Split daily CI using 2 level matrix (#28773)

* update / add new workflow files

* Add comment

* Use env.NUM_SLICES

* use scripts

* use scripts

* use scripts

* Fix

* using one script

* Fix

* remove unused file

* update

* fail-fast: false

* remove unused file

* fix

* fix

* use matrix

* inputs

* style

* update

* fix

* fix

* no model name

* add doc

* allow args

* style

* pass argument

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar
2024-01-31 18:04:43 +01:00
committed by GitHub
parent 95346e9dcd
commit 4735866141
4 changed files with 188 additions and 127 deletions

View File

@@ -27,6 +27,7 @@ env:
TF_FORCE_GPU_ALLOW_GROWTH: true
RUN_PT_TF_CROSS_TESTS: 1
CUDA_VISIBLE_DEVICES: 0,1
NUM_SLICES: 2
jobs:
setup:
@@ -39,7 +40,8 @@ jobs:
image: huggingface/transformers-all-latest-gpu
options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
outputs:
matrix: ${{ steps.set-matrix.outputs.matrix }}
folder_slices: ${{ steps.set-matrix.outputs.folder_slices }}
slice_ids: ${{ steps.set-matrix.outputs.slice_ids }}
steps:
- name: Update clone
working-directory: /transformers
@@ -61,133 +63,27 @@ jobs:
name: Identify models to test
working-directory: /transformers/tests
run: |
echo "matrix=$(python3 -c 'import os; tests = os.getcwd(); model_tests = os.listdir(os.path.join(tests, "models")); d1 = sorted(list(filter(os.path.isdir, os.listdir(tests)))); d2 = sorted(list(filter(os.path.isdir, [f"models/{x}" for x in model_tests]))); d1.remove("models"); d = d2 + d1; print(d)')" >> $GITHUB_OUTPUT
echo "folder_slices=$(python3 ../utils/split_model_tests.py --num_splits ${{ env.NUM_SLICES }})" >> $GITHUB_OUTPUT
echo "slice_ids=$(python3 -c 'd = list(range(${{ env.NUM_SLICES }})); print(d)')" >> $GITHUB_OUTPUT
- name: NVIDIA-SMI
run: |
nvidia-smi
run_tests_single_gpu:
name: Model tests
run_tests_gpu:
name: " "
needs: setup
strategy:
fail-fast: false
matrix:
folders: ${{ fromJson(needs.setup.outputs.matrix) }}
machine_type: [single-gpu]
runs-on: ['${{ matrix.machine_type }}', nvidia-gpu, t4, daily-ci]
container:
image: huggingface/transformers-all-latest-gpu
options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
needs: setup
steps:
- name: Echo folder ${{ matrix.folders }}
shell: bash
# For folders like `models/bert`, set an env. var. (`matrix_folders`) to `models_bert`, which will be used to
# set the artifact folder names (because the character `/` is not allowed).
run: |
echo "${{ matrix.folders }}"
matrix_folders=${{ matrix.folders }}
matrix_folders=${matrix_folders/'models/'/'models_'}
echo "$matrix_folders"
echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
- name: Update clone
working-directory: /transformers
run: git fetch && git checkout ${{ github.sha }}
- name: Reinstall transformers in edit mode (remove the one installed during docker image build)
working-directory: /transformers
run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
- name: NVIDIA-SMI
run: |
nvidia-smi
- name: Environment
working-directory: /transformers
run: |
python3 utils/print_env.py
- name: Show installed libraries and their versions
working-directory: /transformers
run: pip freeze
- name: Run all tests on GPU
working-directory: /transformers
run: python3 -m pytest -v --make-reports=${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }} tests/${{ matrix.folders }}
- name: Failure short reports
if: ${{ failure() }}
continue-on-error: true
run: cat /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}/failures_short.txt
- name: "Test suite reports artifacts: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports"
if: ${{ always() }}
uses: actions/upload-artifact@v3
with:
name: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports
path: /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}
run_tests_multi_gpu:
name: Model tests
strategy:
fail-fast: false
matrix:
folders: ${{ fromJson(needs.setup.outputs.matrix) }}
machine_type: [multi-gpu]
runs-on: ['${{ matrix.machine_type }}', nvidia-gpu, t4, daily-ci]
container:
image: huggingface/transformers-all-latest-gpu
options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
needs: setup
steps:
- name: Echo folder ${{ matrix.folders }}
shell: bash
# For folders like `models/bert`, set an env. var. (`matrix_folders`) to `models_bert`, which will be used to
# set the artifact folder names (because the character `/` is not allowed).
run: |
echo "${{ matrix.folders }}"
matrix_folders=${{ matrix.folders }}
matrix_folders=${matrix_folders/'models/'/'models_'}
echo "$matrix_folders"
echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
- name: Update clone
working-directory: /transformers
run: git fetch && git checkout ${{ github.sha }}
- name: Reinstall transformers in edit mode (remove the one installed during docker image build)
working-directory: /transformers
run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
- name: NVIDIA-SMI
run: |
nvidia-smi
- name: Environment
working-directory: /transformers
run: |
python3 utils/print_env.py
- name: Show installed libraries and their versions
working-directory: /transformers
run: pip freeze
- name: Run all tests on GPU
working-directory: /transformers
run: python3 -m pytest -v --make-reports=${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }} tests/${{ matrix.folders }}
- name: Failure short reports
if: ${{ failure() }}
continue-on-error: true
run: cat /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}/failures_short.txt
- name: "Test suite reports artifacts: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports"
if: ${{ always() }}
uses: actions/upload-artifact@v3
with:
name: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports
path: /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}
machine_type: [single-gpu, multi-gpu]
slice_id: ${{ fromJSON(needs.setup.outputs.slice_ids) }}
uses: ./.github/workflows/model_jobs.yml
with:
folder_slices: ${{ needs.setup.outputs.folder_slices }}
machine_type: ${{ matrix.machine_type }}
slice_id: ${{ matrix.slice_id }}
secrets: inherit
run_examples_gpu:
name: Examples directory
@@ -407,8 +303,7 @@ jobs:
if: always()
needs: [
setup,
run_tests_single_gpu,
run_tests_multi_gpu,
run_tests_gpu,
run_examples_gpu,
run_pipelines_tf_gpu,
run_pipelines_torch_gpu,
@@ -455,8 +350,7 @@ jobs:
if: always()
needs: [
setup,
run_tests_single_gpu,
run_tests_multi_gpu,
run_tests_gpu,
run_examples_gpu,
run_pipelines_tf_gpu,
run_pipelines_torch_gpu,
@@ -490,7 +384,7 @@ jobs:
sudo apt-get install -y curl
pip install slack_sdk
pip show slack_sdk
python utils/notification_service.py "${{ needs.setup.outputs.matrix }}"
python utils/notification_service.py "${{ needs.setup.outputs.folder_slices }}"
# Upload complete failure tables, as they might be big and only truncated versions could be sent to Slack.
- name: Failure table artifacts