Introduction
In this article, we will go over the steps to configure a PyPI remote repository for packages hosted on data.pyg.org.
Create a new PyPI remote repository and set its URL to https://data.pyg.org.
The Registry URL and Registry Index Location URL Suffix fields can be ignored for this setup, since the installation relies on a direct URL rather than a standard PyPI index.
After creating the repository, we will continue to configure the pip client to work with this repository.
Instead of configuring the client via the Set Me Up instructions, we’ll need to use the -f (--find-links) flag.
This is because https://data.pyg.org is not a standard PyPI “simple index” (it doesn’t follow the /simple/<package>/ structure), so pip can’t treat it as a regular index (--index-url). Instead, we need to provide a more specific location, and -f tells pip where to look for packages during installation.
You can install the package using the following command:
pip install pyg-lib -f https://username:TOKEN@ARTIFACTORY_HOST/artifactory/<REPOSITORY_NAME>/whl/torch-2.9.0+cpu.html
After running the command above, you should be able to see the files cached in the remote-cache repository, for example: