fc417fc802

10 ms·

1.
by fc417fc802 5d ago
True, but as far as that goes the question was a pointless one. Whereas the one that was answered has the potential to lead to productive conversation.
2.
by fc417fc802 5d ago
Depends on the implied meaning. What technical reason? There isn't one, obviously. But what about the practical reason for the behavior we observe? That's explained by the name and marketing. And if your goal is to understand &quo
3.
by fc417fc802 5d ago
The vector that needs to be checked is length n while computing that vector requires n^2 operations. I haven't benchmarked it but I expect the performance overhead to be a rounding error.
4.
by fc417fc802 6d ago
Agreed except for that last bit. Even if the session drags on endlessly the AI keeps using the same repetitive speech patterns (at least IME). Not only that but they stand out to me in that they don't match the nature of the conversati
5.
by fc417fc802 7d ago
Are you sure you don't just have an axe to grind? Because that is an exceedingly uncommon edge case. Typically you only have a few threads and you aren't inundating the allocator with requests thus it is unlikely to make any pract
6.
by fc417fc802 7d ago
Yes, and I believe my point still stands. We thoroughly understand the principle by which a loaded die can be intentionally biased despite not being able to predict the outcome of any given throw due to the system in question being a chaoti
7.
by fc417fc802 7d ago
That's certainly interesting but I don't see how it would change my answer to you. Your question was why projects don't switch. My answer was because doing so seems likely to be a wash at absolute best. Giving it some more th
8.
by fc417fc802 7d ago
But importantly for dice we do understand the overarching principles that give rise to this. And dice don't output coherent sentences. Meanwhile in LLM land the analogous "roll of the dice" can result in a coherent response
9.
by fc417fc802 7d ago
Me: I don't expect to be stabbed. You: But you only might be stabbed. It isn't required to happen only permitted .
10.
by fc417fc802 7d ago
No, they didn't. UB is a cop out and inserting yield is just plain bad. Locking up one or more threads in an implementation defined manner would be the outcome of least surprise (I already know it's going to lock up at least the
11.
by fc417fc802 7d ago
Yes, it was a horrible situation that has been replaced by an only slightly less horrible situation.
12.
by fc417fc802 7d ago
In other words, it's a sane default that you can opt out of on a case by case basis which is the way it should have been all along.
13.
by fc417fc802 7d ago
It's only an incentive to start caring if it's cheaper than circumventing the law. In other words it won't work unless the aforementioned liability loophole is closed. To rephrase the comment you replied to, if being a cowboy
14.
by fc417fc802 8d ago
> ad dollars will be sucked out of the rotting Facebook network Doesn't seem likely to me. People scroll a timeline. You aren't going to replace that with an AI agent so the eyeballs will still be there.
15.
by fc417fc802 8d ago
Hey claude, compare this security review to the current codebase and patch up anything that needs it.
16.
by fc417fc802 8d ago
What's dumb about it? I'd suggest that it's slightly abstract and argue that there's nothing wrong with that. It fits perfectly and it's readily understandable. For example a VPN connection lends itself to being des
17.
by fc417fc802 8d ago
Setting aside whether or not you can pull off a lock free approach here we can be certain of a couple things. There will be at least some overhead that must be paid somewhere even if that's on a separate management thread. And there
18.
by fc417fc802 8d ago
I'm never sure if I should be upset or happy when I've been debugging a problem for long enough that I finally decide to switch something out in order to improve visibility and that immediately solves the problem for entirely unex
19.
by fc417fc802 8d ago
7500 surely? But yeah at this point I feel like the diameter of a silicon atom would make for a better unit of measure.
20.
by fc417fc802 8d ago
Aren't such attractors also problems during training? Presumably alignment constraints would need to apply to continuous learning as well.
21.
by fc417fc802 8d ago
> Seems like AI could help massively there, by removing a huge bottleneck around technical elaboration and application seeking. Ah yes, lets take the existing system that's already barely holding up and flood it with slop. What coul
22.
by fc417fc802 8d ago
That's not a bad idea and I believe it is how any modern not-a-website password based authentication works. Agreed that it does not address the attack vector of someone sniffing your password for site A and then logging into site A wit
23.
by fc417fc802 9d ago
Right but regardless, it's a buggy optimization. The calculations are all fully deterministic when done "properly" and fully consistently but we almost never bother with that because it slows things down and the errors don&#x
24.
by fc417fc802 9d ago
> it won't be until a bunch of people die In the context of rogue misaligned AI won't it be far too late to recover by then? In other words isn't that more or less a doomsday prophecy?
25.
by fc417fc802 9d ago
If this is what misalignment turns out to be I ... might be on board with it? At any rate it's nowhere near as concerning as what I had been expecting.
26.
by fc417fc802 9d ago
I aspire to be at least as useful as bogosort.
27.
by fc417fc802 9d ago
Can't the same also be said of much of the x86 vocabulary at this point? I appreciate that we can upload SPIR-V directly. The API still feels overly obtuse but it's not so bad. SYCL gets close but is language specific.
28.
by fc417fc802 9d ago
I don't see the issue. Attacking a physical lock as you suggest is analogous here to breaking the cryptography in some way. So the NSA busts out its sooper sekrit quantum computer ... The inability to betray is a sharp observation but
29.
by fc417fc802 9d ago
> Aren't there laws regulating bioethics that forbid such questionable practices? Not generally, no. Most of it is by convention coupled with the fact that no reputable journal will publish your work these days if you don't go
30.
by fc417fc802 9d ago
That's a nice way of saying that it's a dependency clusterfuck. I've never understood why we can't just expose the GPU ISA directly the way the CPU does. It's all getting compiled down at the end of the day so someo

More ›