deepspeed.yaml

name: huawei-ascend-npu

defaults:
  run:
    shell: bash -ieo pipefail {0}
on:
  workflow_dispatch:
  pull_request:
    paths:
      - '.github/workflows/deepspeed.yaml'
      - 'requirements/**'
  schedule:
    - cron: "0 0 * * *"
  push:
    paths:
      - '.github/workflows/deepspeed.yaml'

  
concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

permissions:
  contents: read
  issues: write

jobs:
  unit-tests:
    runs-on: [self-hosted, ascend, npu]
    container:
      image: ascendai/cann
      ports:
        - 80
      volumes:
        - /usr/local/dcmi:/usr/local/dcmi
        - /usr/local/bin/npu-smi:/usr/local/bin/npu-smi 
        - /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/
        - /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info
        - /etc/ascend_install.info:/etc/ascend_install.info   
      options: --network host
               --name deepspeed_unit-tests
               --device /dev/davinci6
               --device /dev/davinci_manager
               --device /dev/devmm_svm
               --device /dev/hisi_hdc
               --shm-size "20g"
               --entrypoint /bin/bash
                              
    steps:
    - uses: actions/checkout@v4

    - name: Install pytorch 
      run: |
        npu-smi info
        apt-get update 
        apt-get install sudo
        pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
        source /root/.bashrc

        pip install torch==2.2.0 torchvision==0.17.0 torch_npu==2.2.0 torchaudio==2.2.0 numpy==1.26.4 cloudpickle tornado ml-dtypes
        
        python << EOF 
        if __name__ == '__main__':
            import torch
            import torch_npu
            torch_npu.npu.set_device("npu:0") 
            print(f"Device Name: {torch.npu.get_device_name(0)}")
            print(f"Device Count: {torch.npu.device_count()}")
            print(f"Device Available: {torch.npu.is_available()}")
        EOF

    - name: Install transformers
      run: |
        source /root/.bashrc
        echo "y" | apt-get install git
        git clone https://github.com/huggingface/transformers
        cd transformers
        git rev-parse --short HEAD
        pip install .
             
    - name: Install deepspeed
      uses: nick-fields/retry@v3
      with:
        timeout_minutes: 15
        max_attempts: 3
        retry_on: error
        command: | 
          source /root/.bashrc
          git clone --depth=1 https://github.com/microsoft/DeepSpeed.git
          pip install -r requirements/requirements_deepspeed.txt 
          cd DeepSpeed
          pip install .[1bit,autotuning,inf]   
  
          ds_report

    - name: Python environment
      run: |
        source /root/.bashrc
        pip list

    - name: Unit tests
      run: |
        source /root/.bashrc
        unset TORCH_CUDA_ARCH_LIST # only jit compile for current arch
        cd DeepSpeed/tests/unit/
        
        pytest --verbose accelerator/*
        pytest --verbose autotuning/*
        pytest --verbose checkpoint/test_reshape_checkpoint.py
        pytest --verbose checkpoint/test_moe_checkpoint.py
        pytest --verbose checkpoint/test_shared_weights.py
        pytest --verbose launcher/test_ds_arguments.py launcher/test_run.py
        pytest --verbose model_parallelism/*
        pytest --verbose moe/test_moe_tp.py
        pytest --verbose monitor/*
        pytest --verbose utils/*
        pytest --verbose runtime/test_ds_config_model.py
        pytest --verbose runtime/pipe/test_pipe_schedule.py
        pytest --verbose runtime/zero/test_zero_config.py
        pytest --verbose runtime/zero/test_zero_tiled.py
        pytest --verbose runtime/zero/test_zeropp.py
        pytest --verbose runtime/test_autocast.py
        pytest --verbose runtime/test_data.py
        pytest --verbose runtime/test_runtime_utils.py
        pytest --verbose runtime/activation_checkpointing/*
        pytest --verbose runtime/utils/*
        pytest --verbose runtime/zero/test_zero_dynamic_class.py