achierius

7 ms·

1.
by achierius 4d ago
> despite the absurd fantasies of LLM fans Absurd fantasies like solving a Millennium Prize Problem? How is this not prima facie absurd? And if that's come to pass, why should we believe your bar for anything else?
2.
by achierius 4d ago
Do you seriously think that's that people mean when they quote that?
3.
by achierius 4d ago
Interesting! I admit the AI-written text is rough to read, it could have used a pass or two from an actual human. E.g. "Publishing it unresolved rather than tuning until it matches." -- thanks for not lying, I guess? Fun: > In
4.
by achierius 6d ago
Not 100% surprised that this wasn't picked up as a security issue; denial-of-service is bad, but ultimately doesn't give you a direct path to stealing secrets / hijacking identity / etc. It is pretty egregious though, I
5.
by achierius 7d ago
But Tailwind is still being used? It's just being used by AI now.
6.
by achierius 7d ago
If you don't need to 'understand' to do this, then maybe what you call 'understanding' is less important than you think. When people in the real world reach for a tool, what they care about is whether or not it&#x27
7.
by achierius 7d ago
You're probably correct. FWIW most people with math degrees haven't gone on to do math research regardless, so in that regard this isn't a big change.
8.
by achierius 7d ago
There is a land war in Europe for the first time in generations. The only reason you can say that big things aren't happening is because you're lucky enough to live where they're happening.
9.
by achierius 7d ago
But ultimately this is just an engineering problem, no? Yes, if you just hack steering into a model it's going to hurt performance, because doing so takes the model out of the regime it was trained for and validated in. But if that ste
10.
by achierius 8d ago
> We have AGI and the intelligence abundance is going to be amazing for everyone in the future. Why? These 'geniuses in a datacenter' aren't good, they aren't 'aligned', they don't work for you. They&#x
11.
by achierius 8d ago
I mean it's definitely an outrage, but I find it hard to believe that you went from "yay OpenAI" to "literally destroy the company" over... accusations of academic misconduct?
12.
by achierius 8d ago
Have you read the actual statement https://cims.nyu.edu/~tristanb/statement.pdf ? > I should say here why I interpreted their statement the way I did, the in- terpretation I will discuss below. The route to the Clay
13.
by achierius 8d ago
While I'm generally pretty negative on claims that the labs are 'scamming' the public with misrepresentations of model capabilities, it's hard to see how this wouldn't qualify. - the OpenAI researchers claimed that
14.
by achierius 9d ago
Would you change your mind on this if they became profitable? What would convince you that the labs are real threats worth organizing against? Or are you just dedicated to boosting AI until your dying day?
15.
by achierius 9d ago
Have you read a single paper in ai safety?
16.
by achierius 10d ago
Unfortunately I can't comment much on Apple's situation since I work there :) Regardless what I mean to convey is that ARM would prefer such deviations to be rare, and especially for them to not be visible from userspace. Deviatio
17.
by achierius 10d ago
(Total guess) this feels like an attempt at restricting what parts of the hardware pipeline need to know what; if prefetches need to key off of both the device bit and the nx bit, then both of those need to be piped into the frontend, where
18.
by achierius 10d ago
FWIW if you look at WebKit/JavaScriptCore, the trend has been towards less-frequent code modification, e.g. inline caches are mostly no longer repatched inline. Whole-function reoptimization is still worth the overhead of sys_icache_in
19.
by achierius 10d ago
> while ARM does provide a reference implementation of the architecture, vendors are free to customize it at will, or even roll their own implementations A nitpick, but this is only true for some vendors, depending on their license, and
20.
by achierius 12d ago
Is that just the MiniBrowser?! I have to say I can't imagine daily driving that if so. Why WebKit over a Chromium-based (and more featureful) browser?
21.
by achierius 12d ago
Is there anything that could convince you that actually-bad things are actually-happening? How can you possibly think that these people are somehow working for OpenAI? Is it impossible for you to imagine that there exist people who actually
22.
by achierius 14d ago
I put that poorly: read as 110% and 200%
23.
by achierius 14d ago
No, they're not getting close. Top performing Wasm runtimes (like those in V8 and JSC) are generally between 10% (for pure math workloads) and 2x (conservatively, for ef allocation heavy ones) the speed of equivalent native implementat
24.
by achierius 15d ago
> - environmental like emissions and water usage Those are two very different things, and the former should be a serious concern. If AI does indeed become a double-digit percentage of electricity usage as the AI labs themselves predict,
25.
by achierius 21d ago
No, not necessarily. Sometimes visas are rejected. Unless you're talking about giving everyone who gets a visa a green card?
26.
by achierius 24d ago
AI account?
27.
by achierius 1mo ago
How is that a moat, if anyone else can do the same thing? And how would you get this all into Clang? Nobody wants to use your custom compiler. It's certainly not going to be as fast as Clang!
28.
by achierius 1mo ago
One unintuitive thing I noticed about these curves, at least from the demo on the page, is that they have both areas where their behavior quickly 'snaps' from one configuration to another, and areas where they saturate such that c
29.
by achierius 1mo ago
It'd be really interesting to see whether the winning (losing?) instructions/strategies would be different on other architectures. At least right now the top spot (`fxrstor64` on MMIO, starve PCIe) seems relatively architecture-in
30.
by achierius 1mo ago
Why are you opposed to a general pause?

More ›