oofbey

33 ms·

571.
by oofbey 5y ago
So typically Google. Patting themselves on the back for being smart and proving to themselves they could figure it out. But not actually accomplishing anything useful. Getting it to actually work - that’s too easy and boring for a googler t
572.
by oofbey 5y ago
Anybody using this know if it works well for iPad + pencil for scribbling notes on published research papers?
573.
by oofbey 5y ago
Totally disagree. The encoding scheme could give the model an inductive bias which would make rhyming easier. But the lack of such an inductive bias in no way prevents it from understanding rhymes. This model has 10^11 parameters! The embed
574.
by oofbey 5y ago
Deep learning theory has shown that these huge neural networks can memorize basically any dataset they're given. The useful ones find ways to interpolate everything they've memorized to new contexts.
575.
by oofbey 5y ago
That's true for HTML, but not HTTP. HTML is organic - here's your document, good luck making it work. HTTP is clearly versioned with significantly different changing features at each revision.
576.
by oofbey 5y ago
K-means has a lot going for it, but it's not very good at recovering "natural" clusters. Like if you have a dataset that you know actually is grouped in a certain way, k-means very often will not discover those clusters. Ot
577.
by oofbey 5y ago
Sounds like a pretty good case for enhancing gopher instead of starting over.
578.
by oofbey 5y ago
Sounds like they want to guarantee this never actually catches on. Pretty sure they will succeed.
579.
by oofbey 5y ago
This sounds cool. Current block dedup really seems like a simple (naive) v1 compared to what should be possible.
580.
by oofbey 5y ago
Interesting - I didn't know you couldn't submit GPL-licensed code to the app store. It's surprising to me, but I can imagine a legal reason why it could be difficult for them to support. Still, seems like it should be possi
581.
by oofbey 5y ago
"You're either with us or against us" is a successful rallying cry for extremists, but it's not how most people view the world. I think if you survey FOSS developers, you'll find a LOT of them (including me) use an
582.
by oofbey 5y ago
What units is this? $35 per what? Megawatt-hour?
583.
by oofbey 5y ago
How does building closed source products fight open source? They’re not at odds with each other.
584.
by oofbey 5y ago
Uhh, you know scikit-learn algos are pretty pretty much all ML, right? As is clustering.
585.
by oofbey 5y ago
Google Translate says: Dear Russians, your media is being censored. The Kremlin is lying. Thousands of your soldiers and Ukrainian brothers are dying in Ukraine. Find out the truth on the free web and on the Telegram app. Time to overthrow
586.
by oofbey 5y ago
Oh yeah, I think you're right. If it's all fully-connected (and parts of a transformer are) then more thinner layers use fewer FLOPs than fewer thicker layers, for the same number of parameters. As long as the layers are wide en
587.
by oofbey 5y ago
I hope so, but I'm not convinced. The experimental results don't persuade me that deep transformers are better than the existing extremely wide transformers with tons of parameters. Yes, they show a much better result on the mul
588.
by oofbey 5y ago
Why would evaluation be faster with more narrow layers? If there were fewer tokens it would definitely be faster, because transformers scale by tokens^2, but here "narrow" means number of channels, for presumably the same number
589.
by oofbey 5y ago
> With the enhanced Infineon Doppler radar and the standard algorithm, the module is an ideal solution for individual applications like elderly health care, smart home, and danger alarm. Anybody know what they mean by "the standard
590.
by oofbey 5y ago
Ugh. Yet another visual ML programming tool. They look great for demos. And are a complete disaster when you try to do real work. At any real-world level of complexity, managing boxes and arrows is totally untenable, and you need code.
591.
by oofbey 5y ago
Why is this in their “science” blog?
592.
by oofbey 5y ago
What kind of camera was used to take this?
593.
by oofbey 5y ago
Maybe you don’t remember the software industry in the 90s when Microsoft was at the peak of their power. Nobody had heard of app stores but MS showed the world what a tech bully is. Putting profit over customer happiness, because they coul
594.
by oofbey 5y ago
Security always exists in layers. Policy is a layer - e.g. asking people to do something. Another layer is to have a system to remind/enforce that policy. Writing down passwords is (generally) an obviously stupid thing to do, perhap
595.
by oofbey 5y ago
I would take password-protected keyfiles - I would love it in fact. But AFAIK there's no way for the server to know if the keyfile was password protected or not. Is there? It seems like it wouldn't be reliable - you kinda have
596.
by oofbey 5y ago
Ironically, I think the blog post would have been improved with some explicit self-promotion.
597.
by oofbey 5y ago
The article says to use two-factor-auth (obviously a good idea) but says nothing about HOW you add 2FA to SSH. Does anybody have pointers? I'd love to add 2FA to my bastion hosts, but don't want to put a ton of effort into doing
598.
by oofbey 5y ago
It’s quite strange that the abstract implies Einstein is a founder of QM. More the opposite I think. Einstein remained deeply skeptical of many fundamental aspects of QM - believing in hidden variable theory through his famous statement “
599.
by oofbey 5y ago
I rely heavily on the foot pedal. The single-pedal is too light and really wants to be bolted to something heavy so it doesn't slide around. The triple-food pedal is much nicer, just because it's heavy and stays put. I map all
600.
by oofbey 5y ago
Keras is great for easy problems. But as soon as you want to color outside the lines its opinionated simple model just gets in the way. Every time I've tried to use keras for real work, it was a total pain in the butt. Do your homew

More ›