Installing and using torch_topological

Requirements

torch_topological requires at least Python 3.9. Normally, version resolution should work automatically. The precise mechanism for this depends on your installation method (see below).

Installation via pip

We recommended installing torch_topological using pip. This way, you will always get a release version with a known set of features. It is recommended to use a virtual environment manager such as poetry for handling the dependencies of your project.

$ pip install torch_topological

Installation from source

Installing the package from source requires a virtual environment manager capable of parsing pyproject.toml files. With poetry, for instance, the following steps should be sufficient:

$ git clone git@github.com:aidos-lab/pytorch-topological.git
$ cd pytorch-topological
$ poetry install