• 1 Post
  • 32 Comments
Joined 1 year ago
cake
Cake day: June 19th, 2023

help-circle

  • It cannot “analyze” it. It’s fundamentally not how LLM’s work. The LLM has a finite set of “tokens”: words and word-pieces like “dog”, “house”, but also like “berry” and “straw” or “rasp”. When it reads the input it splits the words into the recognized tokens. It’s like a lookup table. The input becomes “token15, token20043, token1923, token984, token1234, …” and so on. The LLM “thinks” of these tokens as coordinates in a very high dimensional space. But it cannot go back and examine the actual contents (letters) in each token. It has to get the information about the number or “r” from somewhere else. So it has likely ingested some texts where the number of "r"s in strawberry is discussed. But it can never actually “test” it.

    A completely new architecture or paradigm is needed to make these LLM’s capable of reading letter by letter and keep some kind of count-memory.


  • I have no idea how to fix the problem, but I’ve read somewhere that burn (a relatively new machine learning framework in Rust) is capable of loading models like stable diffusion. As Burn is built with webGPU and all the shader transpiler-stuff that comes with it doesn’t that mean that it can also run easily on (even older) AMD cards? I think what’s lacking is equal performance as nvidia drivers are heavily optimized already.

    Maybe someone knows more here?



  • IIRC most successful VCs invest very early and get out often early-ish too. The real enshittification that dangers the actual position of the company often happen much later. At that point the company is traded publicly and there’s a large anonymous body of shareholders - they only care about profits. VCs are actually a little smarter and care about longer time frames as in that early stage often much larger (relative) growth rates are possible.

    At a late stage (think Google, Twitter, Facebook, Reddit etc today) growth is much more difficult. How could Google grow today? They’ve saturated the search market years ago. So the only way of making more money is by sucking more money out of their existing user base. And they absolutely need to do it, as there’s huge pressure on the managerial class to do it, because the shareholders demand it. If the managerial class doesn’t do this (because often some older idealistic people know it would compromise the quality of the product), or they aren’t capable of doing it - they will get replaced by people who are more willing or capable - even if it’s detrimental for the company when viewed longer-term. VCs i would argue care all about profits, “but”. (they are smart enough to see the big picture. They are also small enough or “few enough” that they can communicate among themselves in order to agree on a more wise plan. That’s why they often get out once most of the possible (easy) growth has been achieved. They either know that now growth is much more difficult, or that the company’s value is much more stagnant - ow might decrease even. They can get out and invest their money in other more promising endeavours.

    The shareholders of large publicly traded companies are not that coordinated as they cannot really agree on anything other than just “growth”. More sophisticated strategies would have to be negotiated (and communicated) among thousands. The only unifying bond among shareholders is that they want profits. Think about it: many shareholders often don’t even know what companies they own as they are often part of other investment packages. Maybe you’re retirement plan has invested in stocks of 50 different companies, or 10 different fonds that have invested in others still. That is a form of dilution (?). It’s very difficult to communicate any strategy more sophisticated than “profits”. (a side effect is also that many people have invested indirectly or wothout knowing in endeavours that make their life more shitty/expensive when they retire - without knowing it.) There isn’t enough nuance in the wants of the masses as to want any more sophisticated strategy than simply “growth”. That’s why only short term growth can be thought.

    Of course sometimes also large companies can grow 2.5x or something like that. But it’s rare and takes more time. The exception makes the rule here. Early stage growth that VCs bank on is much more explosive i think. More like 10x or 100x.

    EDIT: sorry i typed this on mobile and it shows.





  • In this case it’s a giant housing shortage though. The city (and large surrounding area) is Freiburg in the south. Rents are so expensive and available flats are so rare that companies don’t find workers who could actually live there. Also: the comparably good loans don’t mean much when it’s only channeled into a greedy landlord’s pockets.

    Edit: oh no i was wrong it’s Nuremberg - their public transport organization is also “VAG”. But Freiburg has a huge labor shortage due to unaffordable housing and housing shortage.


  • Have you been to Europe? Have you walked the streets of Paris? The US was built with enough space being everywhere. American roads are wider, cities are mostly built like square-grids of roads built in a time when cats existed whereas European cities emerged in the middle ages. They’re tightly packed with little extra space. Sometimes (very rarely) here there are old Cadillacs that can be rented for weddings. Seeing one of these cars on the street is an unreal experience. They’re just so huge. They don’t fit on the streets here - and those are cars from the 60s or 70s. Everything seems tiny compared to them. From a European perspective it’s really stupid to build such large vehicles as driving and parking it is much more complicated when everything is build for small cars. Now that SUVs are becoming popular here too it’s just a really annoying. Less parking space per vehicle etc. On cities like Paris - one of the tightest city on Europe this is just annoying. And i haven’t even written about fuel consumption. Paris has had huge problems with smog in recent years.






  • Oh, i have a brilliant one:

    A few years ago i spent a lot of time converting .flac-files into .ogg-files in order to put on my oldschool iPod. As I did a lot of repetitive typing - entering $dir / for file in flac ; do convert etc / mkdir -p $somewhere/$artist/$album / mv $somewhere/.ogg->$new_dir/ and so on - I thought: “hm lets just write a loop over loops for all the artists here and then all the albums and at the same time create the nested directories somewhere else… hm actually in the home directory… and later love everything on the iPod at once.”

    so i was in my music folder with the artists-folders i wanted to convert. i did something wrong

    So i did my complicated script directly in the shell. I made something wrong and instead of creating a folder “~/artist/album” I created 3 folders in my current working directory: “~”, “artist” and “album”. hmph dammit gotta try again… but first : i have to clean up these useless folders in the current dir. so i type of course this: "$ rm -r ~ artist album " after about 5 seconds of wondering why it took so long i realized my error. o_O I stopped the running command, but it was (of course) too late and i bricked my current installation. All the half-deleted config files made or impossible to start normally and extremely tedious to repair it by hand, so i reinstalled.



  • I worked as a projectionist in 2009 when the cinema got its first digital projector in order to be able to show Avatar in 3D. At the start of the movie no one actually knew if it would work. Due to the movie being encrypted - with every cinema in Germany waiting eagerly for the password - No cinema was able to play the movie. But everywhere cinemas were packed with people. Because of fuckups somewhere in this incredibly stupid system the movie was delayed about half an hour (IIRC) nationwide. With no-one knowing if it would eventually work - especially nice for the people working at the cinema having to deal with angry audience members.

    At the same time the 2D 35mm film-version we also had started without any problems (it was massive and pretty dicey to carry it around).