I tend not to bash on reposting, but come on...
https://lemmy.world/post/12084712
https://lemmy.world/post/12084714
https://lemmy.world/post/12084927
And those are all on this specific sublemmy alone.
I tend not to bash on reposting, but come on...
https://lemmy.world/post/12084712
https://lemmy.world/post/12084714
https://lemmy.world/post/12084927
And those are all on this specific sublemmy alone.
@isnep
gpt3/4 are already trained on reddit data. Not reddit data exclusively, but there's a lot of it in there.
president is statistically the job in the US with the highest rate of mortality.
it rusts less, far less in fact, but it still does rust.
Also there are different grades, i bet they cheaped out
What do you think large language model means? If you want desicion making, you should train a model on data relevant to said desicion making. ^
This is like being confused as to why a hammer does a shit job of driving screws.
oh no, literally undecipherable
it may have started that way, but now it's all SEO
SEO os the bane of the internet. SEO is why i have to scroll through a novella every time i want to check out a recipe
Body camera video equivalent to 25 million copies of “Barbie”
Literally anything but the metric system
I rather my data I've chosen to make public is free and accessible to all, than it being sold to the highest bidder.