Setting PyTorch Environment for CPU

Setting PyTorch Environment for CPU#

This page describes the process to setup a PyTorch environment in your machine.

Follow the steps to install Python and virtual environment tools for your OS: Windows or Linux, then follow the steps to create a venv where you will run the getting started examples.

Alternative, you can setup VS Code in any of the supported OSes.

Pre-requisites#

AMD EPYC™ Processors, AMD Ryzen™ Processors or AMD Ryzen™ AI Processors

Your platform is: Loading...

Windows#

Click here to download Python 3.12.8 or find the latest release of Python here python.org/downloads/windows/ select Windows installer (64-bit)

Note

Python 3.13 has not been verified.

Follow the steps in the Python documentation after executing the installer.

Once you complete the installation, open a Windows PowerShell.

Confirm that Python is installed by checking the version

python -c "import sys; print(sys.version)"

Continue by creating the venv.

Linux#

We will use a Python virtual environment (venv) to install the dependencies.

Open a new terminal and install python3-venv.

sudo apt install python3-pip python3-venv libgl1

Confirm that Python is installed by checking the version

python3 -c "import sys; print(sys.version)"

Continue by creating the venv.

Create venv#

Navigate to a directory where you would like to create the venv, then create the venv.

cd /path/to/directory/
mkdir aupai
python3 -m venv aupai

cd /path/to/directory/
mkdir aupai
python -m venv aupai

Activate the venv.

source aupai/bin/activate

./aupai/Scripts/Activate.ps1

./aupai/Scripts/activate.bat

Confirm that you are inside the venv.

python -c "import sys; print(f'Running on venv: {sys.prefix != sys.base_prefix}')"

Note

You can deactivate the venv with deactivate.

Continue by installing package dependencies.

Visual Studio Code#

If you prefer working from a code editing tool, you can setup VS Studio code.

  1. Install VS Code.

  2. Get started with Python in VS Code

  3. Create a virtual environment

Install Package Dependencies#

Inside the venv execute the following to install the necessary dependencies.

python -m pip install "torch>2.3" "numpy<2" torchvision fvcore
python -m pip install av validators matplotlib jupyterlab jupyter transformers SentencePiece accelerate ultralytics==8.0.196 wheel
python -m pip install tiktoken einops pytest Pillow Requests jupyterlab_myst torchinfo "onnx>=1.16.2" netron tqdm shap kaggle roboflow==1.1.47 pickleshare
python -m pip install jupyter-book sphinxcontrib-mermaid sphinx_design

Note

If you are using VS Code, you can open a new terminal by Ctrl+Shift+` or Terminal > New Terminal. For more information, checkout Install and use packages.

This process can take a few minutes to download and install.

You are now all set to get started with pre-trained models in AMD CPUs.

Copyright (C) 2025 Advanced Micro Devices, Inc. All rights reserved.

SPDX-License-Identifier: MIT

Contents