Project ProtoStar links public research sharing to a climate AI challenge
Hugging Face and research partners announced Project ProtoStar on October 29 to improve reuse of publicly funded scientific datasets, models and code. A scheduled climate challenge will test those resources on a concrete research task.
Spec sheet
- Initiative
- Project ProtoStar
- Challenge host
- Carnegie Mellon University
- Challenge dates
- 2–13 November 2026
- Weather coverage
- 210 US cities
- Participation
- Non-binding and non-exclusive
The initiative targets practical obstacles to reuse, including missing documentation, unclear provenance, incomplete metadata and incompatible formats. Participating institutions commit to releasing suitable research artifacts through open infrastructure. The project describes those commitments as non-binding and non-exclusive, leaving institutions room to use domain-specific repositories as well as shared infrastructure.
Carnegie Mellon’s Climate Risk Intelligence challenge is scheduled for November 2–13 and uses historical weather records from 210 US cities. Organizers plan to compare predictions with statistical baselines and assess climate-adaptation ideas separately. The competition has yet to produce results.
Claim check
Sources
- Announcing Project ProtoStar, October 29, 2026 · Hugging Face and Project ProtoStar partners · 29 Oct
- Project ProtoStar organization and participation page · Project ProtoStar · 31 Oct
- Climate Risk Intelligence challenge page · Carnegie Mellon University Open Source Programs Office · 31 Oct