Thanks for sharing, @clefourrier ! Levels of access is definitely a great framework to assess openness! As more and more models are released, a community contributed (+AI-assisted) initiative to assess and compare the models could be useful. The paper and a few others in these lines provide really good base for discussions, and as the space evolves, we can add more nuances to the different levels of access.
That's very useful, @osanseviero !
I had some thoughts on a community contributed "leaderboard" of sort for where different large models fall in the spectrum of openness across dimensions https://kshitizkhanal.com/?page_id=57 Let me know what you think
