test-mlperf-inference-llama2.yml

# This workflow will install Python dependencies, run tests and lint with a variety of Python versions
# For more information see: https://help.github.com/actions/language-and-framework-guides/using-python-with-github-actions

name: MLPerf inference LLAMA 2 70B

on:
  schedule:
    - cron: "30 19 * * 4"

jobs:
  build_reference:
    if: github.repository_owner == 'gateoverflow'
    runs-on: [ self-hosted, GO-i9, linux, x64 ]
    strategy:
      fail-fast: false
      matrix:
        python-version: [ "3.12" ]
        backend: [ "pytorch" ]
        device: [ "cpu" ]

    steps:
    - name: Install dependencies
      run: |
        source gh_action/bin/deactivate || python3 -m venv gh_action
        source gh_action/bin/activate
        export CM_REPOS=$HOME/GH_CM
        python3 -m pip install cm4mlops
        cm pull repo
        python3 -m pip install "huggingface_hub[cli]"
        huggingface-cli login --token ${{ secrets.HF_TOKEN }} --add-to-git-credential
    - name: Test MLPerf Inference LLAMA 2 70B reference implementation
      run: |
        cm run script --tags=run-mlperf,inference,_submission,_short --submitter="MLCommons" --model=llama2-70b-99 --implementation=reference --backend=${{ matrix.backend }} --category=datacenter --scenario=Offline --execution_mode=test --device=${{ matrix.device }}  --docker --quiet --test_query_count=1 --target_qps=1 --docker_it=no --docker_cm_repo=gateoverflow@cm4mlops --adr.compiler.tags=gcc --hw_name=gh_action --docker_dt=yes  --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --env.CM_MLPERF_MODEL_LLAMA2_70B_DOWNLOAD_TO_HOST=yes --adr.inference-src.tags=_repo.https://github.com/anandhu-eng/inference.git --clean