@abucci@anthony.buc.ci where they now say they use it to train their AI models thry used to say “for language models”, which isn’t all that different (possibly extending the scope from text to images, audio and video?).