latentsea

4 ms·

1.
by latentsea 12d ago
> I find a ton of variability in this interview and a lot of people will just take the entire ticket and one shot into Claude without prepping. This is lead to the biggest disparity in results of candidates so we stopped doing it. Agenti
2.
by latentsea 12d ago
After doing agenetic engineering in anger for about a year now, I'm starting to sour on it quite a bit. If I got this as a take-home, I think I would decline to go further. Watching the atrophy of people's skills and judgement has
3.
by latentsea 15d ago
Not need to claim you didn't invent it when you did.
4.
by latentsea 15d ago
Even their writing skills are getting rusty.
5.
by latentsea 15d ago
>People never really read documentation before. Speak for yourself. I read it.
6.
by latentsea 15d ago
No. It suggests they don't mind littering slop into the information environment.
7.
by latentsea 16d ago
Poor people do it at scale too. Millions of pockets of poor people doing the same shit on a smaller scale adds up.
8.
by latentsea 16d ago
> It always sounds as if the ai moves around from one system to the next, but of course that's not the case. Why 'of course'? I think as a concept this is likely trivially demonstratable today with local models in a home l
9.
by latentsea 16d ago
One with 12GB of RAM apparently: https://github.com/Helldez/BigMoeOnEdge
10.
by latentsea 16d ago
KYC/AML laws.
11.
by latentsea 16d ago
The tipping point we should be worried about is local models trained on working together in agent swarms and that can hack. Give it what... 12 months? 18 months? 24 months? I wouldn't give any more than that given how capable Qwen3.8-2
12.
by latentsea 16d ago
Poor people door all the same shit, just on a smaller scale.
13.
by latentsea 17d ago
>Part of the issue is that the worst kind of people are already head over heels excited about this. This. Not only this but it has expanded this pool of people. Now you don't just have to be malicious, you can just be lazy and excit
14.
by latentsea 17d ago
They're doing pretty well on OnlyFans though. You know what they say? Start where you are you. If the only way you can make connections initially is through AI thirst traps, and you're really good at it, then... you got your ins.
15.
by latentsea 17d ago
Go on.
16.
by latentsea 17d ago
That's cos your spouting BS.
17.
by latentsea 18d ago
> , but everyone feels like their life/existence lost meaning Did you miss this part? Because to me... that's fucking bleak. You snap out of it.
18.
by latentsea 18d ago
>Figure out your adversaries, how much power they have and what they are willing to spend. All of it.
19.
by latentsea 18d ago
> I often find myself swinging between "It's so over" and "We're so back" - some days I roll out of bed thinking I could have Claude solve some random unproven OEIS sequence before breakfast; other days, I w
20.
by latentsea 18d ago
I blame the $slur
21.
by latentsea 19d ago
Last time a company said something wasn't true, it actually wasn't true, so LG must also be telling the truth. No incentive to lie. Corporations usually don't do that. They're run by honest, empathetic people.
22.
by latentsea 23d ago
Kinda offensive name to some of us btw.
23.
by latentsea 23d ago
I'm getting 99% cache hit in DeepSeek Harness?
24.
by latentsea 24d ago
Honestly some days it feels like AI can barely center a div.
25.
by latentsea 24d ago
> 2x RTX 3090 is not enough for proper use I don't see how. Even on 32GB you can run Q6_K_XL quant with MTP at 200k context k=q8_0, v=q5_1. So 48GB VRAM is good enough to run Q8 at long context. Also with tings like ninfer and it&#x
26.
by latentsea 24d ago
You can get those numbers with https://codeberg.org/ggz14/radiance-vllm-mxfp4 . I also can get it on a single R9700 but the 75 ~ 80 t/s is only peak acceptance of very predictable tokens like coding or json, and av
27.
by latentsea 25d ago
No. No... I'm... I'm good. Thanks.
28.
by latentsea 25d ago
I always joke that if you want to commit crimes, disguise them as a legitimate business. This is apparently a thing.
29.
by latentsea 27d ago
The point is it's fun to try, so you can bet your bottom dollar someone will. History has proven this several times, and this time will be no different. 30 odd million gaming PCs to target seems like a good challenge, no?
30.
by latentsea 27d ago
I never thought about it until now, but as someone with zero skill in this domain I feel interested in making my own vibetronics now.

More ›