Installation

Bento requires Python version 3.9 or higher. We recommend using a virtual environment to install Bento to avoid conflicting dependencies with other packages. If you are unfamiliar with virtual environments, we recommend using Miniconda.

PyPI

Bento is available on PyPI:

$ pip install bento-tools

Troubleshooting

If you encounter installation errors, you may need double check that package dependencies are installed correctly.

Development

We currently use Rye for package management.

  1. Install Rye.

  2. Clone the repository and navigate to the root directory.

    gh repo clone ckmah/bento-tools
    cd bento-tools
    
  3. Install the package in editable mode:

       pip install -e .
    

Run tests:

   python -m unittest tests

### Documentation

```console
   pip install -e ".[docs]" # install optional deps for docs
   cd docs
   make livehtml

GPU Support (Optional)

   pip install -e ".[torch]"

Bento currently only uses GPUs to accelerate tensor decomposition (via Tensorly). PyTorch can be installed with the torch extra, but users should consult PyTorch installation instructions for enabling GPU support, as installation varies by platform.