update

.github/actions/llm/example-test-436/action.yml

@@ -0,0 +1,14 @@
+name: 'IPEX-LLM example tests'
+description: 'IPEX-LLM example tests'
+
+runs:
+  using: "composite"
+  steps:
+    - name: Test LLAMA2
+      shell: bash
+      env:
+        INT4_CKPT_DIR: ./llm/ggml-actions/stable
+        LLM_DIR: ./llm
+
+      run: |
+        bash python/llm/dev/test/run-example-tests-436.sh

.github/workflows/llm-nightly-test-436.yml

@@ -0,0 +1,113 @@
+name: LLM Nightly Tests for 4.36
+
+# Cancel previous runs in the PR when you push new commits
+concurrency:
+  group: ${{ github.workflow }}-llm-nightly-test-${{ github.event.pull_request.number || github.run_id }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+# Controls when the action will run.
+on:
+  schedule:
+    - cron: "00 13 * * *" # GMT time, 13:00 GMT == 21:00 China
+  pull_request:
+    branches: [main]
+    paths:
+#      - ".github/workflows/llm-nightly-test.yml"
+#      - ".github/actions/llm/setup-llm-env/action.yml"
+#      - ".github/actions/llm/remove-llm-env/action.yml"
+#      - ".github/actions/llm/convert-test/action.yml"
+  # Allows you to run this workflow manually from the Actions tab
+  workflow_dispatch:
+
+# A workflow run is made up of one or more jobs that can run sequentially or in parallel
+jobs:
+  llm-cpp-build:
+    uses: ./.github/workflows/llm-binary-build.yml
+  llm-nightly-convert-test:
+    needs: llm-cpp-build
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - os: windows
+            instruction: AVX-VNNI-UT
+            python-version: "3.11"
+          - os: ubuntu-20.04-lts
+            instruction: avx512
+            python-version: "3.11"
+    runs-on: [self-hosted, llm, "${{matrix.instruction}}", "${{matrix.os}}"]
+    env:
+      ANALYTICS_ZOO_ROOT: ${{ github.workspace }}
+    steps:
+      - name: Set model directories
+        shell: bash
+        run: |
+          echo "ORIGIN_DIR=$(pwd)/../llm/origin-models" >> "$GITHUB_ENV"
+          echo "INT4_CKPT_DIR=$(pwd)/../llm/nightly-converted-models" >> "$GITHUB_ENV"
+      - name: Create model directories
+        shell: bash
+        run: |
+          if [ ! -d $ORIGIN_DIR ]; then
+            mkdir -p $ORIGIN_DIR
+          fi
+          if [ ! -d $INT4_CKPT_DIR ]; then
+            mkdir -p $INT4_CKPT_DIR
+          fi
+      - name: Set environment variables
+        shell: bash
+        run: |
+          echo "LLAMA_ORIGIN_PATH=${ORIGIN_DIR}/llama-7b-hf" >> "$GITHUB_ENV"
+          echo "GPTNEOX_ORIGIN_PATH=${ORIGIN_DIR}/gptneox-7b-redpajama-bf16" >> "$GITHUB_ENV"
+          echo "BLOOM_ORIGIN_PATH=${ORIGIN_DIR}/bloomz-7b1" >> "$GITHUB_ENV"
+          echo "STARCODER_ORIGIN_PATH=${ORIGIN_DIR}/gpt_bigcode-santacoder" >> "$GITHUB_ENV"
+
+          echo "LLAMA_INT4_CKPT_PATH=${INT4_CKPT_DIR}/bigdl_llm_llama_q4_0.bin" >> "$GITHUB_ENV"
+          echo "GPTNEOX_INT4_CKPT_PATH=${INT4_CKPT_DIR}/bigdl_llm_gptneox_q4_0.bin" >> "$GITHUB_ENV"
+          echo "BLOOM_INT4_CKPT_PATH=${INT4_CKPT_DIR}/bigdl_llm_bloom_q4_0.bin" >> "$GITHUB_ENV"
+          echo "STARCODER_INT4_CKPT_PATH=${INT4_CKPT_DIR}/bigdl_llm_starcoder_q4_0.bin" >> "$GITHUB_ENV"
+      - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # actions/checkout@v3
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install dependencies
+        shell: bash
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install --upgrade setuptools==58.0.4
+          python -m pip install --upgrade wheel
+
+      - name: Download llm binary
+        uses: ./.github/actions/llm/download-llm-binary
+
+      - name: Install IPEX-LLM
+        uses: ./.github/actions/llm/setup-llm-env
+
+      - name: Download original models & convert
+        uses: ./.github/actions/llm/convert-test
+
+      - name: Upload ckpt to ftp
+        shell: bash
+        if: runner.os == 'Linux' && github.event_name == 'schedule'
+        run: |
+          curl -T $LLAMA_INT4_CKPT_PATH ${LLM_FTP_URL}/llm/ggml-actions/nightly/bigdl_llm_llama_7b_q4_0.bin
+          curl -T $GPTNEOX_INT4_CKPT_PATH ${LLM_FTP_URL}/llm/ggml-actions/nightly/bigdl_llm_redpajama_7b_q4_0.bin
+          curl -T $BLOOM_INT4_CKPT_PATH ${LLM_FTP_URL}/llm/ggml-actions/nightly/bigdl_llm_bloom_7b_q4_0.bin
+          curl -T $STARCODER_INT4_CKPT_PATH ${LLM_FTP_URL}/llm/ggml-actions/nightly/bigdl_llm_santacoder_1b_q4_0.bin
+      - name: Delete ckpt
+        shell: bash
+        run: |
+          rm -rf $LLAMA_INT4_CKPT_PATH
+          rm -rf $GPTNEOX_INT4_CKPT_PATH
+          rm -rf $BLOOM_INT4_CKPT_PATH
+          rm -rf $STARCODER_INT4_CKPT_PATH
+
+  llm-unit-tests:
+    needs: llm-cpp-build
+    uses: ./.github/workflows/llm_unit_tests_4.36.yml
+  llm-example-test:
+    needs: llm-cpp-build
+    uses: ./.github/workflows/llm_example_tests_436.yml

.github/workflows/llm_example_tests_436.yml

@@ -0,0 +1,81 @@
+name: LLM Example Test for 4.36
+
+# Cancel previous runs in the PR when you push new commits
+concurrency:
+  group: ${{ github.workflow }}-llm-example-tests-${{ github.event.pull_request.number || github.run_id }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+# Controls when the action will run. 
+on:
+  # schedule:
+  #   - cron: '00 13 * * *' # GMT time, 13:00 GMT == 21:00 China
+  pull_request:
+    branches: [ main ]
+    paths:
+      - '.github/workflows/llm_example_tests.yml'
+      - '.github/workflows/llm-binary-build.yml'
+      - '.github/actions/llm/example-test/action.yml'
+      - '.github/actions/llm/setup-llm-env/action.yml'
+      - '.github/actions/llm/remove-llm-env/action.yml'
+      - '.github/actions/llm/download-llm-binary/action.yml'
+      - 'python/llm/dev/test/run-example-tests.sh'
+      - 'python/llm/example/**'
+  workflow_dispatch:
+  workflow_call:
+
+env:
+  INT4_CKPT_DIR: ./llm/ggml-actions/stable
+  LLM_DIR: ./llm
+
+# A workflow run is made up of one or more jobs that can run sequentially or in parallel
+jobs:
+  llm-cpp-build:
+    uses: ./.github/workflows/llm-binary-build.yml
+  llm-example-test:
+    needs: llm-cpp-build
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.11"]
+        instruction: ["AVX512"]
+    runs-on: [ self-hosted, llm,"${{matrix.instruction}}", ubuntu-20.04-lts ]
+    env:
+      THREAD_NUM: 24
+    steps:
+      - uses: actions/checkout@ee0669bd1cc54295c223e0bb666b733df41de1c5 # actions/checkout@v2
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install --upgrade setuptools==58.0.4
+          python -m pip install --upgrade wheel
+
+#      - name: Download llm binary
+#        uses: ./.github/actions/llm/download-llm-binary
+#
+#      - name: Run LLM install (all) test
+#        uses: ./.github/actions/llm/setup-llm-env
+#        env:
+#          ANALYTICS_ZOO_ROOT: ${{ github.workspace }}
+
+      - name: Install IPEX-LLM from Pypi
+        shell: bash
+        run: |
+          pip install --pre --upgrade ipex-llm[all]
+          pip install transformers==4.36.2
+
+      - name: Run LLM example test
+        uses: ./.github/actions/llm/example-test
+        env:
+          ANALYTICS_ZOO_ROOT: ${{ github.workspace }}
+
+      # - name: Clean up test environment
+      #   uses: ./.github/actions/llm/remove-llm-env
+      #   env:
+      #     ANALYTICS_ZOO_ROOT: ${{ github.workspace }}