Welcome to November, and welcome to the Stubsack, your first port of call for learning fresh Awful you’ll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cut’n’paste it into its own post — there’s no quota for posting and the bar really isn’t that high.
The post Xitter web has spawned soo many “esoteric” right wing freaks, but there’s no appropriate sneer-space for them. I’m talking redscare-ish, reality challenged “culture critics” who write about everything but understand nothing. I’m talking about reply-guys who make the same 6 tweets about the same 3 subjects. They’re inescapable at this point, yet I don’t see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldn’t be surgeons because they didn’t believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I can’t escape them, I would love to sneer at them.
(Credit and/or blame to David Gerard for starting this.)
More people using it, more training data fed into it.
That seems a bit low.
yes but that’s not actually the hard and costly part. the hard and costly part is the actual training action, which runs in the hundred of millions if not more
which, y’know, when lower prices……
Well, they’ve also reached model sizes where scrapes of the entire internet aren’t sufficiently large enough to train the model, so they also need more data. But yeah, they compute cost of actually training the models are prohibitively large.
yeah they’re far past public-only sources, there’s more than a few non-public sources going into this shit too. and still it’s not enough to work and sort out the issues! almost as tho the whole thing is a bit of a silly idea
I’m glad there’s finally some financial reckoning coming, I just hope I’m wrong about how utterly fucking bad it will be
If that paper about astra being able to make sense of other models’ weights has any merit, at some point they became so desperate they actually started adding serialized open weight models to the training data.