davrosthedalek

5 ms·

1.
by davrosthedalek 15d ago
The price is that you essentially glue CPU and GPU together, which limits total compute, from a size and thermal perspective. This is really not a limit because of unified memory -- in principle, PCIe GPUs could read/write main memory
2.
by davrosthedalek 1mo ago
regarding my brain, my mother might disagree on the laundry part.
3.
by davrosthedalek 1mo ago
And I would say it's actually a more secure future demand channel than the hyperscalers. Openai and co can build their own chips, so they won't buy from NVidia. But they probably won't sell them to somebody else, so everyone
4.
by davrosthedalek 1mo ago
Because it breaks the TOS.
5.
by davrosthedalek 2mo ago
Because of regulation, not because they are intrinsically hard to get.
6.
by davrosthedalek 2mo ago
It is actually an interesting conundrum. Is a non-well-aligned frontier level AI a problem? I think it is likely that it is, or at least has a high likelihood to be in the future. Two scenarios for this: Misused by some bad guys. Or the ter
7.
by davrosthedalek 3mo ago
Well, I think it's fair to assume that a) They didn't upload everything before they realized it would work. b) They want to mention this as an advantage for future analyses c) Even if you assume that the attack exfiltrated everyth
8.
by davrosthedalek 3mo ago
To be fair, there are really three threats: a) People do bad stuff because LLM told them a wrong thing. Example: AI told me I should treat my heart attack by putting a fork in the outlet. Maybe similar to seeking medical advice on reddit?
9.
by davrosthedalek 3mo ago
If there is an electrical connection between these downstream boxes and the inference servers beyond the power connection, it does stretch the definition of air gap.
10.
by davrosthedalek 3mo ago
I am somewhat more worried about a Darkstar AI.
11.
by davrosthedalek 3mo ago
Too many spaces around the --. That's not load-bearing.
12.
by davrosthedalek 3mo ago
The grant size for phase 1 is 750k / 9 month. That is actually quite substantial. [Despite that, I generally think more money should go to science, all around. But I have COI here.]
13.
by davrosthedalek 3mo ago
Lightspeed in air is close to c_vacuum. Light speed in fiber is roughly 2/3 c_vacuum. So for transatlantic it might be faster.
14.
by davrosthedalek 3mo ago
True, but I think the US requirements are indeed different. Ukraine must repel invading forces in their own country -> lowish range, mass produced, not necessarily precision strike. US want to project power far away from its shores ->
15.
by davrosthedalek 3mo ago
Das Auge ist (man) mit.
16.
by davrosthedalek 3mo ago
10 Gig ethernet is 10GBps usable rate (before packet overhead). The line rates are higher to accommodate this. For 10GBase-R, it's typically 10.3125 GBps, with a 64/66 encoding. For 10GBase-T, it's 4 lanes with PAM-16 at 800
17.
by davrosthedalek 4mo ago
The question is: If you have enough full body scans of many healthy people, and the statistical tools to model it (beyond "this range is OK"), whether this would reduce these false alarms to an acceptable level. The real crux of i
18.
by davrosthedalek 4mo ago
Just for illustration: Gravitational wave detection is on the femtometer scale. The proton is about that size. We can measure these things, but the machines are, let's say, "big".
19.
by davrosthedalek 4mo ago
This is somewhat speculative, but as I see it, there are two ways to retain excellent people: a) You pay them handsomely b) You do shit they like, they way the like. Sometimes it overlaps, of course. But this is essentially the reason why p
20.
by davrosthedalek 4mo ago
Right, that's now /usr/bin/true !
21.
by davrosthedalek 4mo ago
Why it is probably and regrettably true that few people in camps will change their mind, data analysis can help people who haven't been captured yet to either stay away from the camps or at least fall into the "more correct"
22.
by davrosthedalek 4mo ago
First rsync and now less? What comes next, cat?
23.
by davrosthedalek 4mo ago
But: "After posting this on Hacker News and recieving [sic] almost no substantive input, discussion, or response on the actual content of the article, I decided to rewrite all of the prose in my own voice. If anyone complains about my
24.
by davrosthedalek 4mo ago
Haven't looked at the code, but the allocated memory could be larger than necessary to make "off-by-one" or "off-by-a-few" errors less deadly. Then zeroing it out makes it even less so. Defense in depth. Or it'
25.
by davrosthedalek 4mo ago
Many statistics were presented. In the view of the author (and I think he is correct), none of them show evidence for an increased bug rate from Claude. That is absence of evidence (...for the increased bug rate). The two examples you bring
26.
by davrosthedalek 4mo ago
Yes, I do for example. And the author discussed the use of AI pretty exhaustively in point 0 of the post.
27.
by davrosthedalek 4mo ago
No. It's a description of the result of the maybe underpowered study. the underpowered study did not find evidence. Evidence is absent. Because it is underpowered, it's not evidence that the effect is absent. The claim is not &quo
28.
by davrosthedalek 4mo ago
He can't make the evidence of absence claim, but he can absolutely make the absence of evidence claim.
29.
by davrosthedalek 4mo ago
The point is that you should look only at the first smlen entries, which would be 0 for this case.
30.
by davrosthedalek 4mo ago
That's how it used to be. At least at my alma mater, it changed a lot when they switched from Diploma to Master/Bachelor

More ›