_3u10

6 ms·

1.
by _3u10 7d ago
That’s how increasing performance works. You make a model 10x faster, then you make it think 2x as much. Its cost is now 1/10th per token, and 1/5th per task. Basically they have shitty hardware so they have to do a lot of optimiz
2.
by _3u10 19d ago
National security excuses enjoy widespread bipartisan support. Want to execute a citizen without trial? National security. Want to execute his child without trial? National security.
3.
by _3u10 19d ago
It’s not for users it’s for people who think 1/4 of the traffic a raspberry pi could handle is scale. Or don’t know what a server is.
4.
by _3u10 19d ago
Just use whatever service that costs 10x as much to make it do what it was advertised to do in the first place, like DAX for dynamo or cloudfront for S3 in case you hit “scale” like 2000 req/sec
5.
by _3u10 2mo ago
He gave them the unlock code, now it’s unlocked.
6.
by _3u10 2mo ago
Umm restaurants ARE like that, why do you think there’s so many burger and pizza joints?
7.
by _3u10 2mo ago
You can’t train on our models trained on human text, that is stealing the IP we stole first!
8.
by _3u10 2mo ago
I don't live in a "first world" country, where I live people have cars, some even chinese ones. I would assume in the first world people can have cars and air conditioning, maybe not having AC, a car, or a home is a first wor
9.
by _3u10 2mo ago
Ask AI about castor beans and barley, it will stop all that nonsense.
10.
by _3u10 2mo ago
If you wait til the deadline then you create synergies and alignment with management as to whether it’s done.
11.
by _3u10 2mo ago
So we should allocate electricity to less productive uses to benefit people the most? People should just keep driving inefficient cars rather than having gas prices increase so they can evaluate whether getting a more efficient car makes se
12.
by _3u10 2mo ago
Usually a product not working with 99.995% of the eco system is a no go.
13.
by _3u10 2mo ago
What about 36 bit registers
14.
by _3u10 3mo ago
Yes but now they have a moral imperative to collect their ad money and aren’t a vested interest. “I’m shocked shocked to find out my pay cheque and stock money came from selling ads.” “here are your winnings sir”
15.
by _3u10 3mo ago
But it’s undemocratic to ask the people whether it’s working, the democratic way is to ask the media if they are doing a good job and then tell the people they are wrong.
16.
by _3u10 3mo ago
How can you tell the difference in a democratic system between something being popular and it being democratic?
17.
by _3u10 3mo ago
Yes it is the deliberate mechanism of deciding by popularity.
18.
by _3u10 3mo ago
lol. Wait til they find out that algorithms work on popularity. Aka democracy.
19.
by _3u10 3mo ago
Server side rendering on a serverless platform is all you need to know about “modern” front ends.
20.
by _3u10 3mo ago
They don’t, they wouldn’t need tariffs if they did. America mostly produces cheaper ag commodities than the EU but more expensive than South America. Deepseek is already a better search engine than Google. (Not sure if it does Google search
21.
by _3u10 3mo ago
I don’t think they have. I used to think that Claude sucked because of the harness so I tried it on opencode and it was still trash. I’m sure if the primary problem I was solving was multi file complex exploits it would be better but even t
22.
by _3u10 3mo ago
I don’t think the ai market has been around long enough to prove winner take all, I see it more like a commodity which is why the winner quickly switched from OpenAI to Anthropic at least on coding. I generally find the Chinese models to be
23.
by _3u10 3mo ago
Why I prefer Asuncion to Dubai in a nutshell. Chauffeur / Valet > parking apps Maids > dishwashers, laundry, roomba, cooking Fixers > everything else
24.
by _3u10 3mo ago
The outputs belong to whoever purchased them. What are they complaining about?
25.
by _3u10 3mo ago
I’ve rarely experienced this. Typically what is requested is code that has unpredictable pauses, takes unbounded time, has two kinds of null, etc.
26.
by _3u10 3mo ago
It’s actually pretty easy to distinguish AI from real text because all AI generated texts have 2.4 children. In aggregate it is indistinguishable statistically but for a given text it’s remarkably easy.
27.
by _3u10 3mo ago
Not really. You can still not use AI. The solid data is that RAM is more expensive than before… so there’s a small case to be made there because of you need a new computer you can’t avoid that.
28.
by _3u10 3mo ago
No, typically Brazilians go to Paraguay for their education, most of their technology comes from Paraguay too.
29.
by _3u10 3mo ago
No, they need the same arch, but you can distill them into a single model. And yes, if you use the API directly Claude will often say it’s an open weight model (likely the ones it was distilled from)
30.
by _3u10 3mo ago
Yeah, who needs censorship when Canadians attend no kings protests about a democratically elected leader of another country and not King Charles. Ask Claude a simple question, which is a more democratic country El Salvador or Canada. It’s s

More ›