Possible Worlds Research is launched!

[minimalparts]

Hi everybody, I am pleased to announce the launch of the Possible Worlds Research community, here on GitHub.

This project is for people who believe in developing AI responsibly and are curious about exploring many types of algorithms and data formats. It is for people who think about the impact of their code on the world.

The repositories in this organisation contain useful code to build and explore Natural Language Processing systems. As the organisation grows, we hope to host many different tools, from data collection helpers to entire system architectures. We pledge to host code that runs on minimal hardware, so that everybody can participate.

Possible Worlds Research is at the moment a loose collective of like-minded people. We hope it can grow in the future, and why not, become a fully-fledged non-profit dedicated to small and fair AI. We welcome contributions towards this goal.

So if you want a break from bigger data, bigger models, and exaggerated predictions about the capabilities of neural architectures, join us here, discuss your ideas and concerns, and share your code. We are the home of small AI.

Thank you!
Aurelie

[naught101]

Super cool projects you got here. Pears looks like it has a great vision. Some of the other repos look interesting too.

A couple of thoughts that probably aren't big enough for their own discussions:

• Have you considered hosting somewhere that is not corporate controlled, instead of github? E.g. Codeberg (or one of the others I can't remember right now)?
• Is there a roadmap for Pears? The website looks promising, but it's not obvious how to use it. I guess the software isn't ready yet, so perhaps something like an aspirational release date would be nice.

[minimalparts]

Hey @naught101, thanks for dropping in!

About hosting: yes, it's on our very long to-do list. I must confess we haven't found the time and energy yet, but we keep thinking about it so it will eventually happen.

PeARS is actually getting there. You can already try out some instances at https://ai-pears.org/, https://queer-pears.org/ or https://www.searchfoss.nl/. But we really need to get going with the documentation so that people can give a go at setting up their own instance. At the moment, the docs are in this state: https://pears.readthedocs.io/en/latest/index.html. They need quite a lot more love and care. (Btw, the main organisation for PeARS is at https://github.com/PeARSearch if you wish to follow the progress of the Federated version.)