This proposal aims to enhance the OSI's Open Source AI Definition by providing clear guidelines on the distinct elements of AI systems and their openness.
OpenLLM France is not a law firm and does not offer legal services or advice. This work (license draft and related information) is provided on an "as-is" basis, and OpenLLM France disclaims all liability for damages resulting from their use.
Note: as a starting point, take in consideration the Definition of Open Source AI itself; and a checklist to evaluate licenses. We follow the definition of AI adopted by UNESCO (https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449):
An AI system is a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. AI systems are designed to operate with varying levels of autonomy.
It's time to clarify the definition of #OpenSourceAI and to differentiate ourselves from the #OpenScience, #OpenWeights approaches that blur understanding of the true openness and transparency of today's popular LLMs.
This proposal aims to extend and refine the Open Source AI Definition draft v. 0.0.3 by the Open Source Initiative (OSI). It introduces specific enhancements focusing on the distinct aspects of AI systems: training datasets, source code, and models.
As a continuation of the work carried out by the OpenLLM France community on the CLAIRE model, and in particular on the publication of the model's actefact sets (training datasets, source code, tooling for re-training the model), we'd like to make our contribution to the definition of #OpenSourceAI.
This draft is the work of a number of contributors. The Request For Comment (RFC) is open until January 19, 2024, so that we can make an official contribution to OSI and take advantage of FOSDEM to discuss and move forward in this subject with OSI representatives in Brussels and with all the other players who wish to take part in this initiative.
This proposal try to push transparency, licensing differentiation between academic and commercial use, and the publication of training methodologies.
This proposal aims to enhance the OSI's Open Source AI Definition by providing clear guidelines on the distinct elements of AI systems and their openness. It encourages transparency, ethical practices, and collaborative development in the AI field.
You are invited to comment and contribute to this draft, so that we can collectively come up with the most widely shared definition of #OpenSourceAI, so that these technologies can be deployed industrially and safely. Thank you in advance for your contributions.
You contact join us by mail contact@openllm-france.fr or on Discord.