pr_before_merge.yaml

name: pr-before-merge
on:
  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:
  model_init_tests:
    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_init_tests
      run: |
        source activate ${evo_env_torch21_flash2}
        export PYTHONPATH=$PWD:$PYTHONPATH
        jobname=internlm-${GITHUB_RUN_ID}-${GITHUB_JOB}-${GITHUB_RUN_ATTEMPT}
        srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=$jobname -N 1 -n 8 --ntasks-per-node=8 --gpus-per-task=1 python ./tests/test_training/train_CI.py --config ./tests/test_training/7B_check_init.py --seed=1024
        exit_code=$?
        sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname

  test_forward_output_no_fa:
    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: forward_output_no_fa
      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 ./tests/test_training/test_forward_output_no_fa.py
        exit_code=$?
        sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname

  unit_test_model:
    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: test_embedding_accuracy
      run: |
        source activate ${evo_env_torch21_flash2}
        jobname=embedding-${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 ./tests/test_model/test_embedding.py
        exit_code=$?
        sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname

    - name: test_model_internlm_accuracy
      run: |
        source activate ${evo_env_torch21_flash2}
        jobname=model-${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 ./tests/test_model/test_model_internlm.py
        exit_code=$?
        sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname

    - name: test_norm_accuracy
      run: |
        source activate ${evo_env_torch21_flash2}
        jobname=norm_${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 ./tests/test_model/test_norm.py
        exit_code=$?
        sh ./ci_scripts/common/check_slurm_cancled.sh $exit_code $jobname