I use Ubuntu works great.
get something that can run anaconda and then u dont need to worry about anything
rejectedlesbian
joined 1 year ago
the main use for a leading orgenization in agi is that they will hopefully do it safely. but thats not really what we have been seeing recently from openai or meta.
what we really really want to avoid is the situation where an AI system is able to be profitable enough by itself to pay for its own compute and starts copying itself like crazy.
that sort of system wont have a centralized plug we can pull and copies would mutate and evolve and that can potentially go horribly horribly wrong.
but other than that nightmare scenario having a split in the industry is actually good and the best work came from a time like that. good research does not come from big organizations it comes from small (relatively) independent teams.
the current "leaders" of ai pushing for their specific narrative have made us miss a few things for instance the idea that transformers are the be all end all has made us overestimate VIT for years. (https://arxiv.org/abs/2207.11347)
computers are deterministic so any computer program is deterministic...
quantum computers do change that but AI as we know it today is a program that runs on classical computers.
I would just LOVE to see the technical details.
optimizer OMG no one touched optimizes for decades.
we basically figure its ADAM/SGD and there wasnt really any improvement on it.
I tried finding an improvement to it myself for a few months but failed miserably
if u don't have a test dataset than absolutely no.
sometimes people do split test/validation/train and then the validation dataset can be used for meta learning.
I make 0$ but I am just starting out.
hoping to get a payed internship/entry level job in the next few months
kinda hard without a degree but I am working on it
I literly have a publication and experience of 2 years and can't get an entry level job for months now.
So I think very.