NVIDIA-NeMo
diff --git a/‎.github/workflows/install-test.yml‎
Lines changed: 219 additions & 0 deletions b/‎.github/workflows/install-test.yml‎
Lines changed: 219 additions & 0 deletions
diff --git a/‎docker/Dockerfile.ci‎
Lines changed: 13 additions & 0 deletions b/‎docker/Dockerfile.ci‎
Lines changed: 13 additions & 0 deletions
diff --git a/‎nemo_deploy/deploy_pytriton.py‎
Lines changed: 12 additions & 3 deletions b/‎nemo_deploy/deploy_pytriton.py‎
Lines changed: 12 additions & 3 deletions
diff --git a/‎nemo_deploy/deploy_ray.py‎
Lines changed: 18 additions & 4 deletions b/‎nemo_deploy/deploy_ray.py‎
Lines changed: 18 additions & 4 deletions
diff --git a/‎nemo_deploy/multimodal/query_multimodal.py‎
Lines changed: 38 additions & 5 deletions b/‎nemo_deploy/multimodal/query_multimodal.py‎
Lines changed: 38 additions & 5 deletions
@@ -0,0 +1,219 @@
+# Copyright (c) 2025, NVIDIA CORPORATION.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+
+# This workflow verifies that the basic install works across all supported platforms.
+# For basic install, all imports need to either be successful or appropriately guarded.
+
+name: Installation Test
+
+on:
+  push:
+    branches:
+      - main
+      - "pull-request/[0-9]+"
+env:
+  UV_PROJECT_ENVIRONMENT: "./venv"
+
+jobs:
+  pip-test:
+    runs-on: ${{ matrix.arch }}
+    name: Pip - Python${{ matrix.python-version }} - ${{ matrix.arch == 'ubuntu-latest' && 'AMD64/Linux' || 'ARM64/Darwin' }} - No CUDA
+    strategy:
+      fail-fast: false
+      matrix:
+        arch: ["ubuntu-latest", "macos-latest"]
+        python-version: ["3.10", "3.11", "3.12"]
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Upgrade pip
+        run: |
+          python -m pip install --upgrade pip
+
+      - name: Install project
+        run: |
+          pip install .
+
+      - name: Checkout check-imports
+        uses: actions/checkout@v4
+        with:
+          repository: NVIDIA-NeMo/FW-CI-templates
+          ref: v0.39.0
+          path: FW-CI-templates
+
+      - name: Check imports for nemo_deploy
+        uses: ./FW-CI-templates/.github/actions/check-imports
+        with:
+          package-name: nemo_deploy
+          python-binary: python
+
+      - name: Check imports for nemo_export
+        uses: ./FW-CI-templates/.github/actions/check-imports
+        with:
+          package-name: nemo_export
+          python-binary: python
+
+  ngc-cuda-test:
+    runs-on: linux-amd64-cpu16
+    name: UV - Python${{ matrix.python-version }} - AMD64/Linux - NGC CUDA
+    container:
+      image: nvcr.io/nvidia/cuda:12.8.1-cudnn-devel-ubuntu24.04
+    environment: nemo-ci
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.12"]
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install wget
+        run: |
+          apt-get update
+          apt-get install -y wget git libopenmpi-dev
+
+      - name: Upgrade pip
+        run: |
+          python -m pip install --upgrade pip
+
+      - name: Set up UV
+        uses: astral-sh/setup-uv@v1
+        with:
+          version: 0.7.2
+
+      - name: Set up yq
+        shell: bash
+        run: |
+          wget https://github.com/mikefarah/yq/releases/download/v4.45.4/yq_linux_amd64.tar.gz
+          tar -xzf yq_linux_amd64.tar.gz
+          mkdir -p ./bin
+          mv yq_linux_amd64 ./bin/yq 
+          chmod +x ./bin/yq
+
+      - name: Install project
+        env:
+          PAT: ${{ secrets.PAT }}
+        shell: bash
+        run: |
+          echo -e "machine github.com\n  login token\n  password ${{ secrets.PAT }}" > ~/.netrc
+          chmod 600 ~/.netrc 
+
+          uv venv ${UV_PROJECT_ENVIRONMENT} --system-site-packages
+          source ./venv/bin/activate
+
+          export PATH="./bin/:$PATH"
+
+          uv sync --link-mode copy --locked --all-groups
+
+          bash ./docker/common/install_conflicting_deps.sh
+
+          uv pip install --no-deps -e .
+
+      - name: Checkout check-imports
+        uses: actions/checkout@v4
+        with:
+          repository: NVIDIA-NeMo/FW-CI-templates
+          ref: v0.39.0
+          path: FW-CI-templates
+
+      - name: Check imports for nemo_deploy
+        uses: ./FW-CI-templates/.github/actions/check-imports
+        with:
+          package-name: nemo_deploy
+          python-binary: ./venv/bin/python
+
+      - name: Check imports for nemo_export
+        uses: ./FW-CI-templates/.github/actions/check-imports
+        with:
+          package-name: nemo_export
+          python-binary: ./venv/bin/python
+
+  uv-test:
+    runs-on: ${{ matrix.arch }}
+    name: UV - Python ${{ matrix.python-version }} - ${{ matrix.arch == 'ubuntu-latest' && 'AMD64/Linux' || 'ARM64/Darwin' }} - No CUDA
+    environment: nemo-ci
+    strategy:
+      fail-fast: false
+      matrix:
+        arch: ["ubuntu-latest", "macos-latest"]
+        python-version: ["3.12"]
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Set up UV
+        uses: astral-sh/setup-uv@v1
+        with:
+          version: 0.7.2
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install dependencies with UV
+        env:
+          PAT: ${{ secrets.PAT }}
+        shell: bash
+        run: |
+          echo -e "machine github.com\n  login token\n  password ${{ secrets.PAT }}" > ~/.netrc
+          chmod 600 ~/.netrc 
+
+          export PATH="${UV_PROJECT_ENVIRONMENT}/bin/:$PATH"
+
+          uv venv ${UV_PROJECT_ENVIRONMENT} --system-site-packages
+
+          uv sync --link-mode copy --locked --only-group build
+
+          uv sync --link-mode copy --locked --all-groups
+
+          uv pip install --no-deps -e .
+
+      - name: Checkout check-imports
+        uses: actions/checkout@v4
+        with:
+          repository: NVIDIA-NeMo/FW-CI-templates
+          ref: v0.39.0
+          path: FW-CI-templates
+
+      - name: Check imports for nemo_deploy
+        uses: ./FW-CI-templates/.github/actions/check-imports
+        with:
+          package-name: nemo_deploy
+          python-binary: ./venv/bin/python
+
+      - name: Check imports for nemo_export
+        uses: ./FW-CI-templates/.github/actions/check-imports
+        with:
+          package-name: nemo_export
+          python-binary: ./venv/bin/python
+
+  install-test-summary:
+    needs: [pip-test, uv-test, ngc-cuda-test]
+    runs-on: ubuntu-latest
+    name: Install test summary
+    steps:
+      - name: Result
+        run: echo Install check successful
@@ -53,6 +53,19 @@ RUN --mount=type=bind,source=pyproject.toml,target=/workspace/pyproject.toml \
         --extra te \
         --no-install-package torch \
         --no-install-package torchvision \
+        --no-install-package triton \
+        --no-install-package nvidia-cublas-cu12 \
+        --no-install-package nvidia-cuda-cupti-cu12 \
+        --no-install-package nvidia-cuda-nvrtc-cu12 \
+        --no-install-package nvidia-cuda-runtime-cu12 \
+        --no-install-package nvidia-cudnn-cu12 \
+        --no-install-package nvidia-cufft-cu12 \
+        --no-install-package nvidia-cufile-cu12 \
+        --no-install-package nvidia-curand-cu12 \
+        --no-install-package nvidia-cusolver-cu12 \
+        --no-install-package nvidia-cusparse-cu12 \
+        --no-install-package nvidia-cusparselt-cu12 \
+        --no-install-package nvidia-nccl-cu12 \
         --all-groups \
         ${INFERENCE_FRAMEWORK:+--extra $INFERENCE_FRAMEWORK}
 
 
@@ -15,13 +15,19 @@
 
 import logging
 
-from pytriton.model_config import ModelConfig
-from pytriton.triton import Triton, TritonConfig
-
 from nemo_deploy.deploy_base import DeployBase
+from nemo_export_deploy_common.import_utils import MISSING_TRITON_MSG, UnavailableError
 
 LOGGER = logging.getLogger("NeMo")
 
+try:
+    from pytriton.model_config import ModelConfig
+    from pytriton.triton import Triton, TritonConfig
+
+    HAVE_TRITON = True
+except (ImportError, ModuleNotFoundError):
+    HAVE_TRITON = False
+
 
 class DeployPyTriton(DeployBase):
     """Deploys any models to Triton Inference Server that implements ITritonDeployable interface in nemo_deploy.
@@ -99,6 +105,9 @@ def __init__(
 
     def deploy(self):
         """Deploys any models to Triton Inference Server."""
+        if not HAVE_TRITON:
+            raise UnavailableError(MISSING_TRITON_MSG)
+
         try:
             if self.streaming:
                 triton_config = TritonConfig(
 
@@ -15,11 +15,22 @@
 
 import logging
 
-import ray
-from ray import serve
-from ray.serve import Application
-
 from nemo_deploy.ray_utils import find_available_port
+from nemo_export_deploy_common.import_utils import MISSING_RAY_MSG, UnavailableError
+
+try:
+    import ray
+    from ray import serve
+    from ray.serve import Application
+
+    HAVE_RAY = True
+except (ImportError, ModuleNotFoundError):
+    from unittest.mock import MagicMock
+
+    ray = MagicMock()
+    serve = MagicMock()
+    Application = MagicMock()
+    HAVE_RAY = False
 
 LOGGER = logging.getLogger("NeMo")
 
@@ -61,6 +72,9 @@ def __init__(
         Raises:
             Exception: If Ray is not installed.
         """
+        if not HAVE_RAY:
+            raise UnavailableError(MISSING_RAY_MSG)
+
         # Initialize Ray with proper configuration
 
         try:
 
@@ -16,17 +16,38 @@
 
 import numpy as np
 import requests
-from PIL import Image
-from pytriton.client import ModelClient
 
 from nemo_deploy.utils import str_list2numpy
+from nemo_export_deploy_common.import_utils import (
+    MISSING_DECORD_MSG,
+    MISSING_PIL_MSG,
+    MISSING_TRITON_MSG,
+    UnavailableError,
+)
+
+try:
+    from PIL import Image
+
+    HAVE_PIL = True
+except (ImportError, ModuleNotFoundError):
+    HAVE_PIL = False
 
 try:
     from decord import VideoReader
-except Exception:
-    import logging
 
-    logging.warning("The package `decord` was not installed in this environment.")
+    HAVE_DECORD = True
+except (ImportError, ModuleNotFoundError):
+    HAVE_DECORD = False
+
+try:
+    from pytriton.client import ModelClient
+
+    HAVE_TRITON = True
+except (ImportError, ModuleNotFoundError):
+    from unittest.mock import MagicMock
+
+    ModelClient = MagicMock()
+    HAVE_TRITON = False
 
 
 class NemoQueryMultimodal:
@@ -57,16 +78,25 @@ def __init__(self, url, model_name, model_type):
     def setup_media(self, input_media):
         """Setup input media."""
         if self.model_type == "video-neva":
+            if not HAVE_DECORD:
+                raise UnavailableError(MISSING_DECORD_MSG)
+
             vr = VideoReader(input_media)
             frames = [f.asnumpy() for f in vr]
             return np.array(frames)
         elif self.model_type == "lita" or self.model_type == "vita":
+            if not HAVE_DECORD:
+                raise UnavailableError(MISSING_DECORD_MSG)
+
             vr = VideoReader(input_media)
             frames = [f.asnumpy() for f in vr]
             subsample_len = self.frame_len(frames)
             sub_frames = self.get_subsampled_frames(frames, subsample_len)
             return np.array(sub_frames)
         elif self.model_type in ["neva", "vila", "mllama"]:
+            if not HAVE_PIL:
+                raise UnavailableError(MISSING_PIL_MSG)
+
             if input_media.startswith("http") or input_media.startswith("https"):
                 response = requests.get(input_media, timeout=5)
                 media = Image.open(BytesIO(response.content)).convert("RGB")
@@ -106,6 +136,9 @@ def query(
         lora_uids=None,
     ):
         """Run query."""
+        if not HAVE_TRITON:
+            raise UnavailableError(MISSING_TRITON_MSG)
+
         prompts = str_list2numpy([input_text])
         inputs = {"input_text": prompts}