Bolo: Curated, Verified, and Ready-to-run Inference Pipelines for HuggingFace Models

Bolo gives you curated, verified, and ready-to-run inference pipelines for HuggingFace models: one call, no glue code.

Install

pip install pybolo

Each model also needs its own dependencies, listed in its requirements.txt; see the website for how to install them.

Python API

import bolo

bolo.fetch_templates()                                 # optional: pre-download the templates (otherwise fetched on first use)
bolo.list_params("<REPO_ID>")                          # show the template's parameters and defaults
result = bolo.pipeline("<REPO_ID>", device="cuda:0")   # run inference, overriding any parameter

CLI

bolo fetch-templates               # optional: pre-download the templates (otherwise fetched on first use)
bolo run <REPO_ID> device=cuda:0   # run inference, parameters as KEY=VALUE

Platforms

bolo is developed and tested on Red Hat Enterprise Linux machine equipped with NVIDIA A40 GPU, with CUDA version 13.0.

Contributing

Contributions are welcome! Please open an issue or pull request on GitHub.

• Each issue must be labeled with a label.
• Every bug issue will only be closed when PyPI is updated.
• For bug issue, please refer to tracker issue for progress.
• For TODO issue, please refer to Notion sprint for progress (each TODO issue contains the specific sprint page).

License

MIT — see LICENSE for details.

Acknowledgement

• Thanks Jojo and her sister for designing the mascot.