Feat/refactor process group #1369
Workflow file for this run
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: unit-tests | |
| on: | |
| push: | |
| branches: | |
| - "develop" | |
| - "main" | |
| paths-ignore: | |
| - "cmds/**" | |
| - "**.md" | |
| pull_request: | |
| branches: | |
| - "develop" | |
| - "main" | |
| paths-ignore: | |
| - "cmds/**" | |
| - "**.md" | |
| env: | |
| WORKSPACE_PREFIX: $(echo $GITHUB_WORKSPACE |cut -d '/' -f 1-4) | |
| SLURM_PARTITION: llm_s | |
| jobs: | |
| unit_tests_core_pipeline: | |
| runs-on: [t_cluster] | |
| timeout-minutes: 20 | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - uses: actions/checkout@v3 | |
| - name: core_pipeline | |
| run: | | |
| source activate ${evo_env_torch21_flash2} | |
| export PYTHONPATH=$PWD:$PYTHONPATH | |
| jobname=${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT} | |
| srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 1 --gres=gpu:8 python -m pytest -s -v ./tests/test_core/test_pipeline.py | |
| exit_code=$? | |
| sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname | |
| unit_tests_utils_storage_manager: | |
| runs-on: [t_cluster] | |
| timeout-minutes: 10 | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - uses: actions/checkout@v3 | |
| - name: utils_storage_manager | |
| run: | | |
| source activate ${evo_env_torch21_flash2} | |
| export PYTHONPATH=$PWD:$PYTHONPATH | |
| jobname=${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT} | |
| srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 1 --gres=gpu:8 python -m pytest -s -v ./tests/test_utils/test_storage_manager.py | |
| exit_code=$? | |
| sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname | |
| unit_tests_model_fused_precision: | |
| runs-on: [t_cluster] | |
| timeout-minutes: 10 | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - uses: actions/checkout@v3 | |
| - name: model_fused_precision | |
| run: | | |
| source activate ${evo_env_torch21_flash2} | |
| export PYTHONPATH=$PWD:$PYTHONPATH | |
| jobname=${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT} | |
| srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 1 --gres=gpu:8 python -m pytest -s -v ./tests/test_model/test_fused_precision/test_fused_precision.py | |
| exit_code=$? | |
| sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname | |
| unit_tests_data_batch_sampler: | |
| runs-on: [t_cluster] | |
| timeout-minutes: 20 | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - uses: actions/checkout@v3 | |
| - name: data_batch_sample | |
| run: | | |
| source activate ${evo_env_torch21_flash2} | |
| export PYTHONPATH=$PWD:$PYTHONPATH | |
| jobname=${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT} | |
| srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 1 --gres=gpu:8 python -m pytest -s -v ./tests/test_data/test_batch_sampler.py | |
| exit_code=$? | |
| sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname | |
| unit_tests_utils_timeout: | |
| runs-on: [t_cluster] | |
| timeout-minutes: 10 | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - uses: actions/checkout@v3 | |
| - name: utils_timeout | |
| run: | | |
| source activate ${evo_env_torch21_flash2} | |
| export PYTHONPATH=$PWD:$PYTHONPATH | |
| jobname=${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT} | |
| srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 1 --gres=gpu:1 python -m pytest -s -v ./tests/test_utils/test_timeout.py | |
| exit_code=$? | |
| sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname | |
| unit_tests_utils_model_checkpoint: | |
| runs-on: [t_cluster] | |
| timeout-minutes: 10 | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - uses: actions/checkout@v3 | |
| - name: utils_model_checkpoint | |
| run: | | |
| source activate ${evo_env_torch21_flash2} | |
| export PYTHONPATH=$PWD:$PYTHONPATH | |
| jobname=${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT} | |
| srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 1 --gres=gpu:2 python -m pytest -s -v ./tests/test_utils/test_model_checkpoint.py | |
| exit_code=$? | |
| sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname | |
| notify_to_feishu: | |
| if: ${{ always() && !cancelled() && contains(needs.*.result, 'failure') && (github.ref_name == 'develop' || github.ref_name == 'main') }} | |
| needs: [ | |
| unit_tests_core_pipeline, | |
| unit_tests_utils_storage_manager, | |
| unit_tests_model_fused_precision, | |
| unit_tests_data_batch_sampler, | |
| unit_tests_utils_timeout, | |
| unit_tests_utils_model_checkpoint | |
| ] | |
| runs-on: [t_cluster] | |
| steps: | |
| - name: mask env | |
| run: | | |
| echo "::add-mask::${{env.WORKSPACE_PREFIX}}" | |
| echo "::add-mask::$path_prefix" | |
| - name: notify | |
| run: | | |
| curl -X POST -H "Content-Type: application/json" -d '{"msg_type":"post","content":{"post":{"zh_cn":{"title":"'${{ github.repository }}' GitHubAction Failed","content":[[{"tag":"text","text":""},{"tag":"a","text":"Please click here for details ","href":"https://github.com/'${{ github.repository }}'/actions/runs/'${GITHUB_RUN_ID}'"},{"tag":"at","user_id":"'${{ secrets.USER_ID }}'"}]]}}}}' ${{ secrets.WEBHOOK_URL }} |