 256
+
+template<typename T>
+using to_t_cuda_t = void (*)(const void * __restrict__ x, T * __restrict__ y, int64_t k, cudaStream_t stream);
+
+typedef to_t_cuda_t<float> to_fp32_cuda_t;
+typedef to_t_cuda_t<half> to_fp16_cuda_t;
+
+to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type);
+
+to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type);
--- a/llama/ggml-cuda/cpy.cu
+++ b/llama/ggml-cuda/cpy.cu
@@ -0,0 +1,515 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "cpy.cuh"
+
+typedef void (*cpy_kernel_t)(const char * cx, char * cdst);
+
+static __device__ void cpy_1_f32_f32(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    float * dsti = (float *) cdsti;
+
+    *dsti = *xi;
+}
+
+static __device__ void cpy_1_f32_f16(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    half * dsti = (half *) cdsti;
+
+    *dsti = __float2half(*xi);
+}
+
+static __device__ void cpy_1_f16_f16(const char * cxi, char * cdsti) {
+    const half * xi = (const half *) cxi;
+    half * dsti = (half *) cdsti;
+
+    *dsti = *xi;
+}
+
+static __device__ void cpy_1_f16_f32(const char * cxi, char * cdsti) {
+    const half * xi = (const half *) cxi;
+    float * dsti = (float *) cdsti;
+
+    *dsti = *xi;
+}
+
+template <cpy_kernel_t cpy_1>
+static __global__ void cpy_f32_f16(const char * cx, char * cdst, const int ne,
+                                   const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+                                   const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
+                                   const int nb12, const int nb13) {
+    const int64_t i = blockDim.x*blockIdx.x + threadIdx.x;
+
+    if (i >= ne) {
+        return;
+    }
+
+    // determine indices i03/i13, i02/i12, i01/i11, i00/i10 as a function of index i of flattened tensor
+    // then combine those indices with the corresponding byte offsets to get the total offsets
+    const int64_t i03 = i/(ne00 * ne01 * ne02);
+    const int64_t i02 = (i - i03*ne00*ne01*ne02 )/ (ne00*ne01);
+    const int64_t i01 = (i - i03*ne00*ne01*ne02  -  i02*ne01*ne00) / ne00;
+    const int64_t i00 = i - i03*ne00*ne01*ne02 - i02*ne01*ne00 - i01*ne00;
+    const int64_t x_offset = i00*nb00 + i01*nb01 + i02*nb02 + i03 * nb03;
+
+    const int64_t i13 = i/(ne10 * ne11 * ne12);
+    const int64_t i12 = (i - i13*ne10*ne11*ne12) / (ne10*ne11);
+    const int64_t i11 = (i - i13*ne10*ne11*ne12 - i12*ne10*ne11) / ne10;
+    const int64_t i10 = i - i13*ne10*ne11*ne12 - i12*ne10*ne11 - i11*ne10;
+    const int64_t dst_offset = i10*nb10 + i11*nb11 + i12*nb12 + i13 * nb13;
+
+    cpy_1(cx + x_offset, cdst + dst_offset);
+}
+
+static __device__ void cpy_blck_f32_q8_0(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    block_q8_0 * dsti = (block_q8_0 *) cdsti;
+
+    float amax = 0.0f; // absolute max
+
+    for (int j = 0; j < QK8_0; j++) {
+        const float v = xi[j];
+        amax = fmaxf(amax, fabsf(v));
+    }
+
+    const float d = amax / ((1 << 7) - 1);
+    const float id = d ? 1.0f/d : 0.0f;
+
+    dsti->d = d;
+
+    for (int j = 0; j < QK8_0; ++j) {
+        const float x0 = xi[j]*id;
+
+        dsti->qs[j] = roundf(x0);
+    }
+}
+
+static __device__ void cpy_blck_f32_q4_0(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    block_q4_0 * dsti = (block_q4_0 *) cdsti;
+
+    float amax = 0.0f;
+    float vmax = 0.0f;
+
+    for (int j = 0; j < QK4_0; ++j) {
+        const float v = xi[j];
+        if (amax < fabsf(v)) {
+            amax = fabsf(v);
+            vmax = v;
+        }
+    }
+
+    const float d  = vmax / -8;
+    const float id = d ? 1.0f/d : 0.0f;
+
+    dsti->d = d;
+
+    for (int j = 0; j < QK4_0/2; ++j) {
+        const float x0 = xi[0       + j]*id;
+        const float x1 = xi[QK4_0/2 + j]*id;
+
+        const uint8_t xi0 = min(15, (int8_t)(x0 + 8.5f));
+        const uint8_t xi1 = min(15, (int8_t)(x1 + 8.5f));
+
+        dsti->qs[j]  = xi0;
+        dsti->qs[j] |= xi1 << 4;
+    }
+}
+
+static __device__ void cpy_blck_f32_q4_1(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    block_q4_1 * dsti = (block_q4_1 *) cdsti;
+
+    float vmin = FLT_MAX;
+    float vmax = -FLT_MAX;
+
+    for (int j = 0; j < QK4_1; ++j) {
+        const float v = xi[j];
+
+        if (v < vmin) vmin = v;
+        if (v > vmax) vmax = v;
+    }
+
+    const float d  = (vmax - vmin) / ((1 << 4) - 1);
+    const float id = d ? 1.0f/d : 0.0f;
+
+    dsti->dm.x = d;
+    dsti->dm.y = vmin;
+
+    for (int j = 0; j < QK4_1/2; ++j) {
+        const float x0 = (xi[0       + j] - vmin)*id;
+        const float x1 = (xi[QK4_1/2 + j] - vmin)*id;
+
+        const uint8_t xi0 = min(15, (int8_t)(x0 + 0.5f));
+        const uint8_t xi1 = min(15, (int8_t)(x1 + 0.5f));
+
+        dsti->qs[j]  = xi0;
+        dsti->qs[j] |= xi1 << 4;
+    }
+}
+
+static __device__ void cpy_blck_f32_q5_0(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    block_q5_0 * dsti = (block_q5_0 *) cdsti;
+
+    float amax = 0.0f;
+    float vmax = 0.0f;
+
+    for (int j = 0; j < QK5_0; ++j) {
+        const float v = xi[j];
+        if (amax < fabsf(v)) {
+            amax = fabsf(v);
+            vmax = v;
+        }
+    }
+
+    const float d  = vmax / -16;
+    const float id = d ? 1.0f/d : 0.0f;
+
+    dsti->d = d;
+
+    uint32_t qh = 0;
+    for (int j = 0; j < QK5_0/2; ++j) {
+        const float x0 = xi[0       + j]*id;
+        const float x1 = xi[QK5_0/2 + j]*id;
+
+        const uint8_t xi0 = min(31, (int8_t)(x0 + 16.5f));
+        const uint8_t xi1 = min(31, (int8_t)(x1 + 16.5f));
+
+        dsti->qs[j]  = (xi0 & 0xf) | ((xi1 & 0xf) << 4);
+        qh |= ((xi0 & 0x10u) >> 4) << (j + 0);
+        qh |= ((xi1 & 0x10u) >> 4) << (j + QK5_0/2);
+    }
+    memcpy(dsti->qh, &qh, sizeof(qh));
+}
+
+static __device__ void cpy_blck_f32_q5_1(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    block_q5_1 * dsti = (block_q5_1 *) cdsti;
+
+    float min = xi[0];
+    float max = xi[0];
+
+    for (int j = 1; j < QK5_1; ++j) {
+        const float v = xi[j];
+        min = v < min ? v : min;
+        max = v > max ? v : max;
+    }
+
+    const float d  = (max - min) / 31;
+    const float id = d ? 1.0f/d : 0.0f;
+
+    dsti->dm.x = d;
+    dsti->dm.y = min;
+
+    uint32_t qh = 0;
+    for (int j = 0; j < QK5_1/2; ++j) {
+        const float x0 = (xi[0       + j] - min)*id;
+        const float x1 = (xi[QK5_1/2 + j] - min)*id;
+
+        const uint8_t xi0 = (uint8_t)(x0 + 0.5f);
+        const uint8_t xi1 = (uint8_t)(x1 + 0.5f);
+
+        dsti->qs[j]  = (xi0 & 0xf) | ((xi1 & 0xf) << 4);
+        qh |= ((xi0 & 0x10u) >> 4) << (j + 0);
+        qh |= ((xi1 & 0x10u) >> 4) << (j + QK5_1/2);
+    }
+    memcpy(dsti->qh, &qh, sizeof(qh));
+}
+
+
+static __device__ __forceinline__ int best_index_int8(int n, const int8_t * val, float x) {
+    if (x <= val[0]) return 0;
+    if (x >= val[n-1]) return n-1;
+    int ml = 0, mu = n-1;
+    while (mu-ml > 1) {
+        int mav = (ml+mu)/2;
+        if (x < val[mav]) mu = mav; else ml = mav;
+    }
+    return x - val[mu-1] < val[mu] - x ? mu-1 : mu;
+}
+
+static __device__ void cpy_blck_f32_iq4_nl(const char * cxi, char * cdsti) {
+    const float * xi = (const float *) cxi;
+    block_iq4_nl * dsti = (block_iq4_nl *) cdsti;
+
+    float amax = 0.0f;
+    float vmax = 0.0f;
+
+    for (int j = 0; j < QK4_NL; ++j) {
+        const float v = xi[j];
+        if (amax < fabsf(v)) {
+            amax = fabsf(v);
+            vmax = v;
+        }
+    }
+
+    float d = vmax / kvalues_iq4nl[0];
+    const float id = d ? 1.0f/d : 0.0f;
+
+    float sumqx = 0, sumq2 = 0;
+    for (int j = 0; j < QK4_NL/2; ++j) {
+        const float x0 = xi[0        + j]*id;
+        const float x1 = xi[QK4_NL/2 + j]*id;
+        const uint8_t xi0 = best_index_int8(16, kvalues_iq4nl, x0);
+        const uint8_t xi1 = best_index_int8(16, kvalues_iq4nl, x1);
+        dsti->qs[j] = xi0 | (xi1 << 4);
+        const float v0 = kvalues_iq4nl[xi0];
+        const float v1 = kvalues_iq4nl[xi1];
+        const float w0 = xi[0        + j]*xi[0        + j];
+        const float w1 = xi[QK4_NL/2 + j]*xi[QK4_NL/2 + j];
+        sumqx += w0*v0*xi[j] + w1*v1*xi[QK4_NL/2 + j];
+        sumq2 += w0*v0*v0 + w1*v1*v1;
+    }
+
+    dsti->d = sumq2 > 0 ? sumqx/sumq2 : d;
+}
+
+template <cpy_kernel_t cpy_blck, int qk>
+static __global__ void cpy_f32_q(const char * cx, char * cdst, const int ne,
+                                 const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+                                 const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
+                                 const int nb12, const int nb13) {
+    const int i = (blockDim.x*blockIdx.x + threadIdx.x)*qk;
+
+    if (i >= ne) {
+        return;
+    }
+
+    const int i03 = i/(ne00 * ne01 * ne02);
+    const int i02 = (i - i03*ne00*ne01*ne02 )/ (ne00*ne01);
+    const int i01 = (i - i03*ne00*ne01*ne02  -  i02*ne01*ne00) / ne00;
+    const int i00 = i - i03*ne00*ne01*ne02 - i02*ne01*ne00 - i01*ne00;
+    const int x_offset = i00*nb00 + i01*nb01 + i02*nb02 + i03 * nb03;
+
+    const int i13 = i/(ne10 * ne11 * ne12);
+    const int i12 = (i - i13*ne10*ne11*ne12) / (ne10*ne11);
+    const int i11 = (i - i13*ne10*ne11*ne12 - i12*ne10*ne11) / ne10;
+    const int i10 = i - i13*ne10*ne11*ne12 - i12*ne10*ne11 - i11*ne10;
+    const int dst_offset = (i10/qk)*nb10 + i11*nb11 + i12*nb12 + i13*nb13;
+
+    cpy_blck(cx + x_offset, cdst + dst_offset);
+}
+
+static void ggml_cpy_f16_f32_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    const int num_blocks = (ne + CUDA_CPY_BLOCK_SIZE - 1) / CUDA_CPY_BLOCK_SIZE;
+    cpy_f32_f16<cpy_1_f16_f32><<<num_blocks, CUDA_CPY_BLOCK_SIZE, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_f32_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    const int num_blocks = (ne + CUDA_CPY_BLOCK_SIZE - 1) / CUDA_CPY_BLOCK_SIZE;
+    cpy_f32_f16<cpy_1_f32_f32><<<num_blocks, CUDA_CPY_BLOCK_SIZE, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_f16_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    const int num_blocks = (ne + CUDA_CPY_BLOCK_SIZE - 1) / CUDA_CPY_BLOCK_SIZE;
+    cpy_f32_f16<cpy_1_f32_f16><<<num_blocks, CUDA_CPY_BLOCK_SIZE, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_q8_0_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    GGML_ASSERT(ne % QK8_0 == 0);
+    const int num_blocks = ne / QK8_0;
+    cpy_f32_q<cpy_blck_f32_q8_0, QK8_0><<<num_blocks, 1, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_q4_0_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    GGML_ASSERT(ne % QK4_0 == 0);
+    const int num_blocks = ne / QK4_0;
+    cpy_f32_q<cpy_blck_f32_q4_0, QK4_0><<<num_blocks, 1, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_q4_1_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    GGML_ASSERT(ne % QK4_1 == 0);
+    const int num_blocks = ne / QK4_1;
+    cpy_f32_q<cpy_blck_f32_q4_1, QK4_1><<<num_blocks, 1, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_q5_0_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    GGML_ASSERT(ne % QK5_0 == 0);
+    const int num_blocks = ne / QK5_0;
+    cpy_f32_q<cpy_blck_f32_q5_0, QK5_0><<<num_blocks, 1, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_q5_1_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    GGML_ASSERT(ne % QK5_1 == 0);
+    const int num_blocks = ne / QK5_1;
+    cpy_f32_q<cpy_blck_f32_q5_1, QK5_1><<<num_blocks, 1, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f32_iq4_nl_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    GGML_ASSERT(ne % QK4_NL == 0);
+    const int num_blocks = ne / QK4_NL;
+    cpy_f32_q<cpy_blck_f32_iq4_nl, QK4_NL><<<num_blocks, 1, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+static void ggml_cpy_f16_f16_cuda(
+    const char * cx, char * cdst, const int ne,
+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream) {
+
+    const int num_blocks = (ne + CUDA_CPY_BLOCK_SIZE - 1) / CUDA_CPY_BLOCK_SIZE;
+    cpy_f32_f16<cpy_1_f16_f16><<<num_blocks, CUDA_CPY_BLOCK_SIZE, 0, stream>>>
+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
+}
+
+void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, ggml_tensor * src1) {
+    const int64_t ne = ggml_nelements(src0);
+    GGML_ASSERT(ne == ggml_nelements(src1));
+
+    GGML_ASSERT(ggml_nbytes(src0) <= INT_MAX);
+    GGML_ASSERT(ggml_nbytes(src1) <= INT_MAX);
+
+    const int64_t ne00 = src0->ne[0];
+    const int64_t ne01 = src0->ne[1];
+    const int64_t ne02 = src0->ne[2];
+
+    //GGML_ASSERT(src0->ne[3] == 1);
+
+    const int64_t nb00 = src0->nb[0];
+    const int64_t nb01 = src0->nb[1];
+    const int64_t nb02 = src0->nb[2];
+    const int64_t nb03 = src0->nb[3];
+
+    const int64_t ne10 = src1->ne[0];
+    const int64_t ne11 = src1->ne[1];
+    const int64_t ne12 = src1->ne[2];
+
+    //GGML_ASSERT(src1->ne[3] == 1);
+
+    const int64_t nb10 = src1->nb[0];
+    const int64_t nb11 = src1->nb[1];
+    const int64_t nb12 = src1->nb[2];
+    const int64_t nb13 = src1->nb[3];
+
+    cudaStream_t main_stream = ctx.stream();
+
+    char * src0_ddc = (char *) src0->data;
+    char * src1_ddc = (char *) src1->data;
+
+    if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_F32) {
+        ggml_cpy_f32_f32_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_F16) {
+        ggml_cpy_f32_f16_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q8_0) {
+        ggml_cpy_f32_q8_0_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q4_0) {
+        ggml_cpy_f32_q4_0_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q4_1) {
+        ggml_cpy_f32_q4_1_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q5_0) {
+        ggml_cpy_f32_q5_0_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_IQ4_NL) {
+        ggml_cpy_f32_iq4_nl_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q5_1) {
+        ggml_cpy_f32_q5_1_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F16) {
+        ggml_cpy_f16_f16_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F32) {
+        ggml_cpy_f16_f32_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream);
+    } else {
+        fprintf(stderr, "%s: unsupported type combination (%s to %s)\n", __func__,
+                ggml_type_name(src0->type), ggml_type_name(src1->type));
+        GGML_ABORT("fatal error");
+    }
+}
+
+void ggml_cuda_dup(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    ggml_cuda_cpy(ctx, src0, dst);
+}
+
+void* ggml_cuda_cpy_fn(const ggml_tensor * src0, ggml_tensor * src1) {
+    if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_F32) {
+            return (void*) cpy_f32_f16<cpy_1_f32_f32>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_F16) {
+            return (void*) cpy_f32_f16<cpy_1_f32_f16>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q8_0) {
+            return (void*) cpy_f32_q<cpy_blck_f32_q8_0, QK8_0>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q4_0) {
+            return (void*) cpy_f32_q<cpy_blck_f32_q4_0, QK4_0>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q4_1) {
+            return (void*) cpy_f32_q<cpy_blck_f32_q4_1, QK4_1>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q5_0) {
+            return (void*) cpy_f32_q<cpy_blck_f32_q5_0, QK5_0>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_IQ4_NL) {
+            return (void*) cpy_f32_q<cpy_blck_f32_iq4_nl, QK4_NL>;
+    } else if (src0->type == GGML_TYPE_F32 && src1->type == GGML_TYPE_Q5_1) {
+            return (void*) cpy_f32_q<cpy_blck_f32_q5_1, QK5_1>;
+    } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F16) {
+            return (void*) cpy_f32_f16<cpy_1_f32_f16>;
+    } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F32) {
+            return (void*) cpy_f32_f16<cpy_1_f16_f32>;
+    } else {
+        fprintf(stderr, "%s: unsupported type combination (%s to %s)\n", __func__,
+                ggml_type_name(src0->type), ggml_type_name(src1->type));
+        GGML_ABORT("fatal error");
+    }
+}
--- a/llama/ggml-cuda/cpy.cuh
+++ b/llama/ggml-cuda/cpy.cuh
@@ -0,0 +1,35 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_CPY_BLOCK_SIZE 32
+
+void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, ggml_tensor * src1);
+
+void ggml_cuda_dup(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
+
+void* ggml_cuda_cpy_fn(const ggml_tensor * src0, ggml_tensor * src1);
--- a/llama/ggml-cuda/cross-entropy-loss.cu
+++ b/llama/ggml-cuda/cross-entropy-loss.cu
@@ -0,0 +1,132 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+#include "cross-entropy-loss.cuh"
+#include "sumrows.cuh"
+
+#include <cmath>
+#include <cstdint>
+
+static __global__ void cross_entropy_loss_f32(const float * logits, const float * labels, float * dst, const int nclasses, const int k) {
+    const int warp_id = threadIdx.x / WARP_SIZE;
+    const int lane_id = threadIdx.x % WARP_SIZE;
+    const int i0 = blockDim.x*blockIdx.x + warp_id*WARP_SIZE;
+
+    const int ne_tmp = WARP_SIZE*nclasses;
+
+    extern __shared__ float tmp_all[];
+    float * tmp_logits = tmp_all + (2*warp_id + 0)*ne_tmp;
+    float * tmp_labels = tmp_all + (2*warp_id + 1)*ne_tmp;
+
+    // Each warp first loads ne_tmp logits/labels into shared memory:
+    for (int i = lane_id; i < ne_tmp; i += WARP_SIZE) {
+        const int ig = i0*nclasses + i; // ig == i global
+
+        tmp_logits[i] = ig < k*nclasses ? logits[ig] : 0.0f;
+        tmp_labels[i] = ig < k*nclasses ? labels[ig] : 0.0f;
+    }
+
+    // Each thread in the warp then calculates the cross entropy loss for a single row.
+    // TODO: pad in order to avoid shared memory bank conflicts.
+
+    // Find maximum for softmax:
+    float max = -INFINITY;
+    for (int i = 0; i < nclasses; ++i) {
+        max = fmaxf(max, tmp_logits[lane_id*nclasses + i]);
+    }
+
+    // Calculate log(softmax(logits)) which is just logits - max:
+    float sum = 0.0f;
+    for (int i = 0; i < nclasses; ++i) {
+        float val = tmp_logits[lane_id*nclasses + i] - max;
+        sum += expf(val);
+        tmp_logits[lane_id*nclasses + i] = val;
+    }
+    sum = logf(sum);
+
+    // log(exp(logits - max) / sum) = (logits - max) - log(sum)
+    float loss = 0.0f;
+    for (int i = 0; i < nclasses; ++i) {
+        loss += (tmp_logits[lane_id*nclasses + i] - sum) * tmp_labels[lane_id*nclasses + i];
+    }
+    loss = -warp_reduce_sum(loss) / (float)k;
+
+    __syncthreads();
+
+    if (lane_id == 0) {
+        tmp_all[warp_id] = loss;
+    }
+
+    __syncthreads();
+
+    if (warp_id != 0) {
+        return;
+    }
+
+    loss = lane_id < CUDA_CROSS_ENTROPY_LOSS_BLOCK_SIZE/WARP_SIZE ? tmp_all[lane_id] : 0.0f;
+    loss = warp_reduce_sum(loss);
+
+    if (lane_id != 0) {
+        return;
+    }
+
+    dst[blockIdx.x] = loss;
+}
+
+void ggml_cuda_cross_entropy_loss(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const ggml_tensor * src1 = dst->src[1];
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT(src1->type == GGML_TYPE_F32);
+    GGML_ASSERT( dst->type == GGML_TYPE_F32);
+
+    GGML_ASSERT(ggml_is_contiguous(src0));
+    GGML_ASSERT(ggml_is_contiguous(src1));
+    GGML_ASSERT(ggml_is_contiguous(dst));
+
+    const int64_t ne00  = src0->ne[0];
+    const int64_t nrows = ggml_nrows(src0);
+
+    const float * src0_d = (const float *) src0->data;
+    const float * src1_d = (const float *) src1->data;
+    float       * dst_d  = (float       *) dst->data;
+
+    ggml_cuda_pool & pool = ctx.pool();
+    cudaStream_t stream = ctx.stream();
+
+    const dim3 blocks_dim(CUDA_CROSS_ENTROPY_LOSS_BLOCK_SIZE, 1, 1);
+    const dim3 blocks_num((nrows + CUDA_CROSS_ENTROPY_LOSS_BLOCK_SIZE - 1) / CUDA_CROSS_ENTROPY_LOSS_BLOCK_SIZE, 1, 1);
+    const int shmem = 2*CUDA_CROSS_ENTROPY_LOSS_BLOCK_SIZE*ne00*sizeof(float);
+
+    ggml_cuda_pool_alloc<float> dst_tmp(pool, blocks_num.x);
+
+    cross_entropy_loss_f32<<<blocks_num, blocks_dim, shmem, stream>>>(src0_d, src1_d, dst_tmp.ptr, ne00, nrows);
+
+    // Combine results from individual blocks:
+    sum_rows_f32_cuda(dst_tmp.ptr, dst_d, blocks_num.x, 1, stream);
+}
--- a/llama/ggml-cuda/cross-entropy-loss.cuh
+++ b/llama/ggml-cuda/cross-entropy-loss.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_CROSS_ENTROPY_LOSS_BLOCK_SIZE 256
+
+void ggml_cuda_cross_entropy_loss(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/dequantize.cuh
+++ b/llama/ggml-cuda/dequantize.cuh
@@ -0,0 +1,129 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+static __device__ __forceinline__ void dequantize_q4_0(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
+    const block_q4_0 * x = (const block_q4_0 *) vx;
+
+    const dfloat d = x[ib].d;
+
+    const int vui = x[ib].qs[iqs];
+
+    v.x = vui & 0xF;
+    v.y = vui >> 4;
+
+#ifdef GGML_CUDA_F16
+    v = __hsub2(v, {8.0f, 8.0f});
+    v = __hmul2(v, {d, d});
+#else
+    v.x = (v.x - 8.0f) * d;
+    v.y = (v.y - 8.0f) * d;
+#endif // GGML_CUDA_F16
+}
+
+static __device__ __forceinline__ void dequantize_q4_1(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
+    const block_q4_1 * x = (const block_q4_1 *) vx;
+
+    const dfloat d = __low2half(x[ib].dm);
+    const dfloat m = __high2half(x[ib].dm);
+
+    const int vui = x[ib].qs[iqs];
+
+    v.x = vui & 0xF;
+    v.y = vui >> 4;
+
+#ifdef GGML_CUDA_F16
+    v = __hmul2(v, {d, d});
+    v = __hadd2(v, {m, m});
+#else
+    v.x = (v.x * d) + m;
+    v.y = (v.y * d) + m;
+#endif // GGML_CUDA_F16
+}
+
+static __device__ __forceinline__ void dequantize_q5_0(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
+    const block_q5_0 * x = (const block_q5_0 *) vx;
+
+    const dfloat d = x[ib].d;
+
+    uint32_t qh;
+    memcpy(&qh, x[ib].qh, sizeof(qh));
+
+    const int xh_0 = ((qh >> (iqs +  0)) << 4) & 0x10;
+    const int xh_1 = ((qh >> (iqs + 12))     ) & 0x10;
+
+    v.x = ((x[ib].qs[iqs] & 0xf) | xh_0);
+    v.y = ((x[ib].qs[iqs] >>  4) | xh_1);
+
+#ifdef GGML_CUDA_F16
+    v = __hsub2(v, {16.0f, 16.0f});
+    v = __hmul2(v, {d, d});
+#else
+    v.x = (v.x - 16.0f) * d;
+    v.y = (v.y - 16.0f) * d;
+#endif // GGML_CUDA_F16
+}
+
+static __device__ __forceinline__ void dequantize_q5_1(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
+    const block_q5_1 * x = (const block_q5_1 *) vx;
+
+    const dfloat d = __low2half(x[ib].dm);
+    const dfloat m = __high2half(x[ib].dm);
+
+    uint32_t qh;
+    memcpy(&qh, x[ib].qh, sizeof(qh));
+
+    const int xh_0 = ((qh >> (iqs +  0)) << 4) & 0x10;
+    const int xh_1 = ((qh >> (iqs + 12))     ) & 0x10;
+
+    v.x = ((x[ib].qs[iqs] & 0xf) | xh_0);
+    v.y = ((x[ib].qs[iqs] >>  4) | xh_1);
+
+#ifdef GGML_CUDA_F16
+    v = __hmul2(v, {d, d});
+    v = __hadd2(v, {m, m});
+#else
+    v.x = (v.x * d) + m;
+    v.y = (v.y * d) + m;
+#endif // GGML_CUDA_F16
+}
+
+static __device__ __forceinline__ void dequantize_q8_0(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
+    const block_q8_0 * x = (const block_q8_0 *) vx;
+
+    const dfloat d = x[ib].d;
+
+    v.x = x[ib].qs[iqs + 0];
+    v.y = x[ib].qs[iqs + 1];
+
+#ifdef GGML_CUDA_F16
+    v = __hmul2(v, {d, d});
+#else
+    v.x *= d;
+    v.y *= d;
+#endif // GGML_CUDA_F16
+}
--- a/llama/ggml-cuda/diagmask.cu
+++ b/llama/ggml-cuda/diagmask.cu
@@ -0,0 +1,66 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "diagmask.cuh"
+
+static __global__ void diag_mask_inf_f32(const float * x, float * dst, const int ncols, const int rows_per_channel, const int n_past) {
+    const int col = blockDim.y*blockIdx.y + threadIdx.y;
+    const int row = blockDim.x*blockIdx.x + threadIdx.x;
+
+    if (col >= ncols) {
+        return;
+    }
+
+    const int i = row*ncols + col;
+    //dst[i] = col > (n_past + row % rows_per_channel) ? -INFINITY : x[i];
+    //dst[i] = x[i] - (col > n_past + row % rows_per_channel) * INT_MAX; // equivalent within rounding error but slightly faster on GPU
+    dst[i] = x[i] - (col > n_past + row % rows_per_channel) * FLT_MAX;
+}
+
+static void diag_mask_inf_f32_cuda(const float * x, float * dst, const int ncols_x, const int nrows_x, const int rows_per_channel, const int n_past, cudaStream_t stream) {
+    const dim3 block_dims(1, CUDA_DIAG_MASK_INF_BLOCK_SIZE, 1);
+    const int block_num_x = (ncols_x + CUDA_DIAG_MASK_INF_BLOCK_SIZE - 1) / CUDA_DIAG_MASK_INF_BLOCK_SIZE;
+    const dim3 block_nums(nrows_x, block_num_x, 1);
+    diag_mask_inf_f32<<<block_nums, block_dims, 0, stream>>>(x, dst, ncols_x, rows_per_channel, n_past);
+}
+
+void ggml_cuda_op_diag_mask_inf(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
+    const ggml_tensor * src0 = dst->src[0];
+    const float * src0_d = (const float *)src0->data;
+    float * dst_d = (float *)dst->data;
+    cudaStream_t stream = ctx.stream();
+
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT( dst->type == GGML_TYPE_F32);
+
+    const int64_t ne00 = src0->ne[0];
+    const int64_t ne01 = src0->ne[1];
+    const int nrows0 = ggml_nrows(src0);
+
+    const int n_past = ((int32_t *) dst->op_params)[0];
+
+    diag_mask_inf_f32_cuda(src0_d, dst_d, ne00, nrows0, ne01, n_past, stream);
+}
--- a/llama/ggml-cuda/diagmask.cuh
+++ b/llama/ggml-cuda/diagmask.cuh
@@ -0,0 +1,31 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "common.cuh"
+
+#define CUDA_DIAG_MASK_INF_BLOCK_SIZE 32
+
+void ggml_cuda_op_diag_mask_inf(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
--- a/llama/ggml-cuda/dmmv.cu
+++ b/llama/ggml-cuda/dmmv.cu
@@ -0,0 +1,709 @@
+/**
+ * llama.cpp - commit 8962422b1c6f9b8b15f5aeaea42600bcc2d44177 - do not edit this file
+ *
+ * MIT License
+ *
+ * Copyright (c) 2023-2024 The ggml authors
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "dmmv.cuh"
+#include "dequantize.cuh"
+#include "convert.cuh"
+
+#ifndef K_QUANTS_PER_ITERATION
+#define K_QUANTS_PER_ITERATION 2
+#else
+static_assert(K_QUANTS_PER_ITERATION == 1 || K_QUANTS_PER_ITERATION == 2, "K_QUANTS_PER_ITERATION must be 1 or 2");
+#endif
+
+static __global__ void dequantize_mul_mat_vec_q2_k(const void * __restrict__ vx, const float * __restrict__ yy, float * __restrict__ dst, const int ncols, int nrows) {
+
+    static_assert(16%K_QUANTS_PER_ITERATION == 0, "16 must be divisible by K_QUANTS_PER_ITERATION");
+
+    const int row = blockIdx.x*blockDim.y + threadIdx.y;
+    if (row > nrows) return;
+
+    const int num_blocks_per_row = ncols / QK_K;
+    const int ib0 = row*num_blocks_per_row;
+
+    const block_q2_K * x = (const block_q2_K *)vx + ib0;
+
+    float tmp = 0; // partial sum for thread in warp
+
+    const int tid = threadIdx.x/K_QUANTS_PER_ITERATION;  // 0...31 or 0...15
+    const int ix  = threadIdx.x%K_QUANTS_PER_ITERATION;  // 0 or 0,1
+
+    const int step = 16/K_QUANTS_PER_ITERATION;
+
+    const int im = tid/step;                             // 0 or 1. 0 computes 0..., 1 computes 128...
+    const int in = tid - step*im;                        // 0...15 or 0...7
+
+    const int l0 = K_QUANTS_PER_ITERATION*in;            // 0...15 or 0...14 in steps of 2
+    const int q_offset = 32*im + l0;
+    const int s_offset = 8*im;
+    const int y_offset = 128*im + l0;
+
+    uint32_t aux[4];
+    const uint8_t * d = (const uint8_t *)aux;
+    const uint8_t * m = (const uint8_t *)(aux + 2);
+
+    for (int i = ix; i < num_blocks_per_row; i += K_QUANTS_PER_ITERATION) {
+
+        const float   * y = yy + i * QK_K + y_offset;
+        const uint8_t * q = x[i].qs + q_offset;
+
+        const float dall = __low2half(x[i].dm);
+        const float dmin = __high2half(x[i].dm);
+
+        const uint32_t * a = (const uint32_t *)(x[i].scales + s_offset);
+        aux[0] = a[0] & 0x0f0f0f0f;
+        aux[1] = a[1] & 0x0f0f0f0f;
+        aux[2] = (a[0] >> 4) & 0x0f0f0f0f;
+        aux[3] = (a[1] >> 4) & 0x0f0f0f0f;
+
+        float sum1 = 0, sum2 = 0;
+        for (int l = 0; l < K_QUANTS_PER_ITERATION; ++l) {
+            sum1 += y[l+ 0] * d[0] * ((q[l+ 0] >> 0) & 3)
+                  + y[l+32] * d[2] * ((q[l+ 0] >> 2) & 3)
+                  + y[l+64] * d[4] * ((q[l+ 0] >> 4) & 3)
+                  + y[l+96] * d[6] * ((q[l+ 0] >> 6) & 3)
+                  + y[l+16] * d[1] * ((q[l+16] >> 0) & 3)
+                  + y[l+48] * d[3] * ((q[l+16] >> 2) & 3)
+                  + y[l+80] * d[5] * ((q[l+16] >> 4) & 3)
+                  +y[l+112] * d[7] * ((q[l+16] >> 6) & 3);
+            sum2 += y[l+ 0] * m[0] + y[l+32] * m[2] + y[l+64] * m[4] + y[ l+96] * m[6]
+                  + y[l+16] * m[1] + y[l+48] * m[3] + y[l+80] * m[5] + y[l+112] * m[7];
+
+        }
+        tmp += dall * sum1 - dmin * sum2;
+
+    }
+
+    // sum up partial sums and write back result
+    tmp = warp_reduce_sum(tmp);
+
+    if (threadIdx.x == 0) {
+        dst[row] = tmp;
+    }
+}
+
+static __global__ void dequantize_mul_mat_vec_q3_k(const void * __restrict__ vx, const float * __restrict__ yy, float * __restrict__ dst, const int ncols, int nrows) {
+
+    const int row = blockIdx.x*blockDim.y + threadIdx.y;
+    if (row > nrows) return;
+
+    const int num_blocks_per_row = ncols / QK_K;
+    const int ib0 = row*num_blocks_per_row;
+
+    const block_q3_K * x = (const block_q3_K *)vx + ib0;
+
+    float tmp = 0; // partial sum for thread in warp
+
+    const uint16_t kmask1 = 0x0303;
+    const uint16_t kmask2 = 0x0f0f;
+
+    const int tid = threadIdx.x/K_QUANTS_PER_ITERATION;  // 0...31 or 0...16
+    const int ix  = threadIdx.x%K_QUANTS_PER_ITERATION;  // 0 or 0,1
+
+    const int n  = K_QUANTS_PER_ITERATION;               // iterations in the inner loop
+    const int step = 16/K_QUANTS_PER_ITERATION;
+    const int im = tid/step;                             // 0 or 1. 0 computes 0..., 1 computes 128...
+    const int in = tid - step*im;                        // 0....15 or 0...7
+
+    const uint8_t m = 1 << (4*im);
+
+    const int l0 = n*in;                                 // 0...15 or 0...14 in steps of 2
+    const int q_offset =  32*im + l0;
+    const int y_offset = 128*im + l0;
+
+    uint16_t utmp[4];
+    const int8_t * s = (const int8_t *)utmp;
+
+    const uint16_t s_shift = 4*im;
+
+    for (int i = ix; i < num_blocks_per_row; i += K_QUANTS_PER_ITERATION) {
+
+        const float   * y  = yy + i * QK_K + y_offset;
+        const uint8_t * q = x[i].qs + q_offset;
+        const uint8_t * h = x[i].hmask + l0;
+
+        const uint16_t * a = (const uint16_t *)x[i].scales;
+        utmp[0] = ((a[0] >> s_shift) & kmask2) | (((a[4] >> (s_shift + 0)) & kmask1) << 4);
+        utmp[1] = ((a[1] >> s_shift) & kmask2) | (((a[5] >> (s_shift + 0)) & kmask1) << 4);
+        utmp[2] = ((a[2] >> s_shift) & kmask2) | (((a[4] >> (s_shift + 2)) & kmask1) << 4);
+        utmp[3] = ((a[3] >> s_shift) & kmask2) | (((a[5] >> (s_shift + 2)) & kmask1) << 4);
+
+        const float d = x[i].d;
+
+        float sum = 0;
+        for (int l = 0; l < n; ++l) {
+            sum += y[l+ 0] * (s[0] - 32) * (((q[l] >> 0) & 3) - (h[l] & (m << 0) ? 0 : 4))
+                 + y[l+32] * (s[2] - 32) * (((q[l] >> 2) & 3) - (h[l] & (m << 1) ? 0 : 4))
+                 + y[l+64] * (s[4] - 32) * (((q[l] >> 4) & 3) - (h[l] & (m << 2) ? 0 : 4))
+                 + y[l+96] * (s[6] - 32) * (((q[l] >> 6) & 3) - (h[l] & (m << 3) ? 0 : 4));
+            sum += y[l+16] * (s[1] - 32) * (((q[l+16] >> 0) & 3) - (h[l+16] & (m << 0) ? 0 : 4))
+                 + y[l+48] * (s[3] - 32) * (((q[l+16] >> 2) & 3) - (h[l+16] & (m << 1) ? 0 : 4))
+                 + y[l+80] * (s[5] - 32) * (((q[l+16] >> 4) & 3) - (h[l+16] & (m << 2) ? 0 : 4))
+                + y[l+112] * (s[7] - 32) * (((q[l+16] >> 6) & 3) - (h[l+16] & (m << 3) ? 0 : 4));
+        }
+        tmp += d * sum;
+
+    }
+
+    // sum up partial sums and write back result
+    tmp = warp_reduce_sum(tmp);
+
+    if (threadIdx.x == 0) {
+        dst[row] = tmp;
+    }
+}
+
+static __global__ void dequantize_mul_mat_vec_q4_k(const void * __restrict__ vx, const float * __restrict__ yy, float * __restrict__ dst, const int ncols, int nrows) {
+
+    const int row = blockIdx.x*blockDim.y + threadIdx.y;
+    if (row > nrows) return;
+    const int num_blocks_per_row = ncols / QK_K;
+    const int ib0 = row*num_blocks_per_row;
+
+    const block_q4_K * x = (const block_q4_K *)vx + ib0;
+
+    const uint16_t kmask1 = 0x3f3f;
+    const uint16_t kmask2 = 0x0f0f;
+    const uint16_t kmask3 = 0xc0c0;
+
+    const int tid = threadIdx.x/K_QUANTS_PER_ITERATION;  // 0...31 or 0...16
+    const int ix  = threadIdx.x%K_QUANTS_PER_ITERATION;  // 0 or 0,1
+
+    const int step = 8/K_QUANTS_PER_ITERATION;           // 8 or 4
+
+    const int il  = tid/step;                            // 0...3
+    const int ir  = tid - step*il;                       // 0...7 or 0...3
+    const int n   = 2 * K_QUANTS_PER_ITERATION;          // 2 or 4
+
+    const int im = il/2;  // 0 or 1. 0 computes 0,32 + 128,160, 1 computes 64,96 + 192,224
+    const int in = il%2;
+
+    const int l0 = n*(2*ir + in);
+    const int q_offset = 32*im + l0;
+    const int y_offset = 64*im + l0;
+
+    uint16_t aux[4];
+    const uint8_t * sc = (const uint8_t *)aux;
+
+#if K_QUANTS_PER_ITERATION == 2
+    uint32_t q32[4];
+    const uint8_t * q4 = (const uint8_t *)q32;
+#else
+    uint16_t q16[4];
+    const uint8_t * q4 = (const uint8_t *)q16;
+#endif
+
+    float tmp = 0; // partial sum for thread in warp
+
+    for (int i = ix; i < num_blocks_per_row; i += K_QUANTS_PER_ITERATION) {
+
+        const float   * y1 = yy + i*QK_K + y_offset;
+        const float   * y2 = y1 + 128;
+
+        const float dall = __low2half(x[i].dm);
+        const float dmin = __high2half(x[i].dm);
+
+        const uint16_t * a = (const uint16_t *)x[i].scales;
+        aux[0] = a[im+0] & kmask1;
+        aux[1] = a[im+2] & kmask1;
+        aux[2] = ((a[im+4] >> 0) & kmask2) | ((a[im+0] & kmask3) >> 2);
+        aux[3] = ((a[im+4] >> 4) & kmask2) | ((a[im+2] & kmask3) >> 2);
+
+#if K_QUANTS_PER_ITERATION == 2
+        const uint32_t * q1 = (const uint32_t *)(x[i].qs + q_offset);
+        const uint32_t * q2 = q1 + 16;
+
+        q32[0] = q1[0] & 0x0f0f0f0f;
+        q32[1] = q1[0] & 0xf0f0f0f0;
+        q32[2] = q2[0] & 0x0f0f0f0f;
+        q32[3] = q2[0] & 0xf0f0f0f0;
+
+        float4 s = {0.f, 0.f, 0.f, 0.f};
+        float smin = 0;
+        for (int l = 0; l < 4; ++l) {
+            s.x += y1[l] * q4[l+0]; s.y += y1[l+32] * q4[l+ 4];
+            s.z += y2[l] * q4[l+8]; s.w += y2[l+32] * q4[l+12];
+            smin += y1[l] * sc[2] + y1[l+32] * sc[3] + y2[l] * sc[6] + y2[l+32] * sc[7];
+        }
+        tmp += dall * (s.x * sc[0] + s.y * sc[1] * 1.f/16.f + s.z * sc[4] + s.w * sc[5] * 1.f/16.f) - dmin * smin;
+#else
+        const uint16_t * q1 = (const uint16_t *)(x[i].qs + q_offset);
+        const uint16_t * q2 = q1 + 32;
+
+        q16[0] = q1[0] & 0x0f0f;
+        q16[1] = q1[0] & 0xf0f0;
+        q16[2] = q2[0] & 0x0f0f;
+        q16[3] = q2[0] & 0xf0f0;
+
+        float4 s = {0.f, 0.f, 0.f, 0.f};
+        float smin = 0;
+        for (int l = 0; l < 2; ++l) {
+            s.x += y1[l] * q4[l+0]; s.y += y1[l+32] * q4[l+2];
+            s.z += y2[l] * q4[l+4]; s.w += y2[l+32] * q4[l+6];
+            smin += y1[l] * sc[2] + y1[l+32] * sc[3] + y2[l] * sc[6] + y2[l+32] * sc[7];
+        }
+        tmp += dall * (s.x * sc[0] + s.y * sc[1] * 1.f/16.f + s.z * sc[4] + s.w * sc[5] * 1.f/16.f) - dmin * smin;
+#endif
+
+    }
+
+    // sum up partial sums and write back result
+    tmp = warp_reduce_sum(tmp);
+
+    if (tid == 0) {
+        dst[row] = tmp;
+    }
+}
+
+static __global__ void dequantize_mul_mat_vec_q5_k(const void * __restrict__ vx, const float * __restrict__ yy, float * __restrict__ dst, const int ncols) {
+
+    const int row = blockIdx.x;
+    const int num_blocks_per_row = ncols / QK_K;
+    const int ib0 = row*num_blocks_per_row;
+
+    const block_q5_K * x = (const block_q5_K *)vx + ib0;
+
+    float tmp = 0; // partial sum for thread in warp
+
+    const uint16_t kmask1 = 0x3f3f;
+    const uint16_t kmask2 = 0x0f0f;
+    const uint16_t kmask3 = 0xc0c0;
+
+    const int tid = threadIdx.x/2;  // 0...15
+    const int ix  = threadIdx.x%2;
+
+    const int il  = tid/4;     // 0...3
+    const int ir  = tid - 4*il;// 0...3
+    const int n   = 2;
+
+    const int im = il/2;  // 0 or 1. 0 computes 0,32 + 128,160, 1 computes 64,96 + 192,224
+    const int in = il%2;
+
+    const int l0 = n*(2*ir + in);
+    const int q_offset = 32*im + l0;
+    const int y_offset = 64*im + l0;
+
+    const uint8_t hm1  = 1 << (2*im);
+    const uint8_t hm2  = hm1 << 4;
+
+    uint16_t aux[4];
+    const uint8_t * sc = (const uint8_t *)aux;
+
+    uint16_t q16[8];
+    const uint8_t * q4 = (const uint8_t *)q16;
+
+    for (int i = ix; i < num_blocks_per_row; i += 2) {
+
+        const uint8_t * ql1 = x[i].qs + q_offset;
+        const uint8_t * qh  = x[i].qh + l0;
+        const float   * y1  = yy + i*QK_K + y_offset;
+        const float   * y2  = y1 + 128;
+
+        const float dall = __low2half(x[i].dm);
+        const float dmin = __high2half(x[i].dm);
+
+        const uint16_t * a = (const uint16_t *)x[i].scales;
+        aux[0] = a[im+0] & kmask1;
+        aux[1] = a[im+2] & kmask1;
+        aux[2] = ((a[im+4] >> 0) & kmask2) | ((a[im+0] & kmask3) >> 2);
+        aux[3] = ((a[im+4] >> 4) & kmask2) | ((a[im+2] & kmask3) >> 2);
+
+        float4 sum = {0.f, 0.f, 0.f, 0.f};
+        float smin = 0;
+        const uint16_t * q1 = (const uint16_t *)ql1;
+        const uint16_t * q2 = q1 + 32;
+        q16[0] = q1[0] & 0x0f0f;
+        q16[1] = q1[8] & 0x0f0f;
+        q16[2] = (q1[0] >> 4) & 0x0f0f;
+        q16[3] = (q1[8] >> 4) & 0x0f0f;
+        q16[4] = q2[0] & 0x0f0f;
+        q16[5] = q2[8] & 0x0f0f;
+        q16[6] = (q2[0] >> 4) & 0x0f0f;
+        q16[7] = (q2[8] >> 4) & 0x0f0f;
+        for (int l = 0; l < n; ++l) {
+            sum.x += y1[l+ 0] * (q4[l +0] + (qh[l+ 0] & (hm1 << 0) ? 16 : 0))
+                   + y1[l+16] * (q4[l +2] + (qh[l+16] & (hm1 << 0) ? 16 : 0));
+            sum.y += y1[l+32] * (q4[l +4] + (qh[l+ 0] & (hm1 << 1) ? 16 : 0))
+                   + y1[l+48] * (q4[l +6] + (qh[l+16] & (hm1 << 1) ? 16 : 0));
+            sum.z += y2[l+ 0] * (q4[l +8] + (qh[l+ 0] & (hm2 << 0) ? 16 : 0))
+                   + y2[l+16] * (q4[l+10] + (qh[l+16] & (hm2 << 0) ? 16 : 0));
+            sum.w += y2[l+32] * (q4[l+12] + (qh[l+ 0] & (hm2 << 1) ? 16 : 0))
+                   + y2[l+48] * (q4[l+14] + (qh[l+16] & (hm2 << 1) ? 16 : 0));
+            smin += (y1[l] + y1[l+16]) * sc[2] + (y1[l+32] + y1[l+48]) * sc[3]
+                  + (y2[l] + y2[l+16]) * sc[6] + (y2[l+32] + y2[l+48]) * sc[7];
+        }
+        tmp += dall * (sum.x * sc[0] + sum.y * sc[1] + sum.z * sc[4] + sum.w * sc[5]) - dmin * smin;
+    }
+
+    // sum up partial sums and write back result
+    tmp = warp_reduce_sum(tmp);
+
+    if (threadIdx.x == 0) {
+        dst[row] = tmp;
+    }
+}
+
+static __global__ void dequantize_mul_mat_vec_q6_k(const void * __restrict__ vx, const float * __restrict__ yy, float * __restrict__ dst, const int ncols, int nrows) {
+
+    static_assert(16%K_QUANTS_PER_ITERATION == 0, "16 must be divisible by K_QUANTS_PER_ITERATION");
+
+    const int row = blockIdx.x*blockDim.y + threadIdx.y;
+    if (row > nrows) return;
+
+    const int num_blocks_per_row = ncols / QK_K;
+    const int ib0 = row*num_blocks_per_row;
+
+    const block_q6_K * x = (const block_q6_K *)vx + ib0;
+
+    const int tid = threadIdx.x/K_QUANTS_PER_ITERATION;  // 0...31 or 0...16
+    const int ix  = threadIdx.x%K_QUANTS_PER_ITERATION;  // 0 or 0, 1
+
+    const int step = 16/K_QUANTS_PER_ITERATION;          // 16 or 8
+
+    const int im = tid/step;                             // 0 or 1. 0 computes 0..., 1 computes 128...
+    const int in = tid - step*im;                        // 0...15 or 0...7
+
+#if K_QUANTS_PER_ITERATION == 1
+    const int l0 = K_QUANTS_PER_ITERATION*in;            // 0...15
+    const int is = 0;
+#else
+    const int l0 = 4 * in;                               // 0, 4, 8, ..., 28
+    const int is = in / 4;
+#endif
+    const int ql_offset = 64*im + l0;
+    const int qh_offset = 32*im + l0;
+    const int s_offset  =  8*im + is;
+    const int y_offset = 128*im + l0;
+
+    float tmp = 0; // partial sum for thread in warp
+
+    for (int i = ix; i < num_blocks_per_row; i += K_QUANTS_PER_ITERATION) {
+
+        const float   * y  = yy + i * QK_K + y_offset;
+        const uint8_t * ql = x[i].ql + ql_offset;
+        const uint8_t * qh = x[i].qh + qh_offset;
+        const int8_t  * s  = x[i].scales + s_offset;
+
+        const float d = x[i].d;
+
+#if K_QUANTS_PER_ITERATION == 1
+        float sum = y[ 0] * s[0] * d * ((int8_t)((ql[ 0] & 0xF) | ((qh[ 0] & 0x03) << 4)) - 32)
+                  + y[16] * s[1] * d * ((int8_t)((ql[16] & 0xF) | ((qh[16] & 0x03) << 4)) - 32)
+                  + y[32] * s[2] * d * ((int8_t)((ql[32] & 0xF) | ((qh[ 0] & 0x0c) << 2)) - 32)
+                  + y[48] * s[3] * d * ((int8_t)((ql[48] & 0xF) | ((qh[16] & 0x0c) << 2)) - 32)
+                  + y[64] * s[4] * d * ((int8_t)((ql[ 0]  >> 4) | ((qh[ 0] & 0x30) >> 0)) - 32)
+                  + y[80] * s[5] * d * ((int8_t)((ql[16]  >> 4) | ((qh[16] & 0x30) >> 0)) - 32)
+                  + y[96] * s[6] * d * ((int8_t)((ql[32]  >> 4) | ((qh[ 0] & 0xc0) >> 2)) - 32)
+                  +y[112] * s[7] * d * ((int8_t)((ql[48]  >> 4) | ((qh[16] & 0xc0) >> 2)) - 32);
+        tmp += sum;
+#else
+        float sum = 0;
+        for (int l = 0; l < 4; ++l) {
+            sum += y[l+ 0] * s[0] * d * ((int8_t)((ql[l+ 0] & 0xF) | (((qh[l] >> 0) & 3) << 4)) - 32)
+                 + y[l+32] * s[2] * d * ((int8_t)((ql[l+32] & 0xF) | (((qh[l] >> 2) & 3) << 4)) - 32)
+                 + y[l+64] * s[4] * d * ((int8_t)((ql[l+ 0]  >> 4) | (((qh[l] >> 4) & 3) << 4)) - 32)
+                 + y[l+96] * s[6] * d * ((int8_t)((ql[l+32]  >> 4) | (((qh[l] >> 6) & 3) << 4)) - 32);
+        }
+        tmp += sum;
+#endif
+
+    }
+
+    // sum up partial sums and write back result
+    tmp = warp_reduce_sum(tmp);
+
+    if (tid == 0) {
+        dst[row] = tmp;
+    }
+}
+
+static __device__ void convert_f16(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
+    const half * x = (const half *) vx;
+
+    // automatic half -> float type cast if dfloat == float
+    v.x = x[ib + iqs + 0];
+    v.y = x[ib + iqs + 1];
+}
+
+static constexpr __device__ dequantize_kernel_t get_dequantize_kernel(ggml_type type) {
+    return type == GGML_TYPE_Q4_0 ? dequantize_q4_0 :
+        type == GGML_TYPE_Q4_1 ? dequantize_q4_1 :
+        type == GGML_TYPE_Q5_0 ? dequantize_q5_0 :
+        type == GGML_TYPE_Q5_1 ? dequantize_q5_1 :
+        type == GGML_TYPE_Q8_0 ? dequantize_q8_0 :
+        type == GGML_TYPE_F16 ? convert_f16 :
+        nullptr;
+}
+
+template <ggml_type type>
+static __global__ void dequantize_mul_mat_vec(const void * __restrict__ vx, const dfloat * __restrict__ y, float * __restrict__ dst, const int ncols, const int nrows) {
+    constexpr int qk = ggml_cuda_type_traits<type>::qk; // quantized weights per x block
+    constexpr int qr = ggml_cuda_type_traits<type>::qr; // number of quantized weights per data value in x block
+    constexpr dequantize_kernel_t dequantize_kernel = get_dequantize_kernel(type);
+
+    const int64_t row = (int64_t)blockIdx.x*blockDim.y + threadIdx.y;
+
+    if (row >= nrows) {
+        return;
+    }
+
+    const int tid = threadIdx.x;
+
+    const int iter_stride = 2*GGML_CUDA_DMMV_X;
+    const int vals_per_iter = iter_stride / WARP_SIZE; // num quantized vals per thread and i iter
+    const int y_offset = qr == 1 ? 1 : qk/2;
+
+// partial sum for each thread
+#ifdef GGML_CUDA_F16
+    half2 tmp = {0.0f, 0.0f}; // two sums for f16 to take advantage of half2 intrinsics
+#else
+    float tmp = 0.0f;
+#endif // GGML_CUDA_F16
+
+    for (int i = 0; i < ncols; i += iter_stride) {
+        const int col = i + vals_per_iter*tid;
+        const int64_t ib = ((int64_t)row*ncols + col)/qk; // x block index
+        const int iqs = (col%qk)/qr; // x quant index
+        const int iybs = col - col%qk; // y block start index
+
+// processing >2 values per i iter is faster for fast GPUs
+#pragma unroll
+        for (int j = 0; j < vals_per_iter; j += 2) {
+            // process 2 vals per j iter
+
+            // dequantize
+            // for qr = 2 the iqs needs to increase by 1 per j iter because 2 weights per data val
+            dfloat2 v;
+            dequantize_kernel(vx, ib, iqs + j/qr, v);
+
+            // matrix multiplication
+            // for qr = 2 the y index needs to increase by 1 per j iter because of y_offset = qk/2
+#ifdef GGML_CUDA_F16
+            tmp += __hmul2(v, {
+                y[iybs + iqs + j/qr + 0],
+                y[iybs + iqs + j/qr + y_offset]
+            });
+#else
+            tmp += v.x * y[iybs + iqs + j/qr + 0];
+            tmp += v.y * y[iybs + iqs + j/qr + y_offset];
+#endif // GGML_CUDA_F16
+        }
+    }
+
+    // sum up partial sums and write back result
+    tmp = warp_reduce_sum(tmp);
+
+    if (tid == 0) {
+#ifdef GGML_CUDA_F16
+        dst[row] = tmp.x + tmp.y;
+#else
+        dst[row] = tmp;
+#endif // GGML_CUDA_F16
+    }
+}
+
+static void dequantize_mul_mat_vec_q4_0_cuda(const void * vx, const dfloat * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % (GGML_CUDA_DMMV_X*2) == 0);
+    const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y;
+    // the number of rows may exceed maximum grid size in the y or z dimensions, use the x dimension instead
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1);
+    dequantize_mul_mat_vec<GGML_TYPE_Q4_0>
+        <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q4_1_cuda(const void * vx, const dfloat * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % (GGML_CUDA_DMMV_X*2) == 0);
+    const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1);
+    dequantize_mul_mat_vec<GGML_TYPE_Q4_1>
+        <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q5_0_cuda(const void * vx, const dfloat * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % (GGML_CUDA_DMMV_X*2) == 0);
+    const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1);
+    dequantize_mul_mat_vec<GGML_TYPE_Q5_0>
+        <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q5_1_cuda(const void * vx, const dfloat * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % (GGML_CUDA_DMMV_X*2) == 0);
+    const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1);
+    dequantize_mul_mat_vec<GGML_TYPE_Q5_1>
+        <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q8_0_cuda(const void * vx, const dfloat * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % (GGML_CUDA_DMMV_X*2) == 0);
+    const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1);
+    dequantize_mul_mat_vec<GGML_TYPE_Q8_0>
+        <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q2_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % QK_K == 0);
+    const int ny = 2; // very slightly faster than 1 even when K_QUANTS_PER_ITERATION = 2
+    const int block_num_y = (nrows + ny - 1) / ny;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(32, ny, 1);
+    dequantize_mul_mat_vec_q2_k<<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q3_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % QK_K == 0);
+    const int ny = 2 / K_QUANTS_PER_ITERATION;
+    const int block_num_y = (nrows + ny - 1) / ny;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(32, ny, 1);
+    dequantize_mul_mat_vec_q3_k<<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q4_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % QK_K == 0);
+    const int ny = 2 / K_QUANTS_PER_ITERATION;
+    const int block_num_y = (nrows + ny - 1) / ny;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(32, ny, 1);
+    dequantize_mul_mat_vec_q4_k<<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void dequantize_mul_mat_vec_q5_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % QK_K == 0);
+    const dim3 block_dims(32, 1, 1);
+    dequantize_mul_mat_vec_q5_k<<<nrows, block_dims, 0, stream>>>(vx, y, dst, ncols);
+}
+
+static void dequantize_mul_mat_vec_q6_K_cuda(const void * vx, const float * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % QK_K == 0);
+    const int ny = 2 / K_QUANTS_PER_ITERATION;
+    const int block_num_y = (nrows + ny - 1) / ny;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(32, ny, 1);
+    dequantize_mul_mat_vec_q6_k<<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+static void convert_mul_mat_vec_f16_cuda(const void * vx, const dfloat * y, float * dst, const int ncols, const int nrows, cudaStream_t stream) {
+    GGML_ASSERT(ncols % (GGML_CUDA_DMMV_X*2) == 0);
+    const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y;
+    const dim3 block_nums(block_num_y, 1, 1);
+    const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1);
+    dequantize_mul_mat_vec<GGML_TYPE_F16>
+        <<<block_nums, block_dims, 0, stream>>>(vx, y, dst, ncols, nrows);
+}
+
+void ggml_cuda_op_dequantize_mul_mat_vec(
+    ggml_backend_cuda_context & ctx,
+    const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst, const char * src0_dd_i, const float * src1_ddf_i,
+    const char * src1_ddq_i, float * dst_dd_i, const int64_t row_low, const int64_t row_high, const int64_t src1_ncols,
+    const int64_t src1_padded_row_size, cudaStream_t stream) {
+    GGML_UNUSED(ctx);
+    const int64_t ne00 = src0->ne[0];
+    const int64_t row_diff = row_high - row_low;
+
+    GGML_ASSERT(src1->type == GGML_TYPE_F32);
+
+    // on some GPUs it is faster to convert src1 to half and to use half precision intrinsics
+#ifdef GGML_CUDA_F16
+    ggml_cuda_pool_alloc<half> src1_dfloat_a(ctx.pool());
+    half * src1_dfloat = nullptr; // dfloat == half
+
+    bool src1_convert_f16 =
+        src0->type == GGML_TYPE_Q4_0 || src0->type == GGML_TYPE_Q4_1 ||
+        src0->type == GGML_TYPE_Q5_0 || src0->type == GGML_TYPE_Q5_1 ||
+        src0->type == GGML_TYPE_Q8_0 || src0->type == GGML_TYPE_F16;
+
+    if (src1_convert_f16) {
+        src1_dfloat = src1_dfloat_a.alloc(ne00);
+        const to_fp16_cuda_t to_fp16_cuda = ggml_get_to_fp16_cuda(src1->type);
+        GGML_ASSERT(to_fp16_cuda != nullptr);
+        to_fp16_cuda(src1_ddf_i, src1_dfloat, ne00, stream);
+    }
+#else
+    const dfloat * src1_dfloat = (const dfloat *) src1_ddf_i; // dfloat == float, no conversion
+#endif // GGML_CUDA_F16
+
+    switch (src0->type) {
+        case GGML_TYPE_Q4_0:
+            dequantize_mul_mat_vec_q4_0_cuda(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q4_1:
+            dequantize_mul_mat_vec_q4_1_cuda(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q5_0:
+            dequantize_mul_mat_vec_q5_0_cuda(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q5_1:
+            dequantize_mul_mat_vec_q5_1_cuda(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q8_0:
+            dequantize_mul_mat_vec_q8_0_cuda(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q2_K:
+            dequantize_mul_mat_vec_q2_K_cuda(src0_dd_i, src1_ddf_i, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q3_K:
+            dequantize_mul_mat_vec_q3_K_cuda(src0_dd_i, src1_ddf_i, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q4_K:
+            dequantize_mul_mat_vec_q4_K_cuda(src0_dd_i, src1_ddf_i, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q5_K:
+            dequantize_mul_mat_vec_q5_K_cuda(src0_dd_i, src1_ddf_i, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_Q6_K:
+            dequantize_mul_mat_vec_q6_K_cuda(src0_dd_i, src1_ddf_i, dst_dd_i, ne00, row_diff, stream);
+            break;
+        case GGML_TYPE_F16:
+            convert_mul_mat_vec_f16_cuda(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
+            break;
+        default:
+            GGML_ABORT("fatal error");
+            break;
+    }
+
+    GGML_UNUSED(src1);
+    GGML_UNUSED(dst);
+    GGML_UNUSED(src1_ddq_i);
+    GGML_UNUSED(src1_ncols);
+    GGML_UNUSED(src1_padded_row_size);
+}
+
+bool ggml_cuda_dmmv_type_supported(ggml_type src0_type) {
+    return src0_type == GGML_TYPE_Q4_0 || src0_type == GGML_TYPE_Q4_1 ||
+        src0_type == GGML_TYPE_Q5_0 || src0_type == GGML_TYPE_Q5_1 ||
+        src0_type == GGML_TYPE_Q8_0 || src0_type == GGML_TYPE_Q2_K ||
+        src0_type == GGML_TYPE_Q3_K || src0_type == GGML_TYPE_Q4_K ||
+        src0_type == GGML_TYPE_Q5_K || src0_type == GGML_TYPE_Q6_K ||
+        src0_type == GGML_TYPE_F16;
+}
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
jmorganca	f443dd7b81	llama: sync llama.cpp to commit 8962422	2024-09-03 21:23:46 -04:00
Jesse Gross	8db94469e0	runner.go: Support GGUF LoRAs The current CGo bindings for loading LoRAs only supports the older GGLA file format, which is no longer supported. This switches to use functions that load the newer GGUF LoRAs.	2024-09-03 21:15:14 -04:00
Jesse Gross	c989321509	runner.go: Don't cast a Go handle to a C void * Cgo handles passing pointers through C to another Go function with integer handles to the Go memory. However, it is not legal to cast this handle to a void * aux data in C. Doing this results in panics about invalid pointers on the stack in certain circumstances. Instead, we should pass a pointer to the handle and pin that in memory. It would probably also be safe to directly pin the Go function pointer and pass that rather than using the handle since it is an opaque blob to C. However, using a handle is the more generally correct solution and there is no need to get clever.	2024-09-03 21:15:14 -04:00
Jesse Gross	e4a091bafd	runner.go: Support resource usage command line options Command line options to the runner that control resource usage (mmap, mlock, tensor split) are used by Ollama but not currently implemented. This implements support for these while ignoring others that have no meaning in this context.	2024-09-03 21:15:14 -04:00
Jeffrey Morgan	fd4ecd1ff5	llama: fix sync script ggml-metal_darwin_arm64.m filename (#6610 )	2024-09-03 21:15:14 -04:00
Jeffrey Morgan	9d8129b8bb	llama: delete unused files (#6523 )	2024-09-03 21:15:14 -04:00
Jesse Gross	c8a1741d9b	runner.go: Update TODOs	2024-09-03 21:15:14 -04:00
Jesse Gross	46a7c682f2	runner.go: Fix embeddings endpoint The embeddings endpoint only takes a single input and provides a single output, instead of multiple as the current implementation expected. Fixing this also allows the implementation to be simplified and a few embedding-specific issues to be addressed.	2024-09-03 21:15:14 -04:00
Jesse Gross	52e88ab7b3	runner.go: Health endpoint comments The health endpoint needs a little more work to show progress as Ollama expects but we can at least return the right status and have comments for the future.	2024-09-03 21:15:14 -04:00
Jesse Gross	4ca8579428	runner.go: Cleanups Variable names in Go style, removing commented out code and typo fixes	2024-09-03 21:15:14 -04:00
Jesse Gross	d022cfc9e6	runner.go: Move pieces[] into sequence pieces[] is used to cache pending responses and is currently being passed around to different functions. Move it into the sequences where it logically belongs.	2024-09-03 21:15:14 -04:00
Jesse Gross	6ccd0644e1	runner.go: Fix deadlock if a connection is closed during decoding If a connection is closed while a sequence is being decoded, tokens will continue to be added to the channel without anyone to read them. This will result in the sender blocking, which will in turn block all other decoding and sending for other sequences. This is not limited to just the connection between Ollama and the runner process. If the connection to the Ollama API is closed by the user then Ollama will close the connection to the runner, triggering this issue.	2024-09-03 21:15:14 -04:00
Jesse Gross	0b73cca386	runner.go: Fix resource leaks when removing sequences There are multiple causes and paths that result in a sequence ending. Not all of these free the sampling context or reset the pieces slice. This factors out the removal code so that all paths release resources.	2024-09-03 21:15:14 -04:00
Jesse Gross	55fb0633db	runner.go: Separate KV cache and context sizes Currently the entire KV cache is shared by all parallel requestors. This gives maximum resource utilization but there is a potential for overflow and unfairness if multiple requests are trying to use significant context. Instead, it is better to have a hard partition of KV cache space.	2024-09-03 21:15:14 -04:00
Jesse Gross	53b600921e	runner.go: Hold mutex for entire time when processing batch It is not safe to hold a mutex only while we are waiting for the condition variable to signal that a new sequence has been added. It's possible that a sequence could be added in the middle of batch processing. For example, if a new sequence is added while Decode() is running, it will get picked up for sampling, despite not having been added to the original batch. This change holds a mutex for the majority of the time when active processing is happening, releasing it only for a brief period each time around the loop. Depending on the workload and the scheduler is may result in unfairness between different requests. However, this was not actually observed in testing. This addresses the correctness issue - better performance and fairness can be achieved with additional improvements in the future.	2024-09-03 21:15:14 -04:00
Jesse Gross	8e1554c91d	runner.go: Scale batches to be processed by numParallel We should process a batch of tokens for each parallel request, rather than having a shared pool. Otherwise, a single request can fill the batch and then subsequent ones will fail or get starved. Server.cpp used the KV cache size allocated for each parallel request as the allocated size for the batch. This is the upper bound for the batch but since we know how many tokens we will actually put in a batch there is no need to over allocate.	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	f52d4b9879	Make new tokenizer logic conditional (#6395 ) Only use the new cgo tokenizer/detokenizer if we're using the new runners	2024-09-03 21:15:14 -04:00
Jesse Gross	76718ead40	runner.go: Support MinP parameter MinP is a user-facing parameter that is exposed that is exposed through the APIs but is not currently plumbed through.	2024-09-03 21:15:14 -04:00
Jesse Gross	90d25d3b0a	runner.go: Check for incomplete UTF-8 character Generated text can contain a partial multi-byte Unicode character at the end. Check for this and hold it over until the next token is produced.	2024-09-03 21:15:14 -04:00
Jesse Gross	477f529d26	runner.go: Implement RepeatLastN to penalize repeated tokens RepeatLastN is a user-facing parameter that is exposed that is exposed through the APIs but is not currently plumbed through.	2024-09-03 21:15:14 -04:00
Jesse Gross	eccd4dd8d2	runner.go: Use correct JSON field names for runners The fields for inference parameters are very similar between the Ollama API and Ollama/runners. However, some of the names are slightly different. For these fields (such as NumKeep and NumPredict), the values from Ollama were never read properly and defaults were always used. In the future, we can share a single interface rather than duplicating structs. However, this keeps the interface consistent with minimal changes in Ollama as long as we continue to use server.cpp	2024-09-03 21:15:14 -04:00
Jesse Gross	69cc5795a7	runner.go: Shift context window when KV cache space is exceeded Currently, once the KV cache is full, text generation stops. Instead, we should shift out the oldest context so that new generation can continue based on more recent context. This uses the algorithm from llama.cpp that is currently used by Ollama with the server.cpp code. There are others but they are never turned on through Ollama, so this restores parity. The algorithm is: - Retain a configurable number of tokens at the beginning (for things like beginning of sequence tokens - Drop the oldest half of the remaining tokens - Shift the remaining new tokens to the back of the cache	2024-09-03 21:15:14 -04:00
Jesse Gross	5a441d227a	runner.go: Don't decode if nothing has been added to the batch If nothing has been added to a batch then decoding will fail if attempted. This can happen, for example, if the run loop is woken up but we realize that we have the generation limit.	2024-09-03 21:15:14 -04:00
Jesse Gross	8aa97b5e83	llama.go: Advance though tokens when processing multiple batches If the number of input tokens exceeds the size of the batch, multiple batches will be submitted but they will all contain the first tokens. This processes the input tokens as expected so that each batch has the next set of tokens.	2024-09-03 21:15:14 -04:00
Jesse Gross	523d84c563	llama.go: Use dynamic buffer for TokenToPiece The cgo binding for llama_token_to_piece uses a fixed 12 byte buffer, which is usually but not always enough to hold a token. This increase the buffer size if needed, similar to what llama.cpp does internally.	2024-09-03 21:15:14 -04:00
Jesse Gross	ed19fad862	llama.go: Make batch memory allocation match configuration Batch size defaults to 512 but is configurable. However, llama.go uses a fixed size buffer, causing crashes is the batch size is increase. This changes the array size to follow the configuration.	2024-09-03 21:15:14 -04:00
Jesse Gross	5d34320b7c	runner.go: Fix off by one in batch size check When adding tokens to a batch, the index is zero based but is checked against being greater than the max batch size. This results in an out-of-bounds access when the final token is added.	2024-09-03 21:15:14 -04:00
Jesse Gross	1c36f36c41	llm: Fix array out-of-bounds memory access when tokenizing tokenize() passes a string length longer than the actual data into llama_tokenize(). This entire string length gets scanned in the C++ code despite there being a NULL terminator in the correct location (because it gets converted into std::string). The result is read of uninitialized memory, which depending on the contents of that memory fails the check for partial multi-byte UTF8 characters. In addition, if there is not enough space in the passed buffer for token output then llama_tokenize() returns the required space as a negative number. We should convert this to a positive number before reallocing. The first problem results in the following splat: libc++abi: terminating due to uncaught exception of type std::invalid_argument: failed to convert utf8 to codepoint SIGABRT: abort PC=0x193cd55f0 m=11 sigcode=0 signal arrived during cgo execution goroutine 27 gp=0x14000708700 m=11 mp=0x14000584908 [syscall]: runtime.cgocall(0x105549e68, 0x140000c6bf8) /opt/homebrew/Cellar/go/1.22.5/libexec/src/runtime/cgocall.go:157 +0x44 fp=0x140000c6bc0 sp=0x140000c6b80 pc=0x104b372c4 github.com/ollama/ollama/llm._Cfunc_llama_tokenize(0x15180f400, 0x152009a00, 0x5aa, 0x140002e8800, 0x5aa, 0x1, 0x1) _cgo_gotypes.go:270 +0x34 fp=0x140000c6bf0 sp=0x140000c6bc0 pc=0x104ef7664 github.com/ollama/ollama/llm.tokenize.func2(0x140001dd800?, 0x152009a00, 0x5aa, 0x1400012cdc0?) /Users/jesse/ollama/llm/llm.go:74 +0x8c fp=0x140000c6c50 sp=0x140000c6bf0 pc=0x104ef83cc github.com/ollama/ollama/llm.tokenize(0x140003f7da0, {0x140001dd800, 0x5a8}) /Users/jesse/ollama/llm/llm.go:74 +0xb4 fp=0x140000c6d90 sp=0x140000c6c50 pc=0x104ef7f94 github.com/ollama/ollama/llm.(llmServer).Tokenize(0x140000c6df8?, {0x105516574?, 0x5a8?}, {0x140001dd800?, 0x140000c6d00?}) /Users/jesse/ollama/llm/server.go:963 +0x2c fp=0x140000c6dc0 sp=0x140000c6d90 pc=0x104ef6b6c github.com/ollama/ollama/llm.LlamaServer.Tokenize-fm({0x105e876f0?, 0x140001e5c70?}, {0x140001dd800?, 0x140000350e0?}) <autogenerated>:1 +0x50 fp=0x140000c6e00 sp=0x140000c6dc0 pc=0x105532fc0 github.com/ollama/ollama/server.chatPrompt({0x105e876f0, 0x140001e5c70}, 0x14000616480, 0x140000c7508, 0x1400013e000, {0x1400014e008, 0x7, 0x7}, {0x0, 0x0, ...}) /Users/jesse/ollama/server/prompt.go:36 +0x2a0 fp=0x140000c7100 sp=0x140000c6e00 pc=0x1055165a0 github.com/ollama/ollama/server.(Server).ChatHandler(0x1400000e9c0, 0x1400011c100) /Users/jesse/ollama/server/routes.go:1340 +0x478 fp=0x140000c7610 sp=0x140000c7100 pc=0x105523318 github.com/ollama/ollama/server.(*Server).ChatHandler-fm(0x9?) <autogenerated>:1 +0x30 fp=0x140000c7630 sp=0x140000c7610 pc=0x105533130	2024-09-03 21:15:14 -04:00
Jesse Gross	0c2f95f3de	runner: Initialize numPredict numPredict is used to enforce a limit on the number of tokens to generate. Is it passed in from Ollama but it is never stored to be checked.	2024-09-03 21:15:14 -04:00
Jesse Gross	ebdf781397	server: Fix double free on runner subprocess error. If the runner subprocess encounters an error, it will close the HTTP connect, which causes Ollama to free the instance of the model that has open. When Ollama exits, it will again try to free the models for all of the runners that were open, resulting in a double free.	2024-09-03 21:15:14 -04:00
Jesse Gross	23c7c1326e	llm: Fix lint	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	8fe30d161c	Fix filename for non darwin arm builds	2024-09-03 21:15:14 -04:00
jmorganca	a483a4c4ed	lint	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	b267ab92b0	Add missing vendor headers to ggml sync	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	189ca38f1d	Wire up native source file dependencies This should make sure incremental builds correctly identify when to rebuild components based on which native files are modified.	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	80db43b7b4	Bump llama sync to 1e6f65	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	47b0e81219	fix dolphin-mistral	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	21947d5c1b	harden integration tests	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	751009a5d7	Runtime selection of new or old runners This adjusts the new runners to comingle with existing runners so we can use an env var to toggle the new runners on.	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	8527028bf4	Implement timings response in Go server This implements the fields necessary for `run --verbose` to generate timing information.	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	e0241118d0	Get embeddings working Truncation doesn't pass, but the other embeddings tests pass	2024-09-03 21:15:14 -04:00
Daniel Hiltgen	f97ee8c506	Fix parallel requests	2024-09-03 21:15:13 -04:00
Daniel Hiltgen	e9dd656ff5	Update sync with latest llama.cpp layout, and run against b3485	2024-09-03 21:15:13 -04:00
Daniel Hiltgen	6c0d892498	Prefix all build artifacts with an OS/ARCH dir This will help keep incremental builds from stomping on each other and make it easier to stitch together the final runner payloads	2024-09-03 21:15:13 -04:00
Daniel Hiltgen	13348e3629	Get linux building Still needs a bit more refinement to (auto)detect cuda/hip and fallback gracefully if not detected.	2024-09-03 21:15:13 -04:00
jmorganca	3d5a08c315	add note in readme	2024-09-03 21:15:13 -04:00
jmorganca	a29851bc9b	clean up metal code	2024-09-03 21:15:13 -04:00
jmorganca	8dda9293fa	fix `Makefile` on windows	2024-09-03 21:15:13 -04:00
jmorganca	b3c62dcafd	remove printing	2024-09-03 21:15:13 -04:00
jmorganca	9b8b7cd9b5	dont apply license to `stb_image.h` and `json.hpp`	2024-09-03 21:15:13 -04:00
jmorganca	1da6c40f4f	lint	2024-09-03 21:15:13 -04:00
jmorganca	76ca2de06e	update sync header	2024-09-03 21:15:13 -04:00
jmorganca	0eabc2e34d	remove unused script	2024-09-03 21:15:13 -04:00
jmorganca	dded27dcfa	fix metal	2024-09-03 21:15:13 -04:00
jmorganca	080b600865	add header to not edit	2024-09-03 21:15:13 -04:00
jmorganca	d6b6de9a5a	add header to not edit	2024-09-03 21:15:13 -04:00
jmorganca	24a741424f	fix build on windows	2024-09-03 21:15:13 -04:00
jmorganca	4d476d894e	fix `Makefile`	2024-09-03 21:15:13 -04:00
jmorganca	bd94ddfc56	fix `README.md`	2024-09-03 21:15:13 -04:00
jmorganca	f1f54c5bd5	fix `README.md`	2024-09-03 21:15:13 -04:00
jmorganca	18662d1180	consistent whitespace	2024-09-03 21:15:13 -04:00
jmorganca	3d1f3569cf	update `.gitattributes`	2024-09-03 21:15:13 -04:00
jmorganca	083a9e9b4e	link metal	2024-09-03 21:15:13 -04:00
jmorganca	d0703eaf44	wip	2024-09-03 21:15:13 -04:00
jmorganca	ce00e387c3	wip meta	2024-09-03 21:15:13 -04:00
jmorganca	763d7b601c	sync	2024-09-03 21:15:13 -04:00
jmorganca	4d0e6c55b0	remove perl docs	2024-09-03 21:15:13 -04:00
jmorganca	3375b82c56	remove build scripts	2024-09-03 21:15:13 -04:00
jmorganca	b8c1065ab6	remove need for perl	2024-09-03 21:15:13 -04:00
jmorganca	a632a04426	fix output	2024-09-03 21:15:13 -04:00
jmorganca	110f37ffb0	arch build	2024-09-03 21:15:13 -04:00
jmorganca	f2f03ff7f2	add temporary makefile	2024-09-03 21:15:13 -04:00
jmorganca	ba0ff1c46a	fix cuda and rocm builds	2024-09-03 21:15:13 -04:00
jmorganca	9966a055e5	fix cgo flags for darwin amd64	2024-09-03 21:15:13 -04:00
jmorganca	7aa7a3c1e5	remove `-fPIC` from `build_hipblas.sh`	2024-09-03 21:15:13 -04:00
jmorganca	de634b7fd7	fix issues with runner	2024-09-03 21:15:13 -04:00
jmorganca	795753be7e	move sync script back in for now	2024-09-03 21:15:13 -04:00
jmorganca	0eed68fed4	llama: sync	2024-09-03 21:15:13 -04:00
jmorganca	783134a3bb	update to `d5c938cd`	2024-09-03 21:15:13 -04:00
jmorganca	74a158a79e	add patches	2024-09-03 21:15:13 -04:00
jmorganca	8f79a2e86a	cleanup stop code	2024-09-03 21:15:13 -04:00
jmorganca	a4d402c403	fix `example`	2024-09-03 21:15:13 -04:00
jmorganca	e1dfc757b3	revert llm changes	2024-09-03 21:15:13 -04:00
jmorganca	7d0a452938	num predict	2024-09-03 21:15:13 -04:00
jmorganca	43efc893d7	basic progress	2024-09-03 21:15:13 -04:00
jmorganca	20afaae020	add more runner params	2024-09-03 21:15:13 -04:00
jmorganca	72f3fe4b94	truncate stop properly	2024-09-03 21:15:13 -04:00
jmorganca	a379d68aa9	wip stop tokens	2024-09-03 21:15:13 -04:00
jmorganca	b2ef3bf490	embeddings	2024-09-03 21:15:12 -04:00
jmorganca	ce15ed6d69	remove dependency on `llm`	2024-09-03 21:15:12 -04:00
jmorganca	c0b94376b2	grammar	2024-09-03 21:15:12 -04:00
jmorganca	72be8e27c4	sampling	2024-09-03 21:15:12 -04:00
jmorganca	d12db0568e	better `example` module, add port	2024-09-03 21:15:12 -04:00
jmorganca	ec17359a68	wip	2024-09-03 21:15:12 -04:00
jmorganca	fbc8572859	add `llava` to `runner`	2024-09-03 21:15:12 -04:00
jmorganca	87af27dac0	fix output in build_hipblas.sh	2024-09-03 21:15:12 -04:00
jmorganca	54f391309f	mods to `build_hipblas.sh` for linux	2024-09-03 21:15:12 -04:00
jmorganca	28bedcd807	wip	2024-09-03 21:15:12 -04:00
jmorganca	922d0acbdb	improve cuda and hipblas build scripts	2024-09-03 21:15:12 -04:00
jmorganca	b22d78720e	cuda linux	2024-09-03 21:15:12 -04:00
Jeffrey Morgan	905568a47f	Update README.md	2024-09-03 21:15:12 -04:00
Jeffrey Morgan	a15ac52fbe	Update README.md	2024-09-03 21:15:12 -04:00
jmorganca	9547aa53ff	disable log file	2024-09-03 21:15:12 -04:00
jmorganca	e29205ad6d	fix readme for llava	2024-09-03 21:15:12 -04:00
jmorganca	a8f91d3cc1	add llava	2024-09-03 21:15:12 -04:00
jmorganca	a9884ae136	llama: add clip dependencies	2024-09-03 21:15:12 -04:00
jmorganca	e37651cca0	add clip and parallel requests to the todo list	2024-09-03 21:15:12 -04:00
jmorganca	593d6836ab	fix cuda build	2024-09-03 21:15:12 -04:00
jmorganca	533a7e7d50	fix build on windows	2024-09-03 21:15:12 -04:00
jmorganca	0873d28b16	fix `ggml-metal.m` build constraints	2024-09-03 21:15:12 -04:00
jmorganca	bb795faa6c	fix `ggml-metal.m`	2024-09-03 21:15:12 -04:00
jmorganca	e86db9381a	`avx2` should only add `avx2`	2024-09-03 21:15:12 -04:00
jmorganca	4a5633e4bc	fix sync script	2024-09-03 21:15:12 -04:00
jmorganca	86f453252b	fix `ggml-metal.m`	2024-09-03 21:15:12 -04:00
jmorganca	dfd8f34806	fix `ggml-metal.m`	2024-09-03 21:15:12 -04:00
jmorganca	beb847b40f	add license headers	2024-09-03 21:15:12 -04:00
jmorganca	785f76d390	pre-patch	2024-09-03 21:15:12 -04:00
jmorganca	9fe48978a8	move `runner` package down	2024-09-03 21:15:12 -04:00
jmorganca	01ccbc07fe	replace static build in `llm`	2024-09-03 21:15:12 -04:00
jmorganca	ec09be97e8	fix build	2024-09-03 21:15:12 -04:00
jmorganca	6129f30479	wip...	2024-09-03 21:15:12 -04:00
jmorganca	eb1aa97961	rename `server` to `runner`	2024-09-03 21:15:12 -04:00
Jeffrey Morgan	5e921e06ac	Update README.md	2024-09-03 21:15:12 -04:00
Jeffrey Morgan	02089baf70	Update README.md	2024-09-03 21:15:12 -04:00
Jeffrey Morgan	870e91be76	Update README.md	2024-09-03 21:15:12 -04:00
Jeffrey Morgan	7ecc8e86c4	Update README.md	2024-09-03 21:15:12 -04:00
jmorganca	b1696e308e	Add missing hipcc flags	2024-09-03 21:15:12 -04:00
jmorganca	c646115b31	fix `.gitattributes`	2024-09-03 21:15:12 -04:00
jmorganca	0110994d06	Initial `llama` Go module	2024-09-03 21:15:12 -04:00
jmorganca	2ef3a217d1	add sync of llama.cpp	2024-09-03 21:15:12 -04:00
				`@@ -1 +0,0 @@`
				`This is here to make sure the build/ directory exists for the go:embed command`