AI••NEXORA Editorial•TechCrunch
Base Labs launches open-weight AI safety partnership with Hugging Face and Goodfire
Summary
Base Labs, the research group Baseten formed earlier this year, has announced an open partnership with Hugging Face and Goodfire. Under the collaboration, Base Labs will develop and publish methods for training and monitoring open-weight models. The initiative aims to share practices to improve safety when using such models. The announcement did not include details on timelines, funding, or specific projects.
Details
Base Labs, a research group set up by Baseten earlier this year, has announced a partnership with Hugging Face and Goodfire focused on AI safety.
According to the provided information, Base Labs will develop and publish methods for training and monitoring open-weight models. These resources are intended to be made publicly available to share safety practices for working with open models.
Hugging Face and Goodfire are listed as partners in the initiative, but the available information does not specify their exact roles, the projects involved, or implementation timelines.
The announcement does not include details on funding, specific research outputs, or examples of models that will be targeted by the methods under development.
The communication frames the effort as an open publication of training and monitoring methodologies, without additional context on next steps or other participants.
Any future disclosures would be needed to clarify the scope and practical application of the proposed safety methods.
NEXORA AI — Why it matters
Publishing open methods for training and monitoring could make safety practices for open models more accessible.
NEXORA PRO
Unlock detailed analysis, key facts, context, story developments, source comparisons, a personalized feed and smart alerts.
Explore NEXORA PRO