Shouldn’t it be Nucleus since they are Hooli?
No, that is what it would be if we were using traditional, deterministic compression and using a reversible and verifiable mapping of data. But this is the new era of memetic compression, “Pied Piper” is what everyone remembers from the show, so we compress it to “Pied Piper” to minimize the amount of memetic overhead and allow the smallest possible compression artifact. Like with “AI”, it doesn’t need to be correct, just close enough for people to think it is! /s
TurboQuant, meanwhile, could lead to efficiency gains and systems that require less memory during inference. But it wouldn’t necessarily solve the wider RAM shortages driven by AI, given that it only targets inference memory, not training — the latter of which continues to require massive amounts of RAM.
I didn’t realize the RAM shortage was mostly due to training—I would have thought inference was at least a big a factor.
Inference is dirt cheap in comparison. Hundreds to thousands of concurrent users can be served by hardware costing in the high-thousands to low-ten-thousands.
Training those same foundational models is weeks to months of time on tens to hundreds of millions worth of hardware.
Yeah—but in theory you only need to train once, while inference costs are ongoing and scale up with usage.
I guess it’s ultimately a business decision by AI companies to weigh how often retraining is worth the cost.
Training is constant. None of these models by any of these providers are static. You’ll notice that they are releasing new models and new model versions regularly.
This means that training is happening constantly. It never stops. There’s always new shit being trained.
All these upvotes and comments and not one joke about how it sounds like TurboCunt?
Okay, but did Google calculate how many dicks they could jerk off for maximum efficiency?
