hgoel

6 ms·

61.
by hgoel 1mo ago
Apparently between technology and soul crushing authoritarianism at an unprecedented scale, you would much rather the latter than the former.
62.
by hgoel 1mo ago
Same with NYC, "going to The City" usually just means going to Manhattan.
63.
by hgoel 1mo ago
Near frontier models are currently able to run on a ~150mm^3 computer cluster on 300W, and most of that volume is cooling.
64.
by hgoel 1mo ago
Scientifically useful physics simulations. Every model absolutely sucks at them. Or, as someone else points out in another thread here, academic writing. It's one of the things newer models seem to have actually gotten worse at. Even w
65.
by hgoel 1mo ago
I understand that part, what I don't quite get is why we can't divide the layers such that each node only needs to talk to at most 2 nodes. I assume I'm misunderstanding how work is distributed between them, but haven't
66.
by hgoel 1mo ago
But they also need to have computer skills if they're going to survive in this world... it's too late to teach them once they're old enough to have the safeguards removed.
67.
by hgoel 1mo ago
DGX Sparks have 200Gbps interlinks for clustering. Apparently switches are the only practical way to scale such clusters beyond 2 nodes even though each Spark has 2 such ports.
68.
by hgoel 1mo ago
If we're banning superintelligence, we might as well ban genies from granting wishes and aliens from outside the solar system from establishing residence on Earth.
69.
by hgoel 1mo ago
Keeping younger students away from computers is dumb, but we do need to go back to the basic computer literacy classes of the early/mid 2000s, e.g. typing, basics of desktop operating systems like the filesystem, using word, powerpoint
70.
by hgoel 1mo ago
They gave access, but considering that they wouldn't even sign the "don't ban open weights" letter, it's clear they would prefer to have tight control over who they bless with that access.
71.
by hgoel 1mo ago
FIFA is a deeply corrupt institution (IDK if they count as specifically European, but if they do, it doesn't change the fact that they're corrupt). The US government is also deeply corrupt.
72.
by hgoel 1mo ago
It's tricky to do a proper calculation because everyone obscures what a token or a prompt costs in subscriptions and on top of that we have the complexity of how much different models think. For example, GLM5.3-flash runs slightly slow
73.
by hgoel 1mo ago
And we're used to insulting and making fun of the prominent members of those circles too, see: "Scam Altman", the large variety of jokes about Musk's timelines, comments about Dario waking up in a cold sweat whenever a p
74.
by hgoel 1mo ago
>But "run a web search, write a draft perspective from three points of view, and structure data around it" will make everything but Claude + GPT struggle. Did you mean with some specific harness? GLM5.3-Flash handled a very sim
75.
by hgoel 1mo ago
AI isn't quite good enough yet to pick problems that are an effective use of our limited resources.
76.
by hgoel 1mo ago
It isn't even a matter of having rock solid AI. I wonder how massive of a test suite would be needed to catch regressions...
77.
by hgoel 1mo ago
The core will have to still be a solid battle tested thing. Imagine having to worry about if your vibed browser is quietly dropping some elements. Firefox and Chromium browsers are valuable because they are the targets, if a site doesn&#x
78.
by hgoel 1mo ago
Brings back memories of writing a Chip-8 emulator when bored out of my mind in an introductory programming class in undergrad. It's such an underrated project. Producing something that can actually run some full games in a first/s
79.
by hgoel 2mo ago
I am referring to doing exactly that. Realistic, scientifically useful simulations still require tuning all sorts of parameters based on physical intuition and understanding of the system being simulated. Both Sol and Fable/Opus 5 fail
80.
by hgoel 2mo ago
Scientific physics simulations - even the frontier models just engage in rationalization of obviously unphysical results instead of understanding the system. They have the rote knowledge but fail to apply it unless their hand is held throug
81.
by hgoel 2mo ago
Only reason to spend a bunch of money on hardware to run LLMs locally is if it's a hobby to you to an extent that even renting the GPUs temporarily won't satisfy you.
82.
by hgoel 2mo ago
I think the Western labs are burning through funding and compute to maintain the lead at any cost, efficiency be damned.
83.
by hgoel 2mo ago
The cybersecurity refusals, the biology refusals: https://blog.stephenturner.us/p/benchmarking-ai-biosecurity-... and the sexual activity refusals are all a form of censorship. Their purpose is the same as what the Chi
84.
by hgoel 2mo ago
This argument seems somewhat strange to me, and at odds with the recent model releases. It feels like claiming that because a current generation laptop CPU is slower than a current generation HEDT CPU, we should ignore that the laptop CPU i
85.
by hgoel 2mo ago
Yes, there's that, and also things like shameless empty nesters suggesting that they shouldn't have to pay taxes that support the education system.
86.
by hgoel 2mo ago
Expecting the modern American people to support any welfare system is a fairy tale. A large chunk of them will be too busy worrying about the possibility of even one undeserving person benefiting even if it means thousands starve to death.
87.
by hgoel 2mo ago
I get the impression that they're focusing more on efficiency than raw intelligence. While I assume that all AI labs realize that the AGI "race" is mostly bs, the Chinese labs aren't stuck in a trap where they need to ke
88.
by hgoel 2mo ago
Where does the government enforced censoring of Fable's capabilities lay for you? How would you distinguish if, say, a model were being trained to have a certain bias rather than just refusing to respond?
89.
by hgoel 2mo ago
I believe that at minimum, for usable performance, you need to be able to hold the 125B params + 51B ngrams in some sort of RAM. Ideally VRAM, but the benefit of the MoE design is better performance with unified memory since most of that RA
90.
by hgoel 2mo ago
Do you find subscriptions to be meaningfully faster? I didn't really feel too much of a speed difference compared to Opus.

More ›