trenchgun

3 ms·

1.
by trenchgun 2mo ago
Do you think Chinese lab researchers have not read Yudkovsky, or are not AGI pilled? Hate to break it to you, but they are racing to AGI with pretty same reasoning as western labs. Just recently DeepSeek researcher in his post was saying th
2.
by trenchgun 2mo ago
Yep. As is "can't know".
3.
by trenchgun 2mo ago
You do not even need a subscription. Even ChatGPT free account has some quota for Codex/Work use.
4.
by trenchgun 3mo ago
Why delete comments! Constrained grammar! Make invalid state unrepresentable.
5.
by trenchgun 3mo ago
Interesting, so they shipped slop? Succumbed to their own AI psychosis?
6.
by trenchgun 3mo ago
I did write small programs myself, but mostly with Claude Code. It was very pleasant to work in it with Claude Code. Implemented a shell and 40 of coreutils. Did not really get very far with proving properties of them, but had working progr
7.
by trenchgun 4mo ago
Lean4 is a general purpose programming language.
8.
by trenchgun 4mo ago
CMD+F "reward hacking" -> 0 hits. Fail.
9.
by trenchgun 4mo ago
They are already proven false through Jacobian conjecture having been proved false.
10.
by trenchgun 4mo ago
Yes, assuming the stuff it depends on is already formalized.
11.
by trenchgun 4mo ago
This is a product I have wanted to exist for quite a while, and tried to create myself, but lack the skill. Thank you for building this!
12.
by trenchgun 4mo ago
Then it is really not about language design... I would rather recommend https://www.plai.org/
13.
by trenchgun 4mo ago
I think mass deportation of people with wrong skin color goes beyond "stricter view on immigration" >large scale remigration https://en.wikipedia.org/wiki/Remigration
14.
Insert is a programming language for self-modifying code(github.com)
2 points by trenchgun 5mo ago|0 comments
15.
by trenchgun 5mo ago
>But when people think of decentralized training, they don’t first think of gigantic datacenters, owned by the same company, training models across large distances. Instead, they imagine thousands of small datacenters, or individual cons
16.
by trenchgun 5mo ago
Yes, and then you need to have the datacenter. Do you get a permit? How long does it take to build it?
17.
by trenchgun 5mo ago
This pattern reads heavily LLM in style, but also... that is spot on.
18.
by trenchgun 5mo ago
So... you did not read it, or did not understand what you read.
19.
by trenchgun 5mo ago
Where?
20.
by trenchgun 6mo ago
But everything else still works fine, right?
21.
by trenchgun 6mo ago
Article is mostly LLM written slop.
22.
by trenchgun 7mo ago
CEO observes performative work, and his inference will be that means more people need to be fired. Let only the AI native, customer obsessed 10x engineers(/ AI swarm managers) remain.
23.
by trenchgun 7mo ago
This is LLM written comment
24.
by trenchgun 7mo ago
How is this useful?
25.
by trenchgun 7mo ago
Which of the 78 copilots? https://teybannerman.github.io/strategy/2026/03/31/how-many-...
26.
by trenchgun 7mo ago
You’re a literature cycle behind. ‘Middle-layer shared representations exist’ is the observed phenomenon; ‘why exactly they form’ is the theory. You are also confusing ‘mechanistic explanation still incomplete’ with ‘empirical phenomenon un
27.
by trenchgun 8mo ago
Yes but this is probabilistic. Skill, documentation etc help by giving it the information it needs. You are then in the more correct probability distribution. Fine for docs, tips etc, but not good enough for mandatory things.
28.
by trenchgun 8mo ago
Yes this + also APIs! Due to LLMs needing APIs suddenly things that were strictly behind GUI can easily opened up for programmatic use. Double edged sword, due to infosec etc concerns, but sure is nice for personal productivity automation.
29.
by trenchgun 8mo ago
Do you run security review by agents over this?
30.
by trenchgun 8mo ago
Claude default frontend skills suck :D

More ›