On 2026-07-31 10:47, Thanos Apollo wrote:
Jean Louis <[email protected]> writes:
You can modify and study datasets on any average computer.
You can modify and study inference software on any computer.
I'm sure you have downloaded the PBs of data needed and spent your time
modifying and studying it.
https://huggingface.co/datasets
Yes, I do download datasets and run verification on such using Emacs
Lisp libraries and have spent immense time understanding it. And
generating datasets on each elementary object I have be it PDF, or image
or video. I am looking what licenses are there, which model authors lie
about being truly free as in free software, whether they stand to their
own published principles that apparently are aligned with free software
principles as in GNU.
Some of large LLMs made by universities were liberated from proprietary
license to free software licenses due to the work, and I keep doing it.
I have 1060 GPU in other computer, one 17 year old is able to study it
there in other room.
Ah, how many models have you trained with that?
That's not the point, and you know it. We discussed the possibility to
study and modify software, and also to study the inference, it is all
possible with or without the GPU.
It was not in the context of training some large model on it. It would
take too long time.
But you can do tiny LLM on your own CPU. You could try Karpathy's Nano
Chat: https://github.com/karpathy/nanochat for beginning.
If you have nothing substantive to add to the discussion, you can
simply
say so. But dismissing my points with a patronizing remark like that
is
not an argument—it's just rudeness.
It is uncalled for, rude and anti-social behavior to use LLMs in a
public conversation. You are just sending me walls of text with 0
substance and expect a reply from me.
All my text is with substance. If you perceive it was "walls of text" it
means you couldn't find time or peace to actually read and understand
it. All my references I gave you I have in my computer, my typing and
reading speed is high. Excuse me that I had high expectations for this
conversation. I thought we were discussing real issues: copyleft, SaaSS,
local models, and how the GNU Project should handle LLM-generated
contributions. Those deserve more than dismissive one-liners and false
accusations.
--
Jean Louis
---
via emacs-tangents mailing list
(https://lists.gnu.org/mailman/listinfo/emacs-tangents)
