mrtesthah

8 ms·

91.
by mrtesthah 5mo ago
https://aeon.co/essays/what-if-jobs-are-not-the-solution-but...
92.
by mrtesthah 5mo ago
I would say that if this sounds untenable to you, then you may want to consider that the way we architect software has itself been untenable for a while. What Mythos can accomplish today in public, an APT unit can already accomplish in secr
93.
by mrtesthah 5mo ago
It was a security contractor who leaked the creds. So this is already the increased privatization end-game.
94.
by mrtesthah 5mo ago
All the frontier LLMs are multi-modal.
95.
by mrtesthah 5mo ago
Gemini 3.5 is imminent. https://www.voxelbench.ai/leaderboard
96.
by mrtesthah 5mo ago
I love all the customization features but I cannot use them with Privacy Pass sadly. Anonymity is more important to me.
97.
by mrtesthah 5mo ago
Property destruction is not the same crime as battery/assault/etc. Let’s not call breaking a camera “violence”.
98.
by mrtesthah 5mo ago
> but it’s such a complex theory What's the theory now? I didn't propose any specific theory -- just noted a mechanism of influence.
99.
by mrtesthah 5mo ago
It makes sense for chocolate given that cocoa flavanols are prebiotic fiber for GABA-secreting bacteria which of course affects the parasympathetic nervous system.
100.
by mrtesthah 5mo ago
What we do know is that there was no evidence a few months prior and that Trump explicitly said that he was ignoring the clear lack of evidence in making his decision to attack.
101.
by mrtesthah 5mo ago
> Coding, one of the most popular uses cases today, would not be great if it say only understood java to a version from years ago etc. This LLM trained only and entirely on pre-1930s texts was able to code Python programs when given onl
102.
by mrtesthah 5mo ago
And? What actually changed during those few months?
103.
by mrtesthah 5mo ago
The US did in fact bomb the Iranian nuclear facilities in June 2025.
104.
by mrtesthah 5mo ago
Nope, that poster is correct. When we showed up with the anti-abortion message at first we failed. The evangelicals who had attended our seminars by tens of thousands when we were launching the first series just were not interested in the
105.
by mrtesthah 5mo ago
This is probably more of an attempt to make computing relevant to that “intersectional” subset of people who only consider a topic to be relevant if it relates to colonialism in some way.
106.
by mrtesthah 5mo ago
That’s not what US intelligence says. The IC continues to assess that Iran is not building a nuclear weapon and Supreme Leader Khamenei has not authorized the nuclear weapons program that he suspended in 2003. https://www.factc
107.
by mrtesthah 6mo ago
And that’s why you get one with a carbon fiber belt instead! https://www.montaguebikes.com/product/allston/ or get a kit for your Brompton: https://www.kinetics-online.co.uk/folding-bikes/brom
108.
by mrtesthah 6mo ago
> "is the RLHF judge happy with the answer." Reinforcement Learning with Verifiable Rewards (RLVR) to improve math and coding success rates seems like an exception.
109.
by mrtesthah 6mo ago
What is intrinsic to an LLM or its training that would prevent it from following the directive that it should not try to convince you of something?
110.
by mrtesthah 6mo ago
An LLM that can't follow instructions wouldn't be able to write code anyway.
111.
by mrtesthah 6mo ago
No, see, we have to leave writing code to fully identity-verified individuals working on behalf of only the largest institutions now because what if they decided to write malware?
112.
by mrtesthah 6mo ago
The core issue is that the LLM is using rhetoric to try to convince or persuade you. That's what you need to tell it not to do.
113.
by mrtesthah 6mo ago
Ostensibly this makes usage meter rate changes more transparent?
114.
by mrtesthah 6mo ago
gpt-5.4 is unmatched. Claude is possibly better in web UI tasks, but not much else.
115.
by mrtesthah 6mo ago
That’s going to be inefficient when most of the generations have broken syntax and can’t even parse.
116.
by mrtesthah 7mo ago
Someone make a TLA+ model for this bad boy
117.
by mrtesthah 7mo ago
That’s the kind of thing Apple’s software explicitly tries to support. There’s a reason this stuff works better with native apps.
118.
by mrtesthah 7mo ago
Maybe so, but GPT 5.4 is absolutely pulling ahead. You can see the differences visually on https://minebench.ai/ .
119.
by mrtesthah 7mo ago
How about this guy? https://en.wikipedia.org/wiki/Srinivasa_Ramanujan
120.
by mrtesthah 7mo ago
Most of this intransigence is due to corruption, starting with a billionaire cabal’s influence over the supreme court. Citizens United has all but granted the power to decide elections to super PACs, and therefore to the billionaire donor c

More ›