KronisLV

5 ms·

31.
by KronisLV 1mo ago
No MTP?
32.
by KronisLV 1mo ago
> Download ollama and give it a try Weren't people pissed off at them for a variety of reasons? Not giving attribution etc. In case anyone wants other good alternatives, there's llama.cpp and vLLM, each harder to setup but open
33.
by KronisLV 1mo ago
It's surprising how on High reasoning it actually isn't that much more expensive than Sol, in addition to being better.
34.
by KronisLV 1mo ago
> The new app is an unbelievable mess. How so? On Windows, I like it (version 26.825.51511). Works fast (like modern Claude Code desktop, their old versions were bad), has pretty okay UI (could have an option for wider main text display,
35.
by KronisLV 1mo ago
You are allowed (and encouraged) to do good things without being obligated to. I think it would be immensely cool if OpenAI donated some of that crapload of AI money to the people behind LibreOffice to help with development - and then havin
36.
by KronisLV 2mo ago
Same, currently on a mix of Kimi Vivace (K3), GLM Max (5.3 and 5.3 Flash) and OpenAI Max (Sol and Terra mostly). I will say that Kimi feels nice but slow, GLM feels faster but has limited tokens (even off-peak) and OpenAI is nice and fast b
37.
by KronisLV 2mo ago
> Surely passing untrusted input to a agent with execution capabilities Oh hey, I know this one! It’s humans and a phishing test, they’ll click on all the links and enter information without checking the domain properly! Personally, I’d
38.
by KronisLV 2mo ago
Cool article, really cool website (the font could also be a bit smaller, if it didn't mess with the fidelity), but the grayscale setup just doesn’t have enough contrast for my eyes to reliably pick up on it at a glance.
39.
by KronisLV 2mo ago
> I’d like out, but the dealers are on every corner. Since the industry stops for noone (and you kinda have to keep learning in software dev) and due to the state of the economy it feels like one big rat race to me, I don't foresee
40.
by KronisLV 2mo ago
Pass data around. Run a local HTTP server or any other type of socket, like how Docker does.
41.
by KronisLV 2mo ago
> In the non China countries. That's a pretty big addressable market.
42.
by KronisLV 2mo ago
It would be okay if it were a bit less AI slop in regards to prose, had more conventional fonts and also removed the obnoxious glow effects. Then again, personal opinion.
43.
by KronisLV 2mo ago
I'll go one step further: anything you can do in a GUI should also be possible to do through a CLI and yes, also with code libraries. If webapps have n-tier architectures, then so should GUI software. (we can lie and say that this is t
44.
by KronisLV 2mo ago
> You could justify the jabs when we could all blame any outage on the migration to Azure, but then they shared numbers around the scale they're dealing with now that everyone is constantly building and pushing with AI. Does nobody
45.
by KronisLV 2mo ago
I guess this is also a really good indicator of just how much more tokens people would use when not limited financially. Wonder how much that tells the labs about pricing. Then again, it's 2.3x not 10x the next paid/cheap model.
46.
by KronisLV 2mo ago
> Sure they can - especially when conferences are designed to have an API. This sounds about as realistic as all of the crypto and blockchain people talking about distributed governance and such. The API part and agents pulling data is f
47.
by KronisLV 2mo ago
> this team, this business These get overlooked so often. The way you build software if you’re at the helm vs the way you need to build it when dealing with a more/less capable team and business, especially if someone else will be d
48.
by KronisLV 2mo ago
> when Anthropic finally collapses their 50% usage increase at the end of August, I'll definitely be forced to switch my workflow I cancelled my Anthropic subscription until they fix how their models write and it's no longer un
49.
by KronisLV 2mo ago
Consider the reputation implications of them banning someone who can get their complaint about it to the front page of HN and into the YouTube drama loop. They’d get swarmed with activist cancellations. At most I suspect the A.I. providers
50.
by KronisLV 2mo ago
I don’t know how 5 can do so much better in benchmarks but absolutely suck to use in practice compared to 4.X. Fable feels better, Kimi and GLM also feel better sometimes but tbh all of them make plenty of annoying mistakes.
51.
by KronisLV 2mo ago
But if you can use cloud models, why wouldn’t you use SOTA? For 2400 USD or less per year you can get pretty huge amounts of benefit out of that (though at the whim of whoever you are giving the money to).
52.
by KronisLV 2mo ago
If there’s news, then yes. This is a pretty great new release for those still stuck on Qwen3.6 35B A3B if they have enough memory but don’t have super powerful compute. I wonder if I could get this running through vLLM on 6x Nvidia L4 - the
53.
by KronisLV 2mo ago
> switch to some newer radically different paradigm. What if there won’t be a critical mass of fediverse users for it to matter enough and centralised social media platforms are all that we’ll ever get? Asking because things don’t always
54.
by KronisLV 2mo ago
> What is the main use case for X these days? Is it still as popular as it used to be? No idea comparatively, but an artist friend of mine tried out Bluesky and then complained about how there's way more of a limited reach and engag
55.
by KronisLV 2mo ago
I’m currently hoping that my Intel Arc B580 doesn’t die on me cause even a 5060 Ti 16 GB has reached like 700 EUR price over here. I don’t think the prices will get better anytime soon. Same with my M1 Air for being on the move, even notebo
56.
by KronisLV 2mo ago
I really liked Bootstrap because it was consistent and usable - making web just a tad easier to navigate. This modern AI website stuff feels obnoxious - annoying short punchy sentences together with paragraphs of text that meander and don’t
57.
by KronisLV 2mo ago
> It doesn’t work for the vast majority of tickets he attempts because he doesn’t have the necessary understanding to even start thinking about if the solution that the autocomplete generates is even remotely workable. And that’s with fa
58.
by KronisLV 2mo ago
> a fragmented national model These approaches will waste billions upon billions of Euros for no good reason. Having different VAT rates per-country already leads to overcomplication in so many places.
59.
by KronisLV 2mo ago
> Why is it always so hard for MS to say "Sorry, we made a mistake". Seems like no corpos ever want to say that. Surely it’s not because of some pride or PR? I’m guessing such an admission might also sometimes have legal risks?
60.
by KronisLV 2mo ago
Someone should add Kanboard for project management tools (like Trello): https://kanboard.org/ Really lightweight and nice!

More ›