Set up your environment

Prepare your host system

Before you can create a reference PyTorch model, you need to set up Python, ExecuTorch, and Arm backend dependencies.

Use one of the following:

• Linux
• macOS with Apple Silicon

Verify that your Python version is 3.10 or later and earlier than 3.14:

    

        
        
python3 --version

    

Create a Python virtual environment

Create a working directory for this Learning Path:

    

        
        
mkdir preparing-models-for-nt
cd preparing-models-for-nt

    

Create a virtual environment:

    

        
        
python3 -m venv venv
source venv/bin/activate
python -m pip install --upgrade pip

    

Clone and install ExecuTorch

Clone ExecuTorch from GitHub and install:

    

        
        
git clone https://github.com/pytorch/executorch.git repo/executorch
cd repo/executorch
./install_executorch.sh

    

Install Arm backend dependencies

From the root of repo/executorch, run:

    

        
        
./examples/arm/setup.sh \
  --i-agree-to-the-contained-eula \
  --disable-ethos-u-deps \
  --enable-mlsdk-deps

    

Source the generated path script in the same shell session:

    

        
        
source ./examples/arm/arm-scratch/setup_path.sh

    

Return to the Learning Path working directory:

    

        
        
cd ../..

    

(Optional) Use a Jupyter notebook

If you prefer to work through the same steps in Jupyter, download the notebook from the arm-learning-paths repository:

    

        
        
wget https://raw.githubusercontent.com/ArmDeveloperEcosystem/arm-learning-paths/main/content/learning-paths/mobile-graphics-and-gaming/preparing-models-for-nt/prepare-models-for-nt.ipynb

    

Install Jupyter Lab in your active virtual environment:

    

        
        
pip install jupyterlab

    

From the preparing-models-for-nt directory, launch Jupyter Lab:

    

        
        
jupyter lab

    

Open the notebook:

    

        
        prepare-models-for-nt.ipynb

        
    

What you’ve accomplished and what’s next

You’ve now set up the Python environment, installed ExecuTorch, and configured Arm backend dependencies for creating and exporting PyTorch models.

Next, you’ll create and export your first test model, either from the Learning Path pages or from the optional Jupyter notebook.