stared

4 ms·

1.
by stared 6d ago
In the line, I recommend „The Egg” by Andy Weir https://www.galactanet.com/oneoff/theegg_mod.html
2.
Claude couldn't hack OpenAI. Then Anthropic shipped Opus 5(thenewstack.io)
13 points by stared 7d ago|1 comments
3.
GPT-6 Astra solves puzzles(quesma.com)
2 points by stared 12d ago|0 comments
4.
Knowledge vs. wisdom: asking AI "What mushroom is that?"(quesma.com)
2 points by stared 17d ago|0 comments
5.
by stared 17d ago
If you like, you can run these tests yourself as well. It is ~$500 per a single combination, and assuming no failed runs.
6.
by stared 17d ago
Nope. If you would like to do so, it is easy (and orders of magnitude cheaper) than running benchmarks.
7.
by stared 17d ago
Before this experiment I tried to run KLD on various context lengths to see a quantization-dependent deterioration. On wikitext2 there was no difference. I concluded these have no long-term dependency and I should use Linux kernel. Still, t
8.
by stared 18d ago
A single result is binary. All we get from a run is which tasks were solved, which weren’t.
9.
by stared 18d ago
Nice! Sometimes the simplest approaches work the best.
10.
by stared 18d ago
I am curious what's the actual formula. I mean, there so many headers and layers, it is tricky to make a choice that will resonate with our intuition . Is it some weighted average? Or maybe ablation test?
11.
by stared 18d ago
Point taken, but there is a much more fundamental issue with it - and precisely why I wrote "very conservative". It is a different problem if we pick two sets from the same data distribution, A and B, and first we have a score on
12.
by stared 18d ago
I wrote this blog post myself, with AI for proofreading (typos and grammar, but not style). There were a few singular sentences for which I had a writer's block, but not much besides that. So, if there are irrelevant remarks, these are
13.
Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses(quesma.com)
285 points by stared 18d ago|140 comments
14.
by stared 18d ago
Nice! One thing I am missing is an easy „go up” a taxonomy group.
15.
by stared 18d ago
Is it a subtle effect that cannot be explained by Newtonian gravity (i.e. a different potential affecting, V(z) in the Hamiltonian)?
16.
by stared 19d ago
Don’t ask. Start from a few blog posts and see traction, read feedback. You will also see hos long it takes - and what is thd difference between an idea and making it real.
17.
I threw a ton of various logic puzzles at Astra and it solved of them(twitter.com)
1 points by stared 20d ago|0 comments
18.
GPT-6 Astra has autonomously completed Portal(twitter.com)
8 points by stared 20d ago|0 comments
19.
by stared 20d ago
An interesting catch! Maybe it suffices to make this pre-title "the ai-design-slop fingerprint · defs 2026.09" lowercase. In any case, from what I see what is AI aesthetics on the website: pre-title and horizontal lines.
20.
Does your landing page look generated?(slop-detect.com)
3 points by stared 20d ago|3 comments
21.
A connectomics milestone: Mapping the complete male fruit fly brain(research.google)
2 points by stared 20d ago|0 comments
22.
by stared 20d ago
With Internet, we live in an always-on culture. It contrasts with times before, when we actually had to wait for a monthly magazine, and even if we wanted to watch a movie, it was aired (say) the next Tue 7PM. While now we have all convenie
23.
Show HN: Genetic Distance Map(p.migdal.pl)
2 points by stared 21d ago|1 comments
24.
by stared 21d ago
While I like this index, calling in "Intelligence" might be confusing - it is a mix of coding and knowledge. Compare and contrast with ARC-AGI, BabaIsBench ( https://quesma.com/benchmarks/babaisbench/ ), o
25.
by stared 22d ago
Compare and contrast with tests on Qwen3.6 27B: * how does quantization hurts pelican-drawing skills ( https://quesma.com/blog/qwen-quantization-quality/ ) * how does it hurt knowledge ( https://quesma.com
26.
Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses(quesma.com)
3 points by stared 22d ago|1 comments
27.
Beef production in Brazil has been the largest driver of global deforestation(ourworldindata.org)
5 points by stared 24d ago|0 comments
28.
by stared 24d ago
Repo for reproducible science: https://github.com/stared/mushroom-hunting-llm-bench
29.
Mushroom hunting with LLMs: what can go wrong?(quesma.com)
56 points by stared 24d ago|76 comments
30.
Witcher 3's DLC Won't Include Russian Voice Acting(thegamer.com)
5 points by stared 24d ago|3 comments

More ›