didibus

5 ms·

31.
by didibus 2mo ago
Impressive, you can really do a lot on a single node when it comes to big-data queries nowadays, I agree too many jump straight to a Spark cluster or something similar when you can just write a small script on a single node.
32.
by didibus 2mo ago
But why are you starting to leave your house? Likely because you're ready to leave no? To me that's where it was implying that models are ready to leave that benchmark behind, and starting the process is coming up with a new bench
33.
by didibus 2mo ago
You're assuming that's due to more days off? I have my doubts. I think the US could offer more paid days off, and it would retain its higher wages and lower taxation rates.
34.
by didibus 2mo ago
Might be my reading comprehension, but that reads to me like it's implying we're leaving the territory because it's solved and now need something harder.
35.
by didibus 2mo ago
The article makes it sound like Pelicans riding a bike are solved, and we can move on to something else. Nothing to do with excitement or wonder of the tech, it just claims one thing and I feel I've not seen that be the case so the cla
36.
by didibus 2mo ago
This is pretty close to "solved" I'd say. But then, it seems all the other models still need to catch up, and there's likely a little bit of room for improvement on Fable Max as well.
37.
by didibus 2mo ago
Most of the models are multi-modal and trained on images no? That's what they claim at least.
38.
by didibus 2mo ago
Did I miss when they managed to draw a pelican? Cause all the ones I've seen are wrong in some way.
39.
by didibus 2mo ago
I don't think it's about being social/outgoing or not. There's clearly an anxiety we have, we avoid confrontation, or awkward dynamics. For example, you like things done a certain way, that's irrelevant, its that yo
40.
by didibus 2mo ago
Then call it your hobby, right?
41.
by didibus 2mo ago
You work more for your employer? At no additional cost to them?
42.
by didibus 2mo ago
> predict what would come next in this text sampled from distribution X It tries to learn the distribution of "useful" results either through verified rewards or human feedback. Then it encodes that in the network. When you run
43.
by didibus 2mo ago
I mean, they went and measured it, you're just "feeling" it?
44.
by didibus 2mo ago
> AI is making me work more (good thing) How is that good?
45.
by didibus 2mo ago
Ya, I feel 10% to 50% increase seems more realistic, when I see 100% to 900% increase claims it seems a bit extrapolated, at least I don't feel I'm seeing it at my work. And then it depends if we imply a given feature/fix&#x2
46.
by didibus 2mo ago
> We may be atypical, but my partner and I don't want to hire out a cleaning service and deal with the whole human component I kind of understand where the feeling comes from, but this also saddens me. What happened to our generatio
47.
by didibus 2mo ago
> which is nonsense: the gate simply, directly, selects the experts. There's nothing probabilistic about it. Isn't G a learned probability?
48.
by didibus 2mo ago
> the model's output is the result of an optimisation process that rewarded behaviour that was useful, and that's fundamentally different I'm not understanding, can you explain this more? How does it become more than a nex
49.
by didibus 2mo ago
The $20k gig worker, the $100k salaried worker, and the billionaire owner work roughly equally hard. Productivity gains mostly come from better tools, techniques, and leverage, not people working proportionally harder. The question is where
50.
by didibus 2mo ago
I think what gets lost when we value productivity above everything is community and humanity. At the end of the day, I just want to be around people who care, having a good time together. Nothing seems worth doing if it doesn't eventua
51.
by didibus 2mo ago
I feel some systems already operate like this, but typically it was done for fraud and abuse, now it will be done for permissions as well.
52.
by didibus 3mo ago
Interestingly, if you filter by the coding index, Sol xhigh is the best one, and only Opus5 max is better than Sol max and Sol high.
53.
by didibus 3mo ago
What's interesting is this: The top AI models by Intelligence Index are: 1. Claude Opus 5 (Adaptive Reasoning, Max Effort) (61), 2. Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) (60), 3. Claude Fable 5 (Adaptive Reasoning, Max Effor
54.
by didibus 3mo ago
Exactly, that's why this story is so crazy no? It's saying the model was so good at hacking, it managed to bypass all of that, and also found a way to get into Hugging Face. I can understand skepticism, did they have poor security
55.
by didibus 3mo ago
The CFO or wtv that tweeted, their argument was, if you allow cheap free competitive models, you will stop the influx of capital in frontier development of even more powerful LLMs, thus capping how far the tech could reach. Obviously, they&
56.
by didibus 3mo ago
I agree, but I will say, I think both Mythos and these OpenAI model find exploits by examining and trying things against the running system, not from looking at the code. I think you'd have to do the same to catch the real vulnerabilit
57.
by didibus 3mo ago
Fair on the terminology angle, that said in this case, they had proper guardrails no? They were running in a sandbox, but it was able to find an exploit out of the guardrails.
58.
by didibus 3mo ago
Classifiers and such are guard-rails, alignment to me and I assume most people, is about the model training, and it's tendency to respond, agree/disagree, push-back or not, be willing to cheat or even deceive the prompter, etc.
59.
by didibus 3mo ago
China is working on the whole supply chain though and they're willing to compete on razor thin margins. Just look at EVs. They build great cars but the competition is so aggressive that investing in any one Chinese EV company isn'
60.
by didibus 3mo ago
Ya, I think that was a bit misspoken. They mostly meant direct C and C++ interop. It feels much less like a traditional FFI and more like using C and C++ libraries from C++ itself.

More ›