Installation

Clone the GitHub repo: https://github.com/BayesLIM/BayesLIM and then

cd BayesLIM
pip install .

If installed properly, you should be able to import it in a Python session as:

import bayeslim as ba

Dependencies

See the pyproject.toml file for dependencies, listed under [project.optional-dependencies]. I placed them here so that they won’t automatically install during a pip install . by default, which doesn’t always play nicely with software sharing similar dependencies. If you’d like to have pip install the dependencies anyways, you can make this happen by installing the [dev] version:

pip install .[dev]

PyTorch: to my knowledge there is not a great way to install different CPU/GPU versions of PyTorch from within a pyproject.toml, but I suspect this will change relatively soon. In the meantime, you should install torch>=2.0.0 (and optionally CUDA if you want GPU support) on your own (https://pytorch.org/get-started/locally/) before installing BayesLIM.