But also, I'm on team "Its really hard to do the same mental job ~20 hours a week". I can do 2hr x 3 cycles x 5 days a week. But that means breaks.. When I did 12 hour days I was terrible at hours 9-12.
Is the suggestion they would work out more if they were paid? I think if obtaining Olympic medals was a function of training more then the avengers would be far higher.
The people high on that leaderboard are objectively ugly. Attractiveness is pretty consistent across individuals and cultures with only minor variations.
Theoretically it can be, though usually not, so the question is what should be the law to cover the general case. It wouldn’t be such a problem if it were easy for them to get around without driving. Either self driving cars, subsidized Ubers, public transit, walkable cities, home delivery, etc.
My opinion is that in the general case people over 70 shouldn’t be driving and I say this as someone who has 4 spritly grandparents in their 90s. I really don’t like how dangerous roads are, a fact that we accept because we did not really have good alternatives, now that we do we should implement them.
Is your point that we should be governed by the exceptions? I think that would be a bad idea. Does he even need a license for a racetrack? I’m sure he could easily afford Uber rides, and just maybe he would like to lower his odds of getting T-boned at an intersection by a geriatric.
The numbers are not arbitrary if they’re based on data, and generalizations are done for the sake of expediency and practicality. If such things are wholly unimportant then sure, capability test all the things.
They are arbitrary. You don't want to bake these things into law. What if people start living to be 150 as of next week because of some miracle drug. It'd be retarded if people lost their license at 70. Don't do things wrong just because you can. This is why software is full of so many bugs. Do shit right the first time, so that we never have to think about this again. jfc
I’m pretty sure laws can be changed easier than lifespans can doubled. You can’t always do things right the first time because knowledge unfolds with time, you’ll always know more later. You are proposing a waterfall design versus an iterative design. It would be easy enough to run an experiment for a few years to see if the lives saved are worth it.
There is nothing intrinsic to LLM prevents reproducibility. You can run them deterministically without adding noise, it would just be a lot slower to have a deterministic order of operations, which takes an already bad idea and makes it worse.
Please tell me how to do this with any of the inference providers or a tool like llama.cpp, and make it work across machines/GPUs. I think you could maybe get close to deterministic output, but you'll always risk having some level of randomness in the output.
It's just arithmetic, and computer arithmetic is deterministic.
On a practical level, existing implementations are nondeterministic because they don't take care to always perform mathematically commutative operations in the same order every time. Floating-point arithmetic is not commutative, so those variations change the output. It's absolutely possible to fix this and perform the operations in the same order every time, implementors just don't bother. It's not very useful, especially when almost everything runs with a non-zero temperature.
I think the whole nondeterminism thing is overblown anyway. Mathematical nondeterminism and practical nondeterminism aren't the same thing. With a compiler, it's not just that identical input produces identical output. It's also that semantically identical input produces semantically identical output. If I add an extra space somewhere whitespace isn't significant in the language I'm using, this should not change the output (aside from debug info that includes column numbers, anyway). My deterministic JSON decoder should not only decode the same values for two runs on identical JSON, a change in one value in the input should produce the same values in the output except for the one that changed.
LLMs inherently fail at this regardless of temperature or determinism.
Just because you can’t do it with your chosen tools it does not mean it cannot be done. I’ve already granted the premise that it is impractical. Unless there is a framework that already guarantees determinism you’ll have to roll your own, which honestly isn’t that hard to do. You won’t get competitive performance but that’s already being sacrificed for determinism so you wouldn’t get that anyway.
Maybe if the law required all knives to be pink they might be too embarrassed to murder someone. One problem then is the switch to acid attacks which are just clear liquids in containers.
It’s becoming a thing, police don’t like to report on it because they don’t want to give people ideas. They didn’t want to report on Glock switches either. I do machining as a hobby and am interested in machining guns from an academic challenge perspective, I’ve not done it because I focus on making things I can’t buy. Guns from an academic perspective are fascinating, we’ve been making them for a long time in just about every possible way, and there is an easy way to measure and communicate quality, I.e. does it shoot and how accurate is it. I think the ban is absurd, the tech to make 3D printers / CNCs is pretty generic and someone sufficiently motivated to make a gun is unlikely to have difficulty putting together the machines to do it.
reply