ci: Update unit tests to run on self-hosted runner

.github/workflows/_run_test.yml

@@ -39,6 +39,9 @@ on:
         description: Failure will cancel all other tests if set to true
         required: false
         default: false
+    secrets:
+      HF_TOKEN:
+        required: true
     outputs:
       conclusion:
         description: Conclusion of main test step
@@ -60,24 +63,26 @@ jobs:
 
         - name: Docker pull image
           run: |
-            docker pull nemoci.azurecr.io/nemo__placeholder_container:${{ github.run_id }}
+            docker pull nemoci.azurecr.io/nemo_reinforcer_container:${{ github.run_id }}
 
         - name: Start container
           run: |
             docker run --rm -d --name nemo_container_${{ github.run_id }} --runtime=nvidia --gpus all --shm-size=64g \
               --env TRANSFORMERS_OFFLINE=0 \
               --env HYDRA_FULL_ERROR=1 \
-              --env HF_HOME=/home/TestData/_placeholder/hf_home \
-              --env _PLACEHOLDER_CI_DIR=/home/TestData/_placeholder \
-              --env _PLACEHOLDER_REPO_DIR=/opt/NeMo-_Placeholder \
-              --volume /mnt/datadrive/TestData/_placeholder/checkpoints:/home/TestData/_placeholder/checkpoints:ro \
-              --volume /mnt/datadrive/TestData/_placeholder/hf_home/hub:/home/TestData/_placeholder/hf_home/hub:ro \
-              nemoci.azurecr.io/nemo__placeholder_container:${{ github.run_id }} \
+              --env HF_HOME=/home/TestData/reinforcer/hf_home \
+              --env REINFORCER_CI_DIR=/home/TestData/reinforcer \
+              --env REINFORCER_REPO_DIR=/opt/NeMo-Reinforcer \
+              --volume /mnt/datadrive/TestData/reinforcer/checkpoints:/home/TestData/reinforcer/checkpoints:ro \
+              --volume /mnt/datadrive/TestData/reinforcer/hf_home/hub:/home/TestData/reinforcer/hf_home/hub \
+              nemoci.azurecr.io/nemo_reinforcer_container:${{ github.run_id }} \
               bash -c "sleep $(( ${{ inputs.TIMEOUT }} * 60 + 60 ))"
 
         - id: main
           name: Run main script
           timeout-minutes: ${{ inputs.TIMEOUT }}
+          env:
+            HF_TOKEN: ${{ secrets.HF_TOKEN }}
           run: |
             # Print the host driver for debugging
             nvidia-smi
@@ -90,14 +95,13 @@ jobs:
 
             cmd=$(cat <<"RUN_TEST_EOF"
             nvidia-smi
-            # Sanity check the driver/cuda combo
-            cudaCheck
-            # In case git commands need to be run inside _Placeholder
-            git config --global --add safe.directory $_PLACHOLDER_REPO_DIR
+
+            # In case git commands need to be run inside Reinforcer
+            git config --global --add safe.directory $REINFORCER_REPO_DIR
             ${{ inputs.SCRIPT }}
             RUN_TEST_EOF
             )
-            docker exec nemo_container_${{ github.run_id }} bash -eux -o pipefail -c "$cmd"
+            docker exec -u root -e HF_TOKEN nemo_container_${{ github.run_id }} bash -eux -o pipefail -c "$cmd"
             ) 2> >(tee err.log)
 
             EXIT_CODE=$?

.github/workflows/cicd-main.yml

@@ -11,7 +11,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-name: "CICD NeMo _Placeholder"
+name: "CICD Reinforcer"
 
 on:
   pull_request:
@@ -101,23 +101,30 @@ jobs:
           pip install pre-commit
           pre-commit install
           pre-commit run --all-files --show-diff-on-failure --color=always
+
+  build-container:
+    if: ${{ needs.pre-flight.outputs.run_ci == 'true' }}
+    needs: [pre-flight]
+    uses: NVIDIA/NeMo-FW-CI-templates/.github/workflows/_build_container.yml@v0.22.7
+    with:
+      build-ref: ${{ github.sha }}
+      image-name: nemo_reinforcer_container
+      dockerfile: docker/Dockerfile
+      image-label: nemo-reinforcer
+      build-args: |
+        MAX_JOBS=32
+        REINFORCER_COMMIT=${{ github.sha }}
 
   unit-tests:
     name: Unit tests
-    needs: [pre-flight]
-    runs-on: ubuntu-latest
+    needs: [build-container, pre-flight]
+    uses: ./.github/workflows/_run_test.yml
     if: ${{ needs.pre-flight.outputs.run_ci == 'true' }}
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@v4
-      - name: Run unit tests
-        run: |
-          pip install uv
-          uv venv -p python3.10 .venv
-          uv pip install --force-reinstall .
-          uv run --group test -- pytest
-
-      - name: after_script
-        if: always()
-        run: |
-          rm -rf .venv
+    with:
+      RUNNER: self-hosted-azure
+      TIMEOUT: 10
+      SCRIPT: |
+        cd ${REINFORCER_REPO_DIR}
+        uv run bash -x ./tests/run_unit.sh
+    secrets:
+      HF_TOKEN: ${{ secrets.HF_TOKEN }}

.github/workflows/release-freeze.yml

@@ -36,8 +36,8 @@ jobs:
   code-freeze:
     uses: NVIDIA/NeMo-FW-CI-templates/.github/workflows/_code_freeze.yml@v0.22.5
     with:
-      library-name: NeMo-_Placeholder
-      python-package: nemo__placeholder
+      library-name: NeMo-reinforcer
+      python-package: nemo_reinforcer
       release-type: ${{ inputs.release-type }}
       freeze-commit: ${{ inputs.freeze-commit }}
       dry-run: ${{ inputs.dry-run }}

.github/workflows/release.yaml

@@ -11,7 +11,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-name: "Release _Placeholder"
+name: "Release Reinforcer"
 
 on:
   workflow_dispatch:
@@ -35,9 +35,9 @@ jobs:
     uses: NVIDIA/NeMo-FW-CI-templates/.github/workflows/_release_library.yml@v0.22.6
     with:
       release-ref: ${{ inputs.release-ref }}
-      python-package: nemo__placeholder
+      python-package: nemo_reinforcer
       python-version: "3.11"
-      library-name: NeMo-_Placeholder
+      library-name: NeMo-Reinforcer
       dry-run: ${{ inputs.dry-run }}
       version-bump-branch: ${{ inputs.version-bump-branch }}
     secrets:

CONTRIBUTING.md

@@ -37,8 +37,8 @@ We follow a direct clone and branch workflow for now:
 
 1. Clone the repository directly:
    ```bash
-   git clone https://github.com/NVIDIA/nemo__placeholder
-   cd nemo-reinforcer
+   git clone https://github.com/NVIDIA/reinforcer
+   cd reinforcer
    ```
 
 2. Create a new branch for your changes:

docker/Dockerfile

@@ -1,7 +1,18 @@
 ARG BASE_IMAGE=anyscale/ray:2.43.0-py312-cu125
 FROM ${BASE_IMAGE}
 
+WORKDIR /opt/NeMo-Reinforcer
+
 RUN sudo apt-get update && sudo apt-get install -y jq
 
-RUN pip install --no-cache-dir uv
 RUN echo "unset RAY_RUNTIME_ENV_HOOK" >> /home/ray/.bashrc
+
+COPY pyproject.toml .
+
+RUN pip install uv && \
+    uv venv -p python3.12 && \
+    uv pip install -r pyproject.toml --extra dev --extra test
+
+COPY . .
+
+RUN uv pip install -e .

nemo_reinforcer/__init__.py

@@ -0,0 +1,13 @@
+from nemo_reinforcer.package_info import (
+    __contact_emails__,
+    __contact_names__,
+    __description__,
+    __download_url__,
+    __homepage__,
+    __keywords__,
+    __license__,
+    __package_name__,
+    __repository_url__,
+    __shortversion__,
+    __version__,
+)

nemo__placeholder/package_info.py → nemo_reinforcer/package_info.py

@@ -24,12 +24,12 @@
 __shortversion__ = ".".join(map(str, VERSION[:3]))
 __version__ = ".".join(map(str, VERSION[:3])) + "".join(VERSION[3:])
 
-__package_name__ = "nemo__placeholder"
+__package_name__ = "nemo_reinforcer"
 __contact_names__ = "NVIDIA"
-__contact_emails__ = "nemo-_placeholder@nvidia.com"
+__contact_emails__ = "nemo-tookit@nvidia.com"
 __homepage__ = "https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/stable/"
-__repository_url__ = "https://github.com/nvidia/nemo__placeholder"
-__download_url__ = "https://github.com/NVIDIA/NeMo__placeholder"
-__description__ = "_placeholder"
+__repository_url__ = "https://github.com/NVIDIA/reinforcer"
+__download_url__ = "https://github.com/NVIDIA/reinforcer/releases"
+__description__ = "NeMo-Reinforcer - a toolkit for model alignment"
 __license__ = "Apache2"
-__keywords__ = "_placeholder"
+__keywords__ = "deep learning, machine learning, gpu, NLP, NeMo, nvidia, pytorch, torch, language, reinforcement learning, RLHF, preference modeling, SteerLM, DPO"