Hacker :vf:

Usuario de :gnu: #GNU con cosas.

Sin 🄯 #Copyleft sólo queda la barbarie.

Retoot ≠ aprobación
Enlazado ≠ aprobación

Last Notes

2026-05-29 19:31:08 UTC

- reply

Process text to do what exactly? Computers are good calculators. You can do calculations to check if a point in euclidean space is bounded within a known volume. Or do calculations to obtain the probability of something happening.
Statistics have a margin of error. Errors can be compounded. A concatenation of trivial snippets based on the probability of them forming a sequence is error compounding, propagation of uncertainty. You end up with a pile of crap. The only thing you can do to mitigate that is using an enormous number of samples to lower the error up to a limit. That's inefficient.
But LLMs are even worse, because they work at token level. Error propagation is greater, so you need even more samples to keep gibberish from happening too often.

2026-05-29 12:58:17 UTC

- reply

Yeah, copyright infringement is inevitable.
I'm of the opinion that if the generated AI code looks no different than from human code then it didn't save you anything. And if the code was trivial enough for an LLM to not need rework it probably could be covered with a template snippet from your crafted collection of templates.
And there's research showing no productivity improvement at all, even negative.

2026-04-10 19:37:46 UTC

- reply

#Vulkan would be welcomed too.

2025-07-22 11:59:40 UTC

- reply

de toda esa parrafada lo único útil es que los cursos para aprender catalán no tienen suficientes plazas.
Por otra parte difícilmente se me ocurriría trabajar por la UE sin conocer el idioma del país o saber si permiten el uso del inglés (ya que no nos enseñan #Esperanto).