Majromax

7 ms·

1.
by Majromax 8d ago
> The cyclone prediction thing is very interesting to me in particular (not quite sure how you go from the ML matrices to "here's a path the cyclone might take") In a high-level view, it's the result of specialized de
2.
by Majromax 8d ago
> It’s crazy we’ll never have forecasts as good as dark sky again. 'Nowcasting' is an area of active research, both with machine learning and with physics-informed or visual flow approaches. Part of the problem from the machine
3.
by Majromax 8d ago
A barometer will give you surface pressure, but that's a field that tends to vary relatively slowly over the surface of the Earth. The calibrated weather stations that exist at every airstrip do a reasonable job of providing these c
4.
by Majromax 13d ago
That line of reasoning has no end. If you use Antigravity on anything other than a Google Chromebook or Pixel, the hardware is a 'product not provided by them'. Is that a TOS violation?
5.
by Majromax 23d ago
If your maximum addressable market is “the whole economy,” as seen in SpaceX filings, then a city-sized call centre (distributed, of course) really is ‘t that much of an ask.
6.
by Majromax 26d ago
> But it’s just as likely to make an output better. No, for any particular output token the model's true logits are definitionally the 'best' that the model can achieve. This is inherently probabilistic. The model's
7.
by Majromax 27d ago
> Hyperscalers don't run computing at some multiple more efficient than on prem. I'd disagree here. I see two avenues for an efficiency multiple, albeit a single-digit multiple: * Client aggregation allows a hyperscaler to ave
8.
by Majromax 28d ago
> As things settle down and commoditize, the value of switching on a dime diminishes as people lock into their favorite models I can imagine just the opposite outcome from the same scenario: as people settle into their favorite but commo
9.
by Majromax 28d ago
You could say the same thing about compilers versus assemblers, high-level languages versus low-level ones, and services and libraries versus monolithic programs. All other things being equal, increasing the speed of some part of the develo
10.
by Majromax 28d ago
> Is the ROI there to pay for the trillions in commitments that have been bet on that ROI? That looks like a clear no at this point. That's only a potential crisis for those who have made concrete investments. On the use side, the
11.
by Majromax 1mo ago
A reasonable guess about the algorithm is 'A Watermark for Large Language Models' ( https://arxiv.org/abs/2301.10226 ). The idea is that each generated token (or bigram) seeds a strong PRNG that splits the voc
12.
by Majromax 1mo ago
> But unless you know the prompt, you don't fully know which choice the model faces. Surely a lot of coding space is wasted compensating for that uncertainty. When you only need to encode one bit, the signal to noise ratio can be ve
13.
by Majromax 1mo ago
> Wouldn't one AI or another detect this deliberate backdoor and report it, as it'll look just like any other security vulnerability, the only difference being the intention? That's precisely the author's point: delib
14.
by Majromax 1mo ago
> In the notebook, I compute a MiniLM embedding of every real Wayfair classification. I compute the embedding of the fake, hypothetical embedding from the LLM. I then dot product the fake embedding into the real ones to find the most sim
15.
by Majromax 1mo ago
> [A]ll of them found some that the others hadn't discovered. Now, correctness issues aren't the same as vulnerabilities, but the same principle about using heuristics to find defects applies. This makes perfect sense, but that
16.
by Majromax 1mo ago
In my view, it's not so much the writing style itself as the lack of 'taste'. Text that clearly seems AI-written has a flat level of exuberance that's just exhausting , kind of like a written version of the 'loudn
17.
by Majromax 2mo ago
> The attacker just needs one exploit chain, whereas the defender needs to block every avenue. Open access to models with no guardrails greatly benefits the attackers more than the defenders. I see it as the opposite, where the attacker
18.
by Majromax 2mo ago
> The saying that stuck with me was "defenders have to be right 100% of the time, while attackers only have to be right once". > You are suggesting this isn't correct? The intuition behind that is applicable only when c
19.
by Majromax 2mo ago
> Dumb question. If "Mythos-class" models are such a problem, then... why not just let it fix everyone's code? In the specific case of cybersecurity, this is a reasonable medium-term outcome. IMO, the cybersecurity risk i
20.
by Majromax 2mo ago
> Anthropic will make the case that their models should be evaluated with the safety layer in front, because that is the only way the model is available whereas open weight models need to pass the same test just on the weights. Worse tha
21.
by Majromax 2mo ago
> I may be crucified for asking this but: is there any proof that slop matters beyond our sensibilities as developers? That's precisely what this benchmark tries to quantify. Since the benchmark incrementally expands the scope of e
22.
by Majromax 2mo ago
> because people have other ideas https://en.wikipedia.org/wiki/Technological_singularity In a weak sense, singularities are common and should be expected every so often. In a mathematical sense, a singularity is w
23.
by Majromax 2mo ago
No, you'd still care. YOLO mode is about instantaneous permissions and access control, inspection of subagent prompts is about retrospective quality control. If the main model is instructing subagents to do a subtly wrong thing, th
24.
by Majromax 3mo ago
If you're deliberately displaying the image in black and white, the colour pattern is interference that should be suppressed. However, this practice was not universal, and archivists have now recreated colour copies ( https:/&#x2F
25.
by Majromax 3mo ago
> I haven't yet had an agent rm -rf files. That happened to me once; I was running one of a few free-tier models in a pi-coding-agent session. The bash tool there is stateless and always begins from the launch directory, but the ag
26.
by Majromax 3mo ago
This blog post is based on a paper ( https://arxiv.org/abs/2602.14740 ). The paper is based on a simulated wargame. The wargame is of the author's own design. The wargame design does not differentiate between ordi
27.
by Majromax 3mo ago
> fronting the inference layer with a caching prompt classifier to determine which model to use, and automatically select the lowest cost model would probably already save alot of money Unfortunately, that doesn't work within a sing
28.
by Majromax 4mo ago
> Seven tokens long input isn't very realistic, is it? The test prompt above was "Why is the sky blue?", so there's the seven tokens. I meant to highlight that because I'd expect processing of a thousand-token i
29.
by Majromax 4mo ago
From the prompt timings above, it seems like 'prompt eval time' is the equivalent to 'processing time for input tokens'. Hyperscalers can perform this evaluation very quickly because evaluation can be significantly paral
30.
by Majromax 4mo ago
> They are all just variations of "insert a canned prompt", varying only along the dimensions of (a) how and where the prompt is installed and from where it is sourced, and (b) which context or contexts the prompt runs in. Ther

More ›