Skip to navigation
Learn

Install

View as Markdown

Prerequisites

Before proceeding with the installation of NVIDIA AITune, ensure your system meets the following criteria:

  • Operating System: Linux (Ubuntu 22.04+ recommended)
  • Python: Version 3.10 or newer
  • PyTorch: Version 2.7 or newer
  • TensorRT: Version 10.5.0 or higher (for TensorRT backend)
  • NVIDIA GPU: Required for GPU-accelerated tuning

You can use NGC Containers for PyTorch which contain all necessary dependencies:

NVIDIA AITune can be installed from pypi.org.

pip install --extra-index-url https://pypi.nvidia.com aitune

PyTorch 2.10 + CUDA 13 (torch210 extra)

To use PyTorch 2.10 with CUDA 13 support, install the torch210 optional dependency group:

pip install --extra-index-url https://pypi.nvidia.com --extra-index-url https://download.pytorch.org/whl/cu130 "aitune[torch210]"

When using uv, add --torch-backend=cu130 so that uv resolves the CUDA 13 variants of the PyTorch packages:

uv pip install --torch-backend=cu130 --extra-index-url https://pypi.nvidia.com "aitune[torch210]"

Installing from Source

# Clone the repository
git clone https://github.com/ai-dynamo/aitune
cd aitune
pip install --extra-index-url https://pypi.nvidia.com .

or use editable mode for development:

pip install --extra-index-url https://pypi.nvidia.com -e .

To install the torch210 extra from source with uv:

uv pip install --torch-backend=cu130 --extra-index-url https://pypi.nvidia.com -e ".[torch210]"