trigger

.github/workflows/self-scheduled.yml

@@ -12,7 +12,7 @@ on:
     - cron: "17 2 * * *"
   push:
     branches:
-      - run_scheduled_ci*
+      - run_no_job_name
 
 env:
   HF_HOME: /mnt/cache
@@ -61,7 +61,7 @@ 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)')" >> $GITHUB_OUTPUT
+          echo "matrix=$(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[:2] + d1[:2]; print(d)')" >> $GITHUB_OUTPUT
 
       - name: NVIDIA-SMI
         run: |
@@ -121,7 +121,7 @@ jobs:
         continue-on-error: true
         run: cat /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}/failures_short.txt
 
-      - name: Test suite reports artifacts
+      - name: "Test suite reports artifacts: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports"
         if: ${{ always() }}
         uses: actions/upload-artifact@v3
         with:
@@ -182,7 +182,7 @@ jobs:
         continue-on-error: true
         run: cat /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}/failures_short.txt
 
-      - name: Test suite reports artifacts
+      - name: "Test suite reports artifacts: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports"
         if: ${{ always() }}
         uses: actions/upload-artifact@v3
         with:
@@ -233,174 +233,13 @@ jobs:
         continue-on-error: true
         run: cat /transformers/reports/${{ matrix.machine_type }}_examples_gpu/failures_short.txt
 
-      - name: Test suite reports artifacts
+      - name: "Test suite reports artifacts: ${{ matrix.machine_type }}_run_examples_gpu"
         if: ${{ always() }}
         uses: actions/upload-artifact@v3
         with:
           name: ${{ matrix.machine_type }}_run_examples_gpu
           path: /transformers/reports/${{ matrix.machine_type }}_examples_gpu
 
-  run_pipelines_torch_gpu:
-    name: PyTorch pipelines
-    strategy:
-      fail-fast: false
-      matrix:
-        machine_type: [single-gpu, multi-gpu]
-    runs-on: ['${{ matrix.machine_type }}', nvidia-gpu, t4, daily-ci]
-    container:
-      image: huggingface/transformers-pytorch-gpu
-      options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
-    needs: setup
-    steps:
-      - name: Update clone
-        working-directory: /transformers
-        run: git fetch && git checkout ${{ github.sha }}
-
-      - name: Reinstall transformers in edit mode (remove the one installed during docker image build)
-        working-directory: /transformers
-        run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
-
-      - name: NVIDIA-SMI
-        run: |
-          nvidia-smi
-
-      - name: Environment
-        working-directory: /transformers
-        run: |
-          python3 utils/print_env.py
-
-      - name: Show installed libraries and their versions
-        working-directory: /transformers
-        run: pip freeze
-
-      - name: Run all pipeline tests on GPU
-        working-directory: /transformers
-        run: |
-          python3 -m pytest -n 1 -v --dist=loadfile --make-reports=${{ matrix.machine_type }}_tests_torch_pipeline_gpu tests/pipelines
-
-      - name: Failure short reports
-        if: ${{ failure() }}
-        continue-on-error: true
-        run: cat /transformers/reports/${{ matrix.machine_type }}_tests_torch_pipeline_gpu/failures_short.txt
-
-      - name: Test suite reports artifacts
-        if: ${{ always() }}
-        uses: actions/upload-artifact@v3
-        with:
-          name: ${{ matrix.machine_type }}_run_tests_torch_pipeline_gpu
-          path: /transformers/reports/${{ matrix.machine_type }}_tests_torch_pipeline_gpu
-
-  run_pipelines_tf_gpu:
-    name: TensorFlow pipelines
-    strategy:
-      fail-fast: false
-      matrix:
-        machine_type: [single-gpu, multi-gpu]
-    runs-on: ['${{ matrix.machine_type }}', nvidia-gpu, t4, daily-ci]
-    container:
-      image: huggingface/transformers-tensorflow-gpu
-      options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
-    needs: setup
-    steps:
-      - name: Update clone
-        working-directory: /transformers
-        run: |
-          git fetch && git checkout ${{ github.sha }}
-
-      - name: Reinstall transformers in edit mode (remove the one installed during docker image build)
-        working-directory: /transformers
-        run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
-
-      - name: NVIDIA-SMI
-        run: |
-          nvidia-smi
-
-      - name: Environment
-        working-directory: /transformers
-        run: |
-          python3 utils/print_env.py
-
-      - name: Show installed libraries and their versions
-        working-directory: /transformers
-        run: pip freeze
-
-      - name: Run all pipeline tests on GPU
-        working-directory: /transformers
-        run: |
-          python3 -m pytest -n 1 -v --dist=loadfile --make-reports=${{ matrix.machine_type }}_tests_tf_pipeline_gpu tests/pipelines
-
-      - name: Failure short reports
-        if: ${{ always() }}
-        run: |
-          cat /transformers/reports/${{ matrix.machine_type }}_tests_tf_pipeline_gpu/failures_short.txt
-
-      - name: Test suite reports artifacts
-        if: ${{ always() }}
-        uses: actions/upload-artifact@v3
-        with:
-          name: ${{ matrix.machine_type }}_run_tests_tf_pipeline_gpu
-          path: /transformers/reports/${{ matrix.machine_type }}_tests_tf_pipeline_gpu
-
-  run_all_tests_torch_cuda_extensions_gpu:
-    name: Torch CUDA extension tests
-    strategy:
-      fail-fast: false
-      matrix:
-        machine_type: [single-gpu, multi-gpu]
-    runs-on: ['${{ matrix.machine_type }}', nvidia-gpu, t4, daily-ci]
-    needs: setup
-    container:
-      image: huggingface/transformers-pytorch-deepspeed-latest-gpu
-      options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
-    steps:
-      - name: Update clone
-        working-directory: /workspace/transformers
-        run: git fetch && git checkout ${{ github.sha }}
-
-      - name: Reinstall transformers in edit mode (remove the one installed during docker image build)
-        working-directory: /workspace/transformers
-        run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
-
-      - name: Remove cached torch extensions
-        run: rm -rf /github/home/.cache/torch_extensions/
-
-      # To avoid unknown test failures
-      - name: Pre build DeepSpeed *again*
-        working-directory: /workspace
-        run: |
-          python3 -m pip uninstall -y deepspeed
-          DS_BUILD_CPU_ADAM=1 DS_BUILD_FUSED_ADAM=1 python3 -m pip install deepspeed --global-option="build_ext" --global-option="-j8" --no-cache -v --disable-pip-version-check
-
-      - name: NVIDIA-SMI
-        run: |
-          nvidia-smi
-
-      - name: Environment
-        working-directory: /workspace/transformers
-        run: |
-          python utils/print_env.py
-
-      - name: Show installed libraries and their versions
-        working-directory: /workspace/transformers
-        run: pip freeze
-
-      - name: Run all tests on GPU
-        working-directory: /workspace/transformers
-        run: |
-          python -m pytest -v --make-reports=${{ matrix.machine_type }}_tests_torch_cuda_extensions_gpu tests/deepspeed tests/extended
-
-      - name: Failure short reports
-        if: ${{ failure() }}
-        continue-on-error: true
-        run: cat /workspace/transformers/reports/${{ matrix.machine_type }}_tests_torch_cuda_extensions_gpu/failures_short.txt
-
-      - name: Test suite reports artifacts
-        if: ${{ always() }}
-        uses: actions/upload-artifact@v3
-        with:
-          name: ${{ matrix.machine_type }}_run_tests_torch_cuda_extensions_gpu_test_reports
-          path: /workspace/transformers/reports/${{ matrix.machine_type }}_tests_torch_cuda_extensions_gpu
-
   run_extract_warnings:
     name: Extract warnings in CI artifacts
     runs-on: ubuntu-22.04
@@ -410,9 +249,6 @@ jobs:
       run_tests_single_gpu,
       run_tests_multi_gpu,
       run_examples_gpu,
-      run_pipelines_tf_gpu,
-      run_pipelines_torch_gpu,
-      run_all_tests_torch_cuda_extensions_gpu
     ]
     steps:
       - name: Checkout transformers
@@ -458,9 +294,6 @@ jobs:
       run_tests_single_gpu,
       run_tests_multi_gpu,
       run_examples_gpu,
-      run_pipelines_tf_gpu,
-      run_pipelines_torch_gpu,
-      run_all_tests_torch_cuda_extensions_gpu,
       run_extract_warnings
     ]
     steps:
@@ -478,7 +311,7 @@ jobs:
           CI_SLACK_CHANNEL_ID: ${{ secrets.CI_SLACK_CHANNEL_ID }}
           CI_SLACK_CHANNEL_ID_DAILY: ${{ secrets.CI_SLACK_CHANNEL_ID_DAILY }}
           CI_SLACK_CHANNEL_DUMMY_TESTS: ${{ secrets.CI_SLACK_CHANNEL_DUMMY_TESTS }}
-          CI_SLACK_REPORT_CHANNEL_ID: ${{ secrets.CI_SLACK_CHANNEL_ID_DAILY }}
+          CI_SLACK_REPORT_CHANNEL_ID: ${{ secrets.CI_SLACK_CHANNEL_DUMMY_TESTS }}
           ACCESS_REPO_INFO_TOKEN: ${{ secrets.ACCESS_REPO_INFO_TOKEN }}
           CI_EVENT: scheduled
           CI_SHA: ${{ github.sha }}

utils/get_ci_error_statistics.py

@@ -10,6 +10,32 @@ from collections import Counter
 import requests
 
 
+def get_jobs(workflow_run_id, token=None):
+    """Extract jobs in a GitHub Actions workflow run"""
+
+    headers = None
+    if token is not None:
+        headers = {"Accept": "application/vnd.github+json", "Authorization": f"Bearer {token}"}
+
+    url = f"https://api.github.com/repos/huggingface/transformers/actions/runs/{workflow_run_id}/jobs?per_page=100"
+    result = requests.get(url, headers=headers).json()
+    jobs = []
+
+    try:
+        jobs.extend(result["jobs"])
+        pages_to_iterate_over = math.ceil((result["total_count"] - 100) / 100)
+
+        for i in range(pages_to_iterate_over):
+            result = requests.get(url + f"&page={i + 2}", headers=headers).json()
+            jobs.extend(result["jobs"])
+
+        return jobs
+    except Exception:
+        print(f"Unknown error, could not fetch links:\n{traceback.format_exc()}")
+
+    return []
+
+
 def get_job_links(workflow_run_id, token=None):
     """Extract job names and their job links in a GitHub Actions workflow run"""
 

utils/notification_service.py

@@ -24,7 +24,7 @@ import time
 from typing import Dict, List, Optional, Union
 
 import requests
-from get_ci_error_statistics import get_job_links
+from get_ci_error_statistics import get_jobs
 from get_previous_daily_ci import get_last_daily_ci_reports
 from slack_sdk import WebClient
 
@@ -938,9 +938,19 @@ if __name__ == "__main__":
         Message.error_out(title, ci_title)
         raise ValueError("Errored out.")
 
-    github_actions_job_links = get_job_links(
+    github_actions_jobs = get_jobs(
         workflow_run_id=os.environ["GITHUB_RUN_ID"], token=os.environ["ACCESS_REPO_INFO_TOKEN"]
     )
+    github_actions_job_links = {job["name"]: job["html_url"] for job in github_actions_jobs}
+
+    artifact_name_to_job_map = {}
+    for job in github_actions_jobs:
+        for step in job["steps"]:
+            if step["name"].startswith("Test suite reports artifacts: "):
+                artifact_name = step["name"][len("Test suite reports artifacts: ") :]
+                artifact_name_to_job_map[artifact_name] = job
+                break
+
     available_artifacts = retrieve_available_artifacts()
 
     modeling_categories = [
@@ -974,32 +984,13 @@ if __name__ == "__main__":
 
     unclassified_model_failures = []
 
-    # This prefix is used to get job links below. For past CI, we use `workflow_call`, which changes the job names from
-    # `Model tests (...)` to `PyTorch 1.5 / Model tests (...)` for example.
-    job_name_prefix = ""
-    if ci_event.startswith("Past CI - "):
-        framework, version = ci_event.replace("Past CI - ", "").split("-")
-        framework = "PyTorch" if framework == "pytorch" else "TensorFlow"
-        job_name_prefix = f"{framework} {version}"
-    elif ci_event.startswith("Nightly CI"):
-        job_name_prefix = "Nightly CI"
-    elif ci_event.startswith("Push CI (AMD) - "):
-        flavor = ci_event.replace("Push CI (AMD) - ", "")
-        job_name_prefix = f"AMD {flavor}"
-    elif ci_event.startswith("Scheduled CI (AMD) - "):
-        flavor = ci_event.replace("Scheduled CI (AMD) - ", "")
-        job_name_prefix = f"AMD {flavor}"
-
     for model in model_results.keys():
         for artifact_path in available_artifacts[f"run_all_tests_gpu_{model}_test_reports"].paths:
             artifact = retrieve_artifact(artifact_path["path"], artifact_path["gpu"])
             if "stats" in artifact:
                 # Link to the GitHub Action job
-                # The job names use `matrix.folder` which contain things like `models/bert` instead of `models_bert`
-                job_name = f"Model tests ({model.replace('models_', 'models/')}, {artifact_path['gpu']}-gpu)"
-                if job_name_prefix:
-                    job_name = f"{job_name_prefix} / {job_name}"
-                model_results[model]["job_link"][artifact_path["gpu"]] = github_actions_job_links.get(job_name)
+                job = artifact_name_to_job_map[artifact_path["path"]]
+                model_results[model]["job_link"][artifact_path["gpu"]] = job["html_url"]
                 failed, success, time_spent = handle_test_results(artifact["stats"])
                 model_results[model]["success"] += success
                 model_results[model]["time_spent"] += time_spent[1:-1] + ", "
@@ -1084,12 +1075,8 @@ if __name__ == "__main__":
 
         for artifact_path in available_artifacts[additional_files[key]].paths:
             # Link to the GitHub Action job
-            job_name = key
-            if artifact_path["gpu"] is not None:
-                job_name = f"{key} ({artifact_path['gpu']}-gpu)"
-            if job_name_prefix:
-                job_name = f"{job_name_prefix} / {job_name}"
-            additional_results[key]["job_link"][artifact_path["gpu"]] = github_actions_job_links.get(job_name)
+            job = artifact_name_to_job_map[artifact_path["path"]]
+            additional_results[key]["job_link"][artifact_path["gpu"]] = job["html_url"]
 
             artifact = retrieve_artifact(artifact_path["path"], artifact_path["gpu"])
             stacktraces = handle_stacktraces(artifact["failures_line"])