If he had one shortly after they launched, the suspension was truly terrible. R1T was clearly where all the work went and it's like no one even drove an R1S before they released it. Thankfully due to SW updates the suspension is absolutely night and day better, it went from making me violently sick to pretty nice. Assuming dudes car wasn't just a lemon, if he had and got rid of the R1S before the suspension got better I can 100% believe he hates the car.
I can understand when people say the code was the way part, but under the assumption that the system design and architecture are clean, code is high quality or the project is greenfield, and there is proper testing and validation. Then, sure, the lines of code aren't the hardest but that's only because that hardest work was front loaded and given a different name. Even then it's still not always easy.
Getting to a state where those assumptions are true is HARD. And takes a lot of careful programming.
Yes. Once it is true, the code is easy to write. But only because a lot of effort went into making it easy. And keeping it easy is also hard. Without focused effort to keep the code clean and easy to modify, it starts to rot.
Particularly if LLMs plateau and intelligence becomes essentially commoditized. If they can't compete on models alone they'll start moving more and more up the stack.
Exactly this. LLMs are already some form of useful, and have some kind/level of intelligence. We don't need to get to AGI before AI has uses. It isn't a step function and it isn't all or nothing. Even if LLMs get no better than they are today, we will still have valid practical applications that use them.
> Even if LLMs get no better than they are today, we will still have valid practical applications that use them.
That's only true when they are heavily, heavily subsidized. If that $200/month spend could or should be $2000/month, or maybe even as high as $20,000/month, would it still be "practical"?
The truth is we don't really know what these tokens actually cost - the actual cost of the data center, the hardware in the datacenter, the fantastic amounts of electricity it uses daily, the warming of the planet, etc, etc...
Are you sure you want to frame them as "practical"?
Also isn't the point moot if fable is so scary it needs to be banned and open weight models are already close on its heels? Even if China never imported another Nvidia chip the models he's so scared of are already out of the bag. At this point democratizing access seems like the best path forward.
Ive been using btrfs snapshots and some auto generated isolation rules plus a git ceiling at the mount root for the btrfs image (have to do this in wsl, stupid work computer). it's worked really well and fable hasn't had any issues with the "sandbox" (obviously not really but it works well enough)
I've been using something similar with ZFS, 15min frequency with autopruning (Sanoid) and the snapdir mounted for the agent. Has worked well and a big plus is being able to tell the agent to just solve it's mistake via restore from the snapdir.
I'd be curious to see the results, especially with some models having 1.5m and 2m context sizes, if the first 75% of the context was filled with unrelated info.
People don't seem to be able to reconcile the fact that there is likely an overbuild and overspend on AI that may be inflating a bubble, and that AI is actually incredibly useful and getting really really good for certain tasks. Both camps are right, except for when they say the other is wrong.
reply