GPhub offers an interactive guide to Gaussian Process (GP) libraries. This project is built using Hugo Framework, GO Template, HTML, CSS, and JS. The guide features a comprehensive dynamic table to provide an easy-to-use and intuitive reference for exploring the GP libraries available to the data science community. Whether new to GP or experienced in the field, the table simplifies the process of selecting the ideal GP tool that best suits one's needs.
The table is available by opening the Table.html file on a web browser of your choice.
Contributions to GPhub are warmly welcomed. There are several ways you can contribute:
- Improving the Guide: Help us enhance the content and accuracy of our guide. This includes updating library information, adding new GP tools, and refining our descriptions to make them more helpful.
- Technical Enhancements: Contribute to the development of the site by improving its design, enhancing the functionality of the dynamic table, or fixing bugs.
- Feedback and Suggestions: Share your insights on how we can improve GPhub. We're always looking for feedback on the usability of the guide, new features that could be added, or any other suggestions you might have.
- Fork the Repository: Start by forking the GPhub repository on GitHub. This allows you to make changes without affecting the main project.
- Create a Pull Request: Once you've made your changes, submit a pull request. Be sure to provide a detailed description of your contributions and why they are beneficial to the project.
- Code Review: After submission, your pull request will be reviewed by the project maintainers. This process ensures that contributions align with the project's standards and goals.
- Integration: Once approved, your contributions will be merged into the project, becoming a part of the GPhub guide.
To add a new library to the table, you need to create a new PackageName.yaml entry in the gphub/data/libraries directory by running:
cd gphub
make -f new_library name=PackageNameThis will generate a new file from gphub/utils/libraries/base.yaml (the schema of record for every library file, shown below) with PackageID and Library.Name already filled in:
PackageID: null
Library:
- Name: null
URL: null
Reference:
- Author: null
URL: null
Language:
- null
License:
- Name: null
URL: null
Repository:
- Name: null
URL: null
Version:
- Name: null
URL: null
Installation:
- null
Developer:
- Tag: null
Name: null
URL: null
Docs:
- Name: null
URL: null
Support:
- Name: null
URL: null
Framework:
- null
GPU: false
Models:
- Types:
- Tag: null
Name: null
URL: null
Complexity:
- null
Trend:
- Types:
- Tag: null
Name: null
URL: null
LengthScale:
- null
Correlation:
- Types:
- Tag: null
Name: null
URL: null
Mixture: false
MixtureModels:
- Types:
- Tag: null
Name: null
URL: null
Likelihood:
- Types:
- Tag: null
Name: null
URL: null
Nugget:
- Tag: null
Name: null
Noise:
- Tag: null
Name: null
EstimationMethods:
- Types:
- Tag: null
Name: null
URL: null
OptimizationMethods:
- Types:
- Tag: null
Name: null
URL: null
Resampling: false
Metrics:
- nullModify the ./gphub/data/libraries/PackageName.yaml file.
Refer to Hugo website for details on how to install Hugo.
brew install hugoThe helper scripts only need PyYAML:
cd gphub
python3 -m venv .venv && .venv/bin/pip install -r utils/requirements.txtPrettier formats the YAML files after every script run; gh, yq and jq are used by the version-update script.
brew install prettier gh yq jq
gh auth loginYou can start your local server via hugo by executing:
cd gphub
hugo serve -DThe table is served at http://localhost:1313
A GitHub Actions workflow (.github/workflows/update-versions.yml) refreshes the Version of every library from its latest GitHub release (or newest tag) every Monday and opens a pull request with the changes and a rebuilt Table.html. Trigger it by hand with:
gh workflow run update-versions.ymlThe same script can be run locally (it uses your gh login):
cd gphub
make -f fetch_versionsLibraries that are not hosted on GitHub (DACE, GPML, MUQ, ooDACE, UQLab, UQpyLab) or have no release and no tag (albatross, AutoGP, GPR) are skipped and must be updated by hand; friedrich and SuperGauss are tracked through crates.io and CRAN respectively.
The validator checks every file against gphub/utils/libraries/base.yaml and the canonical values recognised by the templates (languages, licences, installation methods, metrics, colour tags, and so on). Run it after any edit:
cd gphub
make -f validateYou can regenerate the self-contained Table.html (no dev server needed) by running:
make -f table_htmlThis project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 955393.
Licensed under the MIT license.