Would it be intesting/useful to use jev to generate a block of text like LLMs do ?
Like asking it to pick the n + 1 word given the starting text (using it's choice primitive), but also asking n + 2,n+3 and so on at the same time.
Would it give coherent or useful results ? Or does the fact that it computes it "all at once" means it cannot make one of it's answer influence the other ones ?
Not an expert at all here, but I saw a comment on the jev post saying that it you constrain an LLM suck that it outputs a valid structure, if the token with the highest probability is not the one that you expected because of the structure (and so you pick the valid lower one) this means the LLM was already confused and your answer is less likely to be correct anyways.
You're halfway there. It's a little more messed up than that.
IIRC, this is all little-endian. So assume our 8 bytes labeled A-H. For a 32 bit/4 byte integer, they will be read as DCBA. For a 64 bit/8 byte integer, HGFEDCBA.
So the struct is set up like a = DBCA, b = HGFE. So when we assign 2 and 3 to a and b, it should look like this in memory:
00000010 00000000 00000000 00000000
and
00000011 00000000 00000000 00000000
When we cast that to a uint64 and assign 4 to it, we should wind up with:
That also reminds me of AI-generated text, which is also (when done without caring) barely readable because of the useless noise mixed the relevant information that a human that proofread the text would have thrown out.
Technically, AI can introduce rounds of fact-checking, proofreading, etc.
But even without that, AI-generated text is already better than what most humans can write -- particularly when they need something fast, cheap and good.
(It might not be "excellent" -- but humans wouldn't be able to make it fast and cheap...)
Of course if the text is made with care and re-read by a human it will be okay even if ai generated.
But I frequently encounter two issues:
1. Careless text generation, that was never read and so the human never prompted the LLm for proofreading and so on, this results in a text/readme/documentation that is using way too much words to describe simple or mildly complex things
2. LinkedIn post/corporate-trying-to-seem-friendly type of text that is always writing like it has to convince you of something (frequent user of the "it's not just X - It's y" trope)
I find that for most AI generated texts, it defaults to my 2nd point, and the human often doesn't read it or is happy with it, whereas I would have been much happier reading the shorter version they would have written instead.
Valve is kinda doing that with SteamOS ? It's primarily made for their own hardware, but they also use work time to add features for installs on non valve hardware, and all of their components are open source.
At the end of the day, a sandbox is a slice/cgroup. i.e.: Kernel's ability to create a complete bubble around a process or set of processes.
My beef is not with immutable OSes or sandboxes or else. They all have use cases and reasons.
My primary beef is being locked out of my own system without my consent. I want to own my personal (esp. Linux) systems, that's all. I'm also against erosion of Free Software ecosystem for something looking open, yet non-buildable or modifiable by the end user. This is why I don't like permissive licenses and essential software to be licensed-washed from Free to Open.
Would it be intesting/useful to use jev to generate a block of text like LLMs do ?
Like asking it to pick the n + 1 word given the starting text (using it's choice primitive), but also asking n + 2,n+3 and so on at the same time.
Would it give coherent or useful results ? Or does the fact that it computes it "all at once" means it cannot make one of it's answer influence the other ones ?
reply