Feyd

1. Floating point math is deterministic.
2. Systems don’t have to be programmed with race conditions. That is not a fundamental aspect of an LLM, but a design decision.
3. Systems don’t have to be programmed to tie break with random methods. That is not a fundamental aspect of an LLM, but a design decision.

This is not hard stuff to understand, if you understand computing.

You can, at that will cause the same output on the same input if there is no variation in floating point rounding errors. (True if the same code is running but easy when optimizing to hit a round up/down and if the tokens are very close the output will diverge)

There are more aspects to the randomness such as race conditions and intentionally nondeterministic tiebreaking when tokens have the same probability, apparently.

I actually think LLMs are ill suited for the vast majority of things people are currently using them for, and there are obviously the ethical problems with data centers bringing new fossil fuel power sources online, but the technology is interesting in and of itself

I said I was wrong in that my statement was overly broad and not applicable to the systems most people are using in my initial response to you, then clarified that it is not an intrinsic character of the technology at large but that the implementations that are most used have it. You apparently think that conversations are a battle with winners and losers so the fact you were right that the biggest systems are nondeterministic for reasons outside of temperature configuration means it doesn't matter why, doesn't matter that those factors don't have to apply to every inference system, and doesn't matter that you have no idea what determinism means. In any case talking to you seems like a waste of time, so enjoy your sad victory lap while I block you so I don't make the mistake of engaging you assuming you're an earnest interlocutor in the future.

Deterministic systems are always predictable, even if you never ran the system. Can you determine the output of an LLM with zero temperature without ever having ran it?

You don’t have to understand a deterministic system for it to be deterministic. You are making that up.

And even disregarding the above, no, they are still NOT deterministic systems

I conceded that setting temperature to 0 for an arbitrary system (including all the remote ones most people are using) does not mean it is deterministic after reading about other factors that influence inference in these systems. That does not mean there are not deterministic implementations of LLM inference, and repeating yourself with NO additional information and using CAPS does NOT make you more CORRECT lol.

You also have to run the model with the input to determine what the output will be, no way to determine it BEFORE running. With a deterministic system, if you know the code you can predict the output with 100% accuracy without ever running it.

This is not the definition of determinism. You are adding qualifications.

I did look it up and I see now there are other factors that aren’t under your control if you’re using a remote system, so I’ll amend my statement to say that you can have deterministic inference systems, but the big ones most people use cannot be configured to be by the user.

You can actually set it up to give the same outputs given the same inputs (temperature = 0). The variability is on purpose

Just fucking stop using it? Wtf? Tell you boss to pound sand! They're going to blame you when it goes south anyway so you might as well stay honest.

You're agreeing with the comment you replied to. Why the fuck are you trying to be so smug???

Every time you open LinkedIn in a Chrome-based browser, LinkedIn’s JavaScript executes a silent scan of your installed browser extensions.

Still wrong, but but not quite as scary as “searches their computer for installed software” which makes it sound like it broke out of browser sandbox

Arch Linux for desktop. Whatever for servers, usually ubuntu, though I’ve been tempted to use arch then too when dealing with out of date software.

1. Software is always up to date
   • instead of old with security backports which have always kind of skeeved me out
   • so upstream issue systems are relevant
   • so I get new features without having to wait
2. When I have to turn a wrench the wiki and forums are amazing
3. I feel like I control my system

There are only 5 games in that category of Zelda game in the first place.

• ocarina of time
• majora’s mask
• windwaker
• twilight princess
• skyward sword

Breath of the wild and tears of the kingdom are totally different formula and the rest are not 3d.

For me it kind of goes windwaker -> oot/majora/skyward -> twilight. So I’d say it’s up there, and the fact it’s so foundational should give it some extra points. I’m amazed they haven’t done a remake to up the graphics and give it modern controls tbh

I feel this way about all open world/map slop. I liked the first couple i played but after the novelty wears off it's an entire genre that sacrifices focus for a lot of lower quality content

At this point you're just trolling 🙄

I agree. You are the one that made specific comments about what carrying a first aid kit means for evidence that are completely irrelevant to the trial.

You completely dodged the actual question. Is a first aid kit evidence of planned terrorism?

hammering Wikipedia’s servers and using it as a training ground. They could just effectually download the freely available archive instead of being assholes but nooo

Iirc California had a similar proposal to this. I actually think it’s not a terrible idea at the core. It’s basically an API for parental controls. You set up a device (or account on a device) and say “this is a device for a kid” and that gets used for everything. It actually makes a lot of sense to do something in that direction. Part of the reason people are convinced something needs to be done is because managing parental controls across the different myriad services and apps is a labyrinth thattech savvy patents can barely navigate, and less savvy patents don’t stand a chance.

Unexpected???

Are they? What have they actually done to make it a bad choice? From my memory the last year or so they’ve: Added some AI stuff that I don’t use but isn’t in my way Added tab groups and sidebar tabs which I use constantly Clarified some legalese that didn’t change anything but caused everyone to say the sky is falling for no reason Refused to deprecate manifest v2 Don’t get me wrong; I’m not saying Mozilla is perfect and does everything I want. I wish they weren’t funded by google and I wish they didn’t waste so much money on executives, but not only do you need them to stand up to google by continuing to develop firefox (all the downstream gecko browsers would collapse), you need them to continue to participate in web standards committees so google doesn’t get to run away and do whatever the fuck it wants (no other non-big-tech entity has a seat at the table or would get one) I swear there is astroturfing for anti-mozilla sentiment to undermine their importance in the ecosystem

I’m usually a firefox defender, because usually people are blowing everything they do out of proportion, and straight up making up negatives. That said, I don’t know what a “modern AI browser” is but I know it’s something I don’t want. Hopefully “AI should always be a choice — something people can easily turn off.” Is a serious statement. But also, the AI features they’ve added so far have been out of my way and haven’t bothered me, and if what you’re calling for is to use chromium based browsers (gecko down stream is fine) then you are making the wrong decision for the future of the internet. Firefox, chrome, edge, and safari are the only browsers which have a snowballs chance of being accepted in corporate environments, and supporting google over Mozilla is a losing proposition.

There are plenty of cross platform frameworks and libraries that don’t involve web tech