pyg-lib

• Installation

Installation

We provide pre-built Python wheels for all major OS/PyTorch/CUDA combinations from Python 3.9 till 3.12, see here. Note that currently, Windows wheels are not supported (we are working on fixing this as soon as possible).

To install the wheels, 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 or 2.5.0
• ${CUDA} should be replaced by either cpu, cu102, cu117, cu118, cu121, or cu124

The following combinations are supported:

PyTorch 2.5	cpu	cu117	cu118	cu121	cu124
Linux	✅		✅	✅	✅
Windows	✅		✅	✅	✅
macOS	✅

PyTorch 2.4	cpu	cu117	cu118	cu121	cu124
Linux	✅		✅	✅	✅
Windows	✅		✅	✅	✅
macOS	✅

PyTorch 2.3	cpu	cu117	cu118	cu121	cu124
Linux	✅		✅	✅
Windows	✅		✅	✅
macOS	✅

PyTorch 2.2	cpu	cu117	cu118	cu121	cu124
Linux	✅		✅	✅
Windows	✅		✅	✅
macOS	✅

PyTorch 2.1	cpu	cu117	cu118	cu121	cu124
Linux	✅		✅	✅
Windows	✅		✅	✅
macOS	✅

PyTorch 2.0	cpu	cu117	cu118	cu121	cu124
Linux	✅	✅	✅	✅
Windows	✅	✅	✅
macOS	✅

PyTorch 1.13	cpu	cu117	cu118	cu121	cu124
Linux	✅	✅
Windows	✅	✅
macOS	✅

Form nightly

Nightly wheels are provided for Linux from Python 3.9 till 3.12:

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

From master

pip install ninja wheel
pip install git+https://github.com/pyg-team/pyg-lib.git