* Eli Zaretskii <e...@gnu.org> <868qoufvf0....@gnu.org> Wrote on Mon, 24 Mar 2025 14:46:43 +0200 >> Date: Mon, 24 Mar 2025 14:27:56 +0530 (IST) >> From: Madhu <enom...@meer.net> >> Are you talking of LLMs or of other ML models? > For the purpose of this discussion, what is the difference?
I thought, if the difference was real and the strong form of the differnece takes the form I imagined, that it would explain where The OP(Jean Louis) was coming from. (I haven't read full thread yet.) > If by "LLM" you mean only those models that produce text (either for > human consumption or text of programs), then no, I'm not talking about > LLMs of that kind. Again I may be mistaken but I assumed LLMs have to do with manipulation of language elements and also necessarily involved generation of language elements -- there is no real problem to be solved except consumption of humans and other LLMs. If you have an example of a LLM domain is not generative in the above sense, maybe I can understand. (I haven't touched Machine Learning mathematics since 2006 and it is likely I'm conceptually wrong) > "Clearly definable" is in the eyes of the beholder. Some problems > have only probabilistic solutions, in which case the output is some > kind of weight or score. And these will be assigned based on an overreaching agenda and the power and clout to pull it off (maybe like discussions on emacs-devel?). This it is the opportunity to program socio-economic political meta-models into the real world systems where this technology is to be implemented. [snip] > I'm not sure this aspect is at all relevant to the present discussion. (it was relevant to my sense of what it meant for people "to accept") --- via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)
