this post was submitted on 06 Jan 2025
765 points (99.0% liked)
TechTakes
1544 readers
149 users here now
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
it won't. its backed by microsoft. they can literally afford to burn the cash on this while it becomes profitable, and it will AI has so many low hanging fruits to optimize its insane.
So many low-hanging fruits. Unbelievable fruits. You wouldn’t believe how low they’re hanging.
They are tremendous fruits. Many people are saying this. Fruits.
and we have concepts for even lower hanging fruits. beautiful concepts.
Okay, explain. What kinds of low hanging fruit?
quants are pretty basic. switching from floats to ints (faster instruction sets) are the well known issues. both those are related to information theory, but there are other things I legally can't mention. shrug. suffice to say the model sizes are going to be decreasing dramatically.
edit: the first two points require reworking the base infrastructure to support which is why they havent hit widespread adoption. but the research showing that 3 bits is as good as 64 is intuitive once you tie the original inspiration for some of the AI designs. that reduction alone means you can get 21x reduction in model size is pretty solid.
hahahaha fuck off with this. no, the horseshit you’re fetishizing doesn’t fix LLMs. here’s what quantization gets you:
anyway speaking of basic information theory:
lol
Honestly, the research showing that a schlong that's 3mm wide is just as satisfying as one that's 64 is intuitive once you tie the original inspiration for some of the sex positions.
I have seen these 3 bit ai papers on hacker news a few times. And the takeaway apparently is: the current models are being pretty shitty at what we want them to do, and we can reach a similar (but slightly worse) level of shittyness with 3 bits.
But that doesn't say anything about how both technologies could progress in the future. I guess you can compensate for having only three bits to pass between nodes by just having more nodes. But that doesn't really seem helpful, neither for storage nor compute.
Anyways yeah it always strikes me as a kind of trend that maybe has an application in a very specific niche but is likely bullshit if applied to the general case
If anything that sounds like an indictment? Like, the current models are so incredibly fucking bad that we could achieve the same with three bits and a ham sandwich
Oh it definitely says something about the current models for sure
Far as I can tell, the only real benefit here is significant energy savings, which would take LLMs from "useless waste of a shitload of power" to "useless waste of power".
It's actually super easy to increase the accuracy of LLMs.
I left out all the other details because it's pretty intuitive why it works if you understand why floats have precision issues.
decimal is a severely underappreciated library
the fruit can’t be rotten, you must be picking it wrong
“look, Mme Karen, this is definitely not a rotten tomato. it can’t be a rotten tomato, we don’t sell rotten tomatoes. you can see here on the menu that we don’t have rotten tomatoes on offer. and see here, on your receipt, where it says quinoa salad? absolutely not rotten tomatoes!” explains the manager fervently, avoiding a tableward glance at the pungent red blob with as much will as they can muster
They will only pay as long as it benefits them.