kingstnap

5 ms·

1.
by kingstnap 6d ago
Performance problems in theorem provers is an old topic. I remember watching this and it was fun. https://youtu.be/m-iGCCuHBvY [Talk] 10 years of superlinear slowness in Coq (2022)
2.
by kingstnap 6d ago
This is pretty interesting in a lot of non-surface-level ways. I can see OpenAI pushing for this as a sort of more durable moat compared to the now huge number of agentic harnesses that run on your own machine. This might be getting the foo
3.
by kingstnap 8d ago
Replace hire with marry and this is *exactly* what politics was globally for all of human history. This is not something "behind us" btw.
4.
by kingstnap 10d ago
> hey Claude vibeslop me a CUDA Anubis solver" route is on its way to being fundamentally dead. Lmao yeah no. I don't think a little argon2 is going to change shit all. I mean the thesis of Anubis itself is "scrappers are
5.
by kingstnap 12d ago
In real life, there is always this feedback edge from the results to the methodology. Theoretically it's unscientific to do tweaks like this but in reality this is what actual science is because you need to see the results understand t
6.
by kingstnap 12d ago
Its also available finally to Pro users! Just took 24 hours.
7.
by kingstnap 12d ago
The AI labs have out considerable effort in trying to find and patch lean exploits. They explicitly set agents and have them try to prove false. > Daniel used OpenAI internal models to discover new soundness issues in the official Lean k
8.
by kingstnap 13d ago
Are you guys just posting what you think might end being the link so you can farm upvotes or something?
9.
by kingstnap 14d ago
This seems to be the announcement w/ the trust me bro numbers. https://x.com/Alibaba_Qwen/status/2094968708288680276
10.
by kingstnap 14d ago
They are simply measuring in base 1.7 :^) Uber has grown 2 orders of magnitude!
11.
by kingstnap 14d ago
Hosting is probably ~0% of the cost lol. The costs of the writers is what makes journalism expensive.
12.
by kingstnap 14d ago
The blog post is gone but I can currently use it in the gemini chat website.
13.
by kingstnap 15d ago
Do API prices not affect usage limits for subscriptions? They do in Codex.
14.
by kingstnap 15d ago
Haiku would have to be a banger, with a significant price drop, to make any sense. It's currently priced 33% above Gemini 3.7 Flash, and several multiples of 5.6 Luna.
15.
by kingstnap 16d ago
I think the jobs thing will be a slowly then all at once thing. Firstly, AI is sort of like this to begin with. It's useless for something until it gets hill climbed and is useful for something. AI agents for chip design was an idiotic
16.
by kingstnap 19d ago
Man that claude writing is bad. Really bad.
17.
by kingstnap 20d ago
Smaller models + more effort has strong diminishing returns, especially if your goal is to save money. Sol already lacks judgement. It will absolutely add idiotic tests and comments. Luna is that but worse so if you account for things like
18.
by kingstnap 21d ago
Italy too. It's bizarre trying to hop on the trenitalia trains while a bunch of random people are standing on the platform right in front of the doors trying to get a few last drags in before the train leaves.
19.
by kingstnap 21d ago
A random blight in Canada in that Bell still hasn't enabled IPv6 for residential plans. You can get 8 Gigabit symmetric home internet from Bell but not IPv6.
20.
by kingstnap 21d ago
Whats stopping you? You could buy puts right now. Get a 210 strike put contract and if your thesis is that nvidias current 10 day slide continues you could make some money.
21.
by kingstnap 22d ago
> Data centers in the states analyzed depend on about 3.4 trillion gallons of freshwater annually for electricity, around 12 times the annual water use of Los Angeles, Phoenix, and Washington D.C., combined. > Most electricity is gene
22.
by kingstnap 23d ago
If you think of LLMs as programs. The weights and inference code are very much a binary. While the training code and data are the true source. Since if you want to robustly modify the LLM that's actually what you need. But since "
23.
by kingstnap 29d ago
You don't need to seperate the clean and dirty air. Mixing them isn't as efficient but it should still be roughly 63% efficient (1-1/e).
24.
by kingstnap 1mo ago
> Overall my view is that AI is structurally a technology that tends to concentrate power > Open-weights do help some with this but are nowhere near a sufficient solution Which is exactly why Anthropic contributes nothing to, and ac
25.
by kingstnap 1mo ago
On the individual level it is a discipline problem. But the fact it has become an individual discipline problem is exactly the core issue. If you turn things into individual discipline problems then surprise you get population level issues.
26.
by kingstnap 1mo ago
Counting macros is extremely effective. The general population doesn't have the discipline for it, but amoung sport coaches, actors, models, bodybuilders, this is *the* technique for achieving goals. Anyway what's cute about this
27.
by kingstnap 1mo ago
You would think raw performance wouldn't be a problem given most of whats happening is waiting for network calls and streaming tokens. But modern bloat manages perfectly well to make apps that wait for network calls run poorly enough t
28.
by kingstnap 1mo ago
The $60 billion cursor option that SpaceX bought was exercised on June 16th. The deal is closed.
29.
by kingstnap 1mo ago
It could also partly be a byproduct of examples of claude writing being in the dataset, which of course anthropic has lots and lots of and they do train on.
30.
by kingstnap 1mo ago
Lets play over/under on an AI model proving (or counter exampling) the Riemann hypothesis? I'm not sure what a good mark would be, but considering this result lets put it at 2027-08-10 (One year from today).

More ›