By pointing out the exact things that will likely happen you are oddly enough hedging against (at least some of them) happening!
A) I reckon it's true that smaller models will continue to improve massively through optimization and better and better harnesses, this tech is all still very young and A LOT of resources and (good-)will is being thrown at it.
B) The 1T+ models will be able to sideload and improve upon a lot of the fundamental improvements that happen to the smaller models to speed up incredibly while getting better at tools while (on a gradient) getting -more- things right.
C) More of an observation that I think is worth keeping in mind clearly; Karl Popper's black swan and all, truth in our temporal world IS a gradient!
Worse in my opinion since the look is simply Tesla (whether one likes that or not), no one would have blinked an eyelid if Tesla released this car whereas Ferrari doing so comes off incoherent.
Excellent practical guide and pictures, if OP is around on this thread: well done! Your future self is going to appreciative too when this needs repeating at some point!
These days it's for sure the dev environment that is lacking, hardware is okay (potentially great?!), software abysmal. To run a local llm in a stable manner implies using Vulkan.. any attempt at ROCm is totally hamstrung by haphazard support of hardware alongside with an online presence poisoned by people primarily discussing work-arounds rather than work when it comes to AMD as a platform. Argh.
It's sad that they have gone political whereas their goal should, in my optics, be almost technocratically in favour of their own stated goals of "protecting user privacy from government/corporate surveillance, defending free speech online, enforcing net neutrality, promoting encryption, and combating abusive intellectual property laws".
I do find the down votes odd, comment seem to contribute to the discussion, is it a reflex move because of intense dislike of the sitting US president?
Having your stuff stored in another country is ultimately a voucher of confidence, I can't see Trump or anyone else willingly misusing that trust. I do think the Western leaders need to temper their tendencies for isolationism these days, what's the alternative guys? And why even think that other people/cultures will want you in their swimming-pool if you can't keep your own one clean/functional?
(this comment also covers France recently bringing home their gold)
But then there's this: "When evaluating the complete bun install improvements, it came out speed-wise to about the same as the existing git usage (due to networking being the big bottleneck time-wise despite more cases being slightly faster with ziggit over multiple benchmarks). Except, it's done in 100% zig and those internal improvements pile up as projects consist of more git dependencies. All in all, it seems like a sensible upstream contribution."
Sooo, after burning these 10k+ worth of tokens we find out that it's sensible to use it because the language (zig) feels good as opposed to git itself which now has +20 years of human eval on it. That seems. Well. Yeah...
Yep. that's the route I tried before, no good, maybe it's just that the documentation is past it's sell by date, maybe it's lack of community use.. I'm just not seeing it. Even the article itself describes how to make an exe file... that will then work in Linux? Or is it simply a program that's easier to run on Wine? Loads of text with unclear details throughout it.
With this exact point in mind: I've recently written a pretty straight forward win32 c implementation of a utility with some context dependent window interactions and a tray icon to help monitor and facility reload of config file.
Is there any way I can use the Wine project to facilitate this compiling and running straight under x11/linux environment as a integrated project that doesn't require the end user to fiddle with Wine? I don't mind bundling shared code as needed. Help appreciated, I tried hard and failed at this endeavour priorly.
This point stands to be underlined! Even the least possible friction is more than people at large are willing to deal with, it's only if the system changes are pushed from the top (rumblings in the EU block at the mo) that we'll see casual consumption of Linux in more mainstream context. Having run Linux Mint across a 50+ coworker setting from a sysadmin perspective this is entirely doable, most will not even notice as long as Chrome is in place alongside with something office-like.
I abandoned MacOS back in 2018 since I found it too quirky and poweruser-unfriendly (the main thing that comes to mind is neatly indicated by todays other MacOS related frontpage article on resizing). Now we can add overt instability to the list.
I use another distro but totally appreciate the effort to keep different branches of potential futures alive. Humans have a tendency, in tech and most other domains afaict, to put a lot of eggs in one basket because it's easier/allows-faster-moving-forward.. but that basket may have structural weaknesses that only shows once it has A LOT of eggs in it.
The title actually cringes me out a bit, it reads like early report titles in academia where young students (myself no doubt incl back when) try their hardest at making a title sound clever but in actuality only achieve obscuration of their own material.
Yeah.. this is exactly the overall reality of the ecosystem isn't it? That being said I do hope uv succeeds in their unification effort, there's nothing worse than relying on a smattering of diff package managers and built streams to get basic stuff working. It's like a messy workshop, it works but there's a implicit cost in terms of the lack of clarity and focus for the user. It's a cost I'm not willingly paying.
Talking about Python "somewhere in the middle" - I had a demo of a simple webview gtk app I wanted to run on vanilla Debian setup last night.. so I did the canonical-thing-of-the-month and used uv to instantiate a venv and pull the dependencies. Then attempted to run the code.. mayhem. Errors indicating that the right things were in place but that the code still couldn't run (?) and finally Python Core Dumped.. OK. This is (in some shape or form) what happens every single time I give Python a fresh go for an idea. Eventually Golang is more verbose (and I don't particularly like the mod.go system either) but once things compile.. they run. They don't attempt running or require xyz OS specific hack.
Yup, but 5 to 15% faster year on year is real progress and that's ultimately what the big user base of Python are counting on at this point.. and they seem to be getting it! Full disclaimer: I'm not a heavy Python user exactly due to the performance and build/distribution situation - it's just sad from a user-end perspective (I'm not addressing centralised web deployment here but rather decentralised distribution which I ultimately find more "real" and rewarding).
Anything out there for reference or would you be implementing from theory/ideas here? God speed to you in terms of the project overall, it's exciting to see the beginnings of a rust-like-lang without the headaches!
Sensible take, thank you. When HN get these "our project: from x to y language" frontpage stories I am always thinking that it would be far more exciting with "our project: 38.2% smaller code base by optimizing our dependency use", "our project: performance optimized by 16.4% by basic profiler use" or similar!