• Installation

We provide pre-built Python wheels for all major OS/PyTorch/CUDA/ROCm(HIP) combinations from Python 3.10 till 3.13, see here. Note that currently, Windows wheels are not supported (we are working on fixing this as soon as possible).

To install the wheels for CPU/CUDA backend, simply run

pip install pyg-lib -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html

where

• ${TORCH} should be replaced by either 1.13.0, 2.0.0, 2.1.0, 2.2.0, 2.3.0, 2.4.0, 2.5.0, 2.6.0, 2.7.0, or 2.8.0
• ${CUDA} should be replaced by either cpu, cu102, cu117, cu118, cu121, cu124, cu126, cu128, or cu129

The following combinations are supported:

For ROCM backend, there is an external pyg-rocm-build repository provides wheels and detailed instructions on how to install PyG for ROCm. If you have any questions about it, please open an issue here.

Note: ROCM backend only support Linux up to now.

Nightly wheels are provided for Linux from Python 3.10 till 3.13:

pip install pyg-lib -f https://data.pyg.org/whl/nightly/torch-${TORCH}+${CUDA}.html

pip install ninja wheel
pip install --no-build-isolation git+https://github.com/pyg-team/pyg-lib.git