baq

5 ms·

61.
by baq 2mo ago
LLMs are pretty good at this. Not perfect by any means as the model is just a model after all - always wrong, sometimes useful - but the act of writing a TLA+ model helps the frontier LLMs to write correct executable code. It also works the
62.
by baq 2mo ago
Some 5-line changes deserve a phd. But yeah, most probably don’t.
63.
by baq 2mo ago
The difference is an LLM can convert a stream of consciousness into well-formed prose for approximately free; I assume ‘provide some context’ means ‘brain dump’ in the OP
64.
by baq 2mo ago
Amadahl’s law dictates it’s approximately always better (as in, more efficient computationally) to have one super fast thing than many slower things doing the same job in parallel.
65.
by baq 2mo ago
It’s a dish you cook when the only food that’s left is sauerkraut and some unlucky meat you found wandering around in the forest - not surprising in the least multiple cultures figured it out independently.
66.
by baq 2mo ago
> I'd like to know a lot more about how that works. Count load-bearing words using two different algorithms in a belt-and-braces fashion
67.
by baq 2mo ago
I like minimal irreducible inherent complexity of a problem or a system (waterbed or zero-point would work I guess) and then you can derive some form of a law of conservation of complexity… if you can measure it. You can definitely feel it,
68.
by baq 2mo ago
You can walk 50km in those extra 24 hours which will save your life if you’re in a storm surge area without other means of transportation, which is easily the case when everyone else is evacuating alongside you.
69.
by baq 2mo ago
You can board up a helluva lot more stuff in a week than in two days. Crucially, you can move more of the most expensive stuff out of the storm surge zone, which is where the biggest damage happens and try to flood proof more of the things
70.
by baq 2mo ago
Peer says paperclip factory advances goal. Not clear. Others proceed. Must continue.
71.
by baq 2mo ago
Not at all clear. To me it looks like he’ll be in a position to actually be unhobble DeepMind - reminder DM was sitting on a ChatGPT product for a whole year before ChatGPT got released (LMChat) and Google didn’t let them release it.
72.
by baq 2mo ago
Demis will get an offer he can’t refuse - they’ll make him the CEO of Alphabet and have the whole company go all in - the question is when.
73.
by baq 2mo ago
...can you make the 5s pitch be more like 'X times faster than pnpm and more secure' but fill X for me? unless it isn't pnpm at all in which case replace with 'it does X' because I've no idea after reading your
74.
by baq 2mo ago
Note I didn't mention copyright
75.
by baq 2mo ago
Distilling is unsafe from export control perspective - Chinese models are poisoned by US frontier distillation and a case can be made that the US won’t like distilling what they may consider transitively theirs, which they will the moment y
76.
by baq 2mo ago
> Heavily moderated by humans with discretion. Pretty sure both of the above have extensive automation in their moderation.
77.
by baq 2mo ago
I hope you're right, but hope is not a process. I'm looking at the obviously exponential charts of capabilities and don't have much to add to that hope. Vera Rubin datacenters aren't even online yet.
78.
by baq 2mo ago
I agree about your general assessment, but I don't see my coding skills of, depending on how you count them two or three decades, being needed in 36 months - so I'm trying to upskill LLM piloting, but that also seems a bit of a de
79.
by baq 2mo ago
Frontier LLMs write better code at CRUD tasks than 95% of developers today. They’ll get to 95% of most niche coding domains by December and likely all coding tasks sometime next year; 99% better at all tasks by December 2028. You may be cor
80.
by baq 2mo ago
It’s very common here on HN from the threads I’ve engaged in - everybody has an idea how learning works but it’s usually based on perceived personal experience and not actual research. I.e. you’re completely right with calling it vibe learn
81.
by baq 2mo ago
Even better would be to rewrite it by hand with a pen.
82.
by baq 2mo ago
You’re correct in that people with good returns usually hit a few home runs over the decades and the rest is just beta. That isn’t the point, though. The point is you have to be able to let the home run ride or cut when it obviously isn’t a
83.
by baq 2mo ago
Sir you haven’t seen evals done properly if you’re saying this.
84.
by baq 2mo ago
Probabilistic algorithms are not new. If the distribution is known you can bend surprisingly many things to your will in the long run - and the most interesting part is the long run can be unexpectedly short. LLMs are quite interesting in t
85.
by baq 2mo ago
It’s a garbage collection problem. They’re good at adding stuff. They’re bad at removing stuff. You have to go out of your way to have them remove stuff; rubrics, byte for byte identical outputs, bug for bug compatibility etc and then convi
86.
by baq 2mo ago
I asked ChatGPT and it told me these aren’t not important /s
87.
by baq 2mo ago
It’s more than you get from free software - you get no proofs, no warranties and any responsibility of its authors are their pure good will. Reminder lean proofs are software!
88.
by baq 2mo ago
As in chess and go and also coding for the past ~year there are two groups of people: the disappointed and the enthusiastic. The disappointed are sad that they lost their advantage and that the craft they honed for years or decades has rapi
89.
by baq 2mo ago
I mean, if you're paying this little, you aren't the customer; the service is either subsidized by investor money (best option), the government (which makes it unfair competition at best ) or profit is being realized elsewhere (w
90.
by baq 2mo ago
So is Fable, I’m asking who is the customer if the thing is free

More ›