Installation

Note

Virne’s RL environments use the gymnasium==1.3.0 API. The scientific stack requires NumPy 2 and Python 3.12 or newer.

Create a Virtual Environment

python3 -m venv .venv
source .venv/bin/activate

Install from PyPI

Install the latest Virne release and verify its command-line entry point:

python -m pip install virne
virne --version

The standard installation includes the complete Virne runtime. PyG’s optional compiled extensions are not required for correctness.

Select a CPU or CUDA Build

When you need to select an exact CPU or CUDA build of PyTorch, clone the source repository and use the installation script. It supports CPU environments on Linux and macOS, plus CUDA 12.6, 13.0, and 13.2 on Linux. It installs PyTorch 2.13.0, PyG 2.8.0.post1, Gymnasium 1.3.0, Virne in editable mode, and the matching optional pyg_lib acceleration wheel.

git clone https://github.com/GeminiLight/virne.git
cd virne

# CPU-only PyTorch and PyG
bash install.sh -c cpu

# CUDA 12.6; 13.0 and 13.2 are also supported
bash install.sh -c 12.6

If -c is omitted, the script installs the CPU build. This explicit default avoids selecting a CUDA runtime that is incompatible with the host driver.

Verify the Installation

Keep the virtual environment active and run:

virne --version
python -c "import gymnasium, torch, torch_geometric, virne; print(virne.__version__, gymnasium.__version__, torch.__version__, torch_geometric.__version__)"

The command should print the installed Virne, Gymnasium, PyTorch, and PyG versions. You can then continue to the Quickstart.