test_accelerate_gaudi2.yml

name: Accelerate integration tests

on:
  workflow_dispatch:
  pull_request:
    branches: [main]
  push:
    branches: [main]

concurrency:
  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
  cancel-in-progress: true

jobs:
  accelerate:
    name: Test Accelerate integration
    runs-on: [self-hosted, linux, x64, gaudi2, fast]

    container:
      image: docker://vault.habana.ai/gaudi-docker/1.19.0/ubuntu22.04/habanalabs/pytorch-installer-2.5.1:latest
      options: --runtime=habana --shm-size=64G --cap-add=sys_nice
        --env OMPI_MCA_btl_vader_single_copy_mechanism=none
        --env HABANA_VISIBLE_DEVICES

    steps:
      - name: Checkout to Accelerate
        uses: actions/checkout@v4
        with:
          ref: hpu-support
          repository: huggingface/accelerate

      - name: HL-SMI
        run: |
          echo "HABANA_VISIBLE_DEVICES=${HABANA_VISIBLE_DEVICES}"
          echo "HABANA_VISIBLE_MODULES=${HABANA_VISIBLE_MODULES}"
          hl-smi

      - name: Install Accelerate with DeepSpeed
        run: |
          pip install -e .[testing] git+https://github.com/HabanaAI/DeepSpeed.git@1.19.0

      # Remove when safetensors is released
      - name: Install Rust and build Safetensors
        run: |
          curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
          . "$HOME/.cargo/env"
          rustup update
          git clone https://github.com/huggingface/safetensors
          cd safetensors
          pip install setuptools_rust
          git checkout fa833511664338bfc927fc02653ddb7d38d40be9
          pip install -e bindings/python
          cd ..

      - name: Set environment variables
        run: |
          export PT_ENABLE_INT64_SUPPORT=1
          export PT_HPU_LAZY_MODE=0
          export RUN_SLOW=1

      # move to the end of the file
      - name: Run Examples tests
        run: |
          make test_examples

      - name: Run CLI tests
        run: |
          make test_cli

      - name: Run Core tests
        run: |
          make test_core

      - name: Run Big Modeling tests
        run: |
          make test_big_modeling

      - name: Run FSDP integration tests
        run: |
          make test_fsdp

      - name: Run DeepSpeed integration tests
        run: |
          make test_deepspeed