L0 pytorch wheel for functional test; gitignore

Author: BrianPei

.github/workflows/functional_tests_common.yml

@@ -49,7 +49,7 @@ jobs:
         uses: actions/checkout@v6
         with:
           fetch-depth: 0
-
+      
       # - name: Set safe directory
       #   run: |
       #     git config --global --add safe.directory $PROJECT_ROOT
@@ -66,7 +66,6 @@ jobs:
     
       - name: Setup Python environment
         env:
-          
           NVTE_WITH_MACA: '1'
           NVTE_WITH_CUDA: '0'
           NVCC: /opt/maca/bin/mcc
@@ -80,9 +79,22 @@ jobs:
           pip install -e . --no-deps --no-build-isolation
         timeout-minutes: 60
 
-      - name: Run functional tests
-        id: functional_test
+      - name: L0 Pytorch Wheel
+        id: L0_pytoech_wheel
+        # timeout-minutes: 50
+        env:
+          TE_PATH: .
+          RUN_LOG: /logs/pytorch/wheel
         run: |
+          echo "TE_PATH: ${TE_PATH}"
+          sed -i "s/^cd transformer_engine\/pytorch\s*$/pushd transformer_engine\/pytorch/" qa/L0_pytorch_wheel/test.sh
+          sed -i '44 s/^cd \s*\$TE_PATH\s*$/popd/' qa/L0_pytorch_wheel/test.sh
+
+          cat qa/L0_pytorch_wheel/test.sh
+          # source /opt/miniconda3/etc/profile.d/conda.sh
+          # conda activate flagscale-train
+          pip uninstall -y transformer_engine
+
           set -euo pipefail
           cd $PROJECT_ROOT
 
@@ -101,16 +113,9 @@ jobs:
 
           # Set environment variables
           export PYTHONPATH=$PROJECT_ROOT:${PYTHONPATH:-}
-
-          # Run functional tests via run_tests.sh with explicit platform/device/task/model/case
+          
           set +e
-          bash "$PROJECT_ROOT/tests/test_utils/runners/run_tests.sh" \
-            --platform "$PLATFORM" \
-            --device "$DEVICE" \
-            --type functional \
-            --task "$TASK" \
-            --model "$MODEL" \
-            --list "$CASE"
+          bash qa/L0_pytorch_wheel/test.sh | tee ${RUN_LOG}/pytorch_wheel-${{ github.run_id }}.log
           exit_code=$?
           set -e
 
@@ -122,7 +127,59 @@ jobs:
 
           echo "exit_code=$exit_code" >> $GITHUB_OUTPUT
           exit $exit_code
-        timeout-minutes: 60
+
+      - name: Upload Installation Logs
+        if: always() && steps.L0_pytoech_wheel.outcome == 'failure'
+        uses: actions/upload-artifact@v4
+        with:
+          name: L0-pytorch-logs-${{ github.run_id }}
+          path: /logs/pytorch/wheel
+          retention-days: 7
+          if-no-files-found: warn
+
+      # - name: Run functional tests
+      #   id: functional_test
+      #   run: |
+      #     set -euo pipefail
+      #     cd $PROJECT_ROOT
+
+      #     PLATFORM='${{ inputs.platform }}'
+      #     DEVICE='${{ matrix.test_config.device }}'
+      #     TASK='${{ matrix.test_config.task }}'
+      #     MODEL='${{ matrix.test_config.model }}'
+      #     CASE='${{ matrix.test_config.case }}'
+
+      #     echo "Running functional tests for training"
+      #     echo "Platform: $PLATFORM"
+      #     echo "Device: $DEVICE"
+      #     echo "Task: $TASK"
+      #     echo "Model: $MODEL"
+      #     echo "Case: ${CASE:-all}"
+
+      #     # Set environment variables
+      #     export PYTHONPATH=$PROJECT_ROOT:${PYTHONPATH:-}
+
+      #     # Run functional tests via run_tests.sh with explicit platform/device/task/model/case
+      #     set +e
+      #     bash "$PROJECT_ROOT/tests/test_utils/runners/run_tests.sh" \
+      #       --platform "$PLATFORM" \
+      #       --device "$DEVICE" \
+      #       --type functional \
+      #       --task "$TASK" \
+      #       --model "$MODEL" \
+      #       --list "$CASE"
+      #     exit_code=$?
+      #     set -e
+
+      #     if [ $exit_code -eq 0 ]; then
+      #       echo "✅ Functional tests passed for $PLATFORM/$DEVICE/$TASK/$MODEL/$CASE"
+      #     else
+      #       echo "❌ Functional tests failed for $PLATFORM/$DEVICE/$TASK/$MODEL/$CASE (exit code: $exit_code)"
+      #     fi
+
+      #     echo "exit_code=$exit_code" >> $GITHUB_OUTPUT
+      #     exit $exit_code
+      #   timeout-minutes: 60
 
       - name: Debug - keep container alive on failure
         if: failure()

.gitignore

@@ -41,4 +41,6 @@ compile_commands.json
 tensor_dumps/
 artifacts/
 # Auto-generated build configuration (specific to each environment)
-transformer_engine/plugin/core/_build_config.py
+transformer_engine/plugin/core/_build_config.py
+# Mac OS
+.DS_Store