Anyone downvoting this has:
- Not looked at the repo
- Not looked at the community
- Has a hate boner for the two letters L and M if combined in a certain way.
- Missed out
Sneak peak for people too lazy to read the very short readme:
Does it actually work? About just as much as any reasonable person would expect
You are a nicer person than I am! ❤️
The AI disclaimer is brilliant:
Ironically, aside from the obvious use of LLMs for the actuall project, everything was written by hand. I wanted to use this as a learning opportunity about git filters, so everything you see is my fault.
I remember GPT2TC, LLMs were a very interesting niche little proof of concept software area for a few years before <incoherent exasperated gesticulations>
Well, I’m not going to downvote, but something in my brain is screaming “lossy compression!” and so you might say I’m at least wary.
/dev/null compresses everything perfectly. The hard part is recovery and is left to the reader.
/dev/null is also webscale
infinite compression ratio :o
That’s the joke.
It is a tremendously stupid idea, which is why I found it funny (also, it’s one of those things where name came first and inspired the rest).
But, to take it seriously for a moment (and I’m not trying to defend it, not a fan of LLMs as most of the people here), is it necessarily lossy? I mean, you basically have to include the 5Gb large model for it to work, and you just move the data from the file to hoping the summary can trigger a correct combination of parameters.
I didn’t run any larger tests, and I assume that if you managed to keep the API/function names and behavior, the summary would be actually longer than the actual implementation in most cases anyway, so it’s probably not even a compression (especially if you include the model).
It’s just a food for thought, it’s definitely a bad idea to do something like this, to the point where I’m pretty sure you could get millions from investors if you made a startup working on something like this (and that one already exists), but I do honestly wonder if the fact that you kind of have the data in the model would still count as lossy.
When I first saw this, I thought it was obviously a toy to learn something and as an excuse use an LLM. I wasn’t sure if the use of an LLM was serious or not (e.g. did the originator know the limitations); however, I did take it seriously for the sake of argument after one of the other comments talked about compression.
I decided that I actually like this project quite a bit as an example to introduce information theory to students. What the compression comment gets wrong imo is that compression is about taking an instantiation of information, reducing it to a minimum set of measurable units of infirmation (i.e. bits) to recreate that instantiation.
What this would aim to do, assuming it works (or rather what a competent human could do) is reduce the instantiation to a set of information that can recreate an instantiation of said information (changing the problem into a non-unique inversion problem) to recreate functionality but not an exact duplicate of the input. To the degree that the description converges on one result then one would end up with the exact same input; however, as human language is notoriously verbose and imprecise this would not compress anything.
Kudos.
I think Microsoft have been using something like this for years.
I have read an article about their own engineer has forgotten how to write native application for their own fucking operating system. I wonder even a LLM can make good win32 app.
Haven’t looked at the repo yet, but I hope it just
rm -rf’s your entire git repo.So… Clean build? But with the added bonus of risking some AI destroying the code? Cool, just what the world needed: a worse way of doing something you been doing effectively for a couple of decades
