Dont pre-install example dependencies in docker, install on run-time

Signed-off-by: Keval Morabia <[email protected]>

Author: kevalmorabia97

.gitlab/tests.yml

@@ -56,6 +56,7 @@ example:
   script:
     # Uninstall apex since T5 Int8 (PixArt) + Apex is not supported as per https://github.com/huggingface/transformers/issues/21391
     - if [ "$TEST" = "diffusers" ]; then pip uninstall -y apex; fi
+    - if [ -f examples/$TEST/requirements.txt ]; then pip install -r examples/$TEST/requirements.txt; fi
     - if [ "$TEST_TYPE" = "pytest" ]; then pytest -s tests/examples/$TEST; else bash tests/examples/test_$TEST.sh; fi
 
 example-ada:

README.md

@@ -68,7 +68,7 @@ To install from source in editable mode with all development dependencies or to
 
 ```bash
 # Clone the Model Optimizer repository
-git clone https://github.com/NVIDIA/TensorRT-Model-Optimizer.git
+git clone git@github.com:NVIDIA/TensorRT-Model-Optimizer.git
 cd TensorRT-Model-Optimizer
 
 pip install -e .[dev]

docker/Dockerfile

@@ -18,17 +18,13 @@ RUN ln -s /app/tensorrt_llm /workspace/tensorrt_llm
 ENV LD_LIBRARY_PATH="/usr/lib/x86_64-linux-gnu:/usr/local/tensorrt/targets/x86_64-linux-gnu/lib:${LD_LIBRARY_PATH}" \
     PATH="/usr/local/tensorrt/targets/x86_64-linux-gnu/bin:${PATH}"
 
-# Install modelopt with all optional dependencies and pre-compile CUDA extensions otherwise they take several minutes on every docker run
-RUN pip install -U "nvidia-modelopt[all,dev-test]"
-RUN python -c "import modelopt.torch.quantization.extensions as ext; ext.precompile()"
-
-# Find and install requirements.txt files for all examples excluding windows
+# Install modelopt from source with all optional dependencies and pre-compile CUDA extensions otherwise they take several minutes on every docker run
+# Pre-install llm_ptq requirements.txt
 COPY . TensorRT-Model-Optimizer
+RUN pip install -e "./TensorRT-Model-Optimizer[all,dev-test]"
 RUN rm -rf TensorRT-Model-Optimizer/.git
-RUN find TensorRT-Model-Optimizer/examples -name "requirements.txt" | grep -v "windows" | while read req_file; do \
-        echo "Installing from $req_file"; \
-        pip install -r "$req_file" || exit 1; \
-    done
+RUN python -c "import modelopt.torch.quantization.extensions as ext; ext.precompile()"
+RUN pip install -r TensorRT-Model-Optimizer/examples/llm_ptq/requirements.txt
 
 # Allow users to run without root
 RUN chmod -R 777 /workspace

docs/source/getting_started/_installation_for_Linux.rst

@@ -34,16 +34,16 @@ Environment setup
 
     To use Model Optimizer with full dependencies (e.g. TensorRT/TensorRT-LLM deployment), we recommend using our provided docker image
     which is based on the `TensorRT-LLM <https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tensorrt-llm/containers/release/tags>`_
-    docker image with additional example-specific dependencies installed.
+    docker image with additional dependencies installed.
 
     After installing the `NVIDIA Container Toolkit <https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html>`_,
-    please run the following commands to build the Model Optimizer docker container which has all the necessary
-    dependencies pre-installed for running the examples.
+    please run the following commands to build the Model Optimizer docker container which has all the base
+    dependencies pre-installed. You may need to install additional dependencies from the examples's `requirements.txt` file.
 
     .. code-block:: shell
 
         # Clone the ModelOpt repository
-        git clone https://github.com/NVIDIA/TensorRT-Model-Optimizer.git
+        git clone git@github.com:NVIDIA/TensorRT-Model-Optimizer.git
         cd TensorRT-Model-Optimizer
 
         # Build the docker (will be tagged `docker.io/library/modelopt_examples:latest`)
@@ -60,8 +60,7 @@ Environment setup
 
     For PyTorch, you can also use `NVIDIA NGC PyTorch container <https://catalog.ngc.nvidia.com/orgs/nvidia/containers/pytorch/tags>`_
     and for NVIDIA NeMo framework, you can use the `NeMo container <https://catalog.ngc.nvidia.com/orgs/nvidia/containers/nemo/tags>`_.
-    Both of these containers come with Model Optimizer pre-installed. NeMo container also comes with the HuggingFace and TensorRT-LLM
-    dependencies. Make sure to update the Model Optimizer to the latest version if not already.
+    Both of these containers come with Model Optimizer pre-installed. Make sure to update the Model Optimizer to the latest version if not already.
 
     For ONNX PTQ, you can use the optimized docker image from [onnx_ptq Dockerfile](https://github.com/NVIDIA/TensorRT-Model-Optimizer/tree/main/examples/onnx_ptq/docker).
 

examples/onnx_ptq/docker/Dockerfile

@@ -19,16 +19,16 @@ ENV LD_LIBRARY_PATH="${CUDNN_LIB_DIR}:${TRT_PATH}/lib:/usr/include:${LD_LIBRARY_
 ENV PATH="${TRT_PATH}/bin:${PATH}"
 
 # Copy application code and install requirements
-COPY modelopt modelopt/modelopt
-COPY examples/onnx_ptq modelopt/examples/onnx_ptq
-COPY setup.py modelopt/setup.py
-COPY pyproject.toml modelopt/pyproject.toml
+COPY modelopt TensorRT-Model-Optimizer/modelopt
+COPY examples/onnx_ptq TensorRT-Model-Optimizer/examples/onnx_ptq
+COPY setup.py TensorRT-Model-Optimizer/setup.py
+COPY pyproject.toml TensorRT-Model-Optimizer/pyproject.toml
 
 # Install onnx_ptq requirements
-RUN pip install -r modelopt/examples/onnx_ptq/requirements.txt
+RUN pip install -r TensorRT-Model-Optimizer/examples/onnx_ptq/requirements.txt
 
 # Install modelopt
-RUN pip install -e "./modelopt[hf,onnx]"
+RUN pip install -e "./TensorRT-Model-Optimizer[hf,onnx]"
 
 # Allow users to run without root
 RUN chmod -R 777 /workspace