loudmax

34 ms·

391.
by loudmax 3y ago
Absolutely this. Google could develop an AI that's superior to GPT-4. They may have already. Under their current leadership, it wouldn't make a difference anyway as they wouldn't know what to do with it. Google is on a fa
392.
by loudmax 3y ago
I second this recommendation to start with llama.cpp. It can run on a regular laptop and it gives a sense of what's possible. If you want access to a serious GPU or TPU, then the sensible solution is to rent one in the cloud. If you
393.
by loudmax 3y ago
> So eventually you get agents that have self-preservation and reproductive instincts. I'm not sure that's a given. Artificial Intelligence as it currently exists, doesn't have any volition. AI doesn't have desire o
394.
by loudmax 3y ago
> It provided Google with essentially zero value. Or rather, it provided enormous value. The failure was for Google to actually capture more than a tiny fraction of that value. No amount of engineering brilliance is going to save Google
395.
by loudmax 3y ago
The appeal of a federated platform is that it's decentralized without being siloed. If Bluesky decides it's in their interest, they could probably figure out some way to federate with Mastodon.
396.
by loudmax 3y ago
Sorry if this is off-topic, but those are some really good book recommendations in the demonstration image! If those are coming from Vicuna, that speaks well of it.
397.
by loudmax 3y ago
Cameras designed for this purpose can mitigate the issue, but "solving" it is not easy. Cameras can digitally sign photos to authenticate provenance. You can add timestamps and GPS coordinates to the signature, though you may be
398.
by loudmax 3y ago
Playing around with Stable Diffusion yields a lot of images that are kind of dream-like. The pictures can be fascinating, but a lot of the details are weird or inconsistent, or somehow "off". I suspect this isn't a coincide
399.
by loudmax 3y ago
I feel the same way about not understanding how LLMs work. I've found Hugginface's tutorials a good place to start: https://huggingface.co/course/
400.
by loudmax 3y ago
I think Amara's law applies: "We overestimate the impact of technology in the short-term and underestimate the effect in the long run." The development of effective LLMs really feels like a tectonic shift. We're still i
401.
by loudmax 3y ago
Jeff Geerling is working within the YouTube algorithm and he does post some clickbaity thumbnails. Don't let those thumbnails mislead you. geerlingguy is an active member of the open source community and he produces some absolute top
402.
by loudmax 3y ago
I hope it doesn't hurt any open source projects, but if there's there's one project that has bloat to trim it's GNOME. It's good that there's experimentation happening with Linux desktops, but that experimenta
403.
by loudmax 3y ago
To confuse matters, many of these attacks aren't so much against America, as against the concept of liberal democracy. To uphold liberal democratic values is to support the free speech that makes these attacks possible. The old 20th c
404.
by loudmax 4y ago
Google has (or until very recently had) some of the best researchers in the industry. Google's problem isn't developing new stuff, it's turning stuff they come up with into a viable product and then developing a market aroun
405.
by loudmax 4y ago
You can absolutely build these in C:S and I've seen them as downloadable assets.
406.
by loudmax 4y ago
I have the same suspicion regarding Google Podcasts. The UI is stable and it doesn't seem to try very hard to promote ads or podcasts I'm not already subscribed to. I do worry that someone in Google management will realize that
407.
by loudmax 4y ago
Not assembly. Machine code. The ones and zeroes. Abstractions like assembly code or C or Python are for the benefit of us humans. We can't reason about even fairly trivial x86 binaries because our minds aren't designed for that
408.
by loudmax 4y ago
That could well be. But also check back in a few years when the AI is chasing down subtle bugs introduced by humans.
409.
by loudmax 4y ago
I think the next phase will be to train GPT-type AIs to forego the compiler entirely and just write binary rather than C or code that's designed to be understood by humans. You should be able to point the AI to a binary executable fil
410.
by loudmax 4y ago
A particularly thorny question if data in the training set is copyrighted. Which I presume much of it is, since I was playing with alpaca.cpp and it answered all kinds of pop culture questions I asked it (to varying degrees of accuracy).
411.
by loudmax 4y ago
It's frankly amazing how much information is summarized into those weights. You do need to take whatever it says with a very healthy pinch of salt. It doesn't know everything and it gives incorrect answers with an air of confide
412.
by loudmax 4y ago
I'm relying on a functional democratic government still existing when I retire. It doesn't seem like a high bar, though admittedly less of a certainty than we would have imagined a decade ago.
413.
by loudmax 4y ago
It depends on what you're looking for in a Sim game. There's no Ender Dragon to kill in City:Skylines, so the objective of the game is whatever you make of it. At heart C:S is really a traffic management game. Cash flow usually
414.
by loudmax 4y ago
You won't get high performance out of Python directly, but there are a lot of Python libraries that use C or a powerful low level language underneath. The heavy lifting in so much of machine learning is CUDA, but most people involved
415.
by loudmax 4y ago
That is my outsider impression of what's going on at Google. The ChatGPT interface that Microsoft is rolling out absolutely should have come from Google first. I'm sure they actually have the technology, and they've probabl
416.
by loudmax 4y ago
Rich Hickey's Simple Made Easy presentation is a fantastic introduction to this philosophy: https://www.infoq.com/presentations/Simple-Made-Easy/ Simple isn't the same as easy, and it isn't always o
417.
by loudmax 4y ago
ThinkPad was originally an IBM line of laptops, which they later sold to Lenovo when they got out of the PC business. ThinkPads had a reputation for being durable, but also somewhat repairable, so something of a DIY laptop community develop
418.
by loudmax 4y ago
None of this is strictly incorrect, but it's a mistake to conclude that since all sides are biased the truth is unknowable and you might as well throw up your hands and retreat entirely. In fact, the goal of disinformation isn't
419.
by loudmax 4y ago
Because the demand hasn't been there. If more people learn the benefits of these sensors and decide they want more of them they'll pay more for models that include them and manufacturers will respond to demand. If people don&#x2
420.
by loudmax 4y ago
Thinking about CO2 levels isn't common sense until after you've been keeping an eye on one of these gadgets. After sitting in a meeting room for an hour, people may find it feels warm or stuffy or notice a slight headache, but th

More ›