sdenton4

13 ms·

211.
by sdenton4 1y ago
Ha, I often speak of doing the first 90% of the work, and then moving on to the following 90% of the work...
212.
by sdenton4 1y ago
Yah, aardwolf (the one o played in ancient times) apparently has 35,000 rooms. Collaborative building over decades adds up!
213.
by sdenton4 1y ago
Yeah, it's tricky to talk about in the span of a comment. I work on Things Involving Animals - animals provide an excellent counter-current to discussion around AGI, in numerous ways. One interesting parallel was the gradual redefiniti
214.
by sdenton4 1y ago
I think it's helpful to put on our statistics hats when looking at data like this... We have some observed values and a number of available covariates, which, perhaps, help explain the observed variability. Some legitimate sources of v
215.
by sdenton4 1y ago
The overwhelming majority of animal species never developed (what we would consider) language processing capabilities. So agi doesn't seem like something that evolution is particularly good at producing; more an emergent trait, eventua
216.
by sdenton4 1y ago
Or perhaps the universe is overflowing with minds we haven't found yet!
217.
by sdenton4 1y ago
Already there's a transition from bass (which you can feel in your body) to hearing music through tiny earbuds...
218.
by sdenton4 1y ago
Yeaaaaaah, I kinda doubt there's much coming from evolutionary biases. If it's a matter of clever initialization bias, it's gotta be pretty simple to survive the replication via DNA and procedural generative process in the me
219.
by sdenton4 1y ago
Option 3, nothing changes: people (and institutions) still have income they need to put /somewhere/, and continue putting it in 401ks and index funds because that's the system. Stonks continue to go up, completely unmoored fr
220.
by sdenton4 1y ago
A good analogy might be dentists - are there other spices of the medical profession that could be turned into more focused specialties with less total schooling?
221.
by sdenton4 1y ago
"Guys, we're crushing it on innovation! Even the Economist agrees!"
222.
by sdenton4 1y ago
Indeed - human judges suck on average. And you can prompt an llm judge to look for particular kinds of problems, then throw the ensemble of judges at an output to nitpick. (Essentially, bake in a diversity of biases through a collection of
223.
by sdenton4 1y ago
Hire an expert and an apprentice to work with them. Now you've got 10k workers.
224.
by sdenton4 1y ago
Yes, and also: Is it good for a factory for when the average worker has three jobs and only gets six hours of sleep? Sounds like a recipe for recalls...
225.
by sdenton4 1y ago
Yep, makes sense - conversion to text and then aligning the text with the audio is a very reasonable way to handle large volumes of speech data. For bioacoustics, we tend to have a loooooot of variation for which there is no real notation,
226.
by sdenton4 1y ago
Solar panel cost per watt has been dropping exponentially for decades as well... Partly these are matters of economies of scale - reduction in production costs at scale - and partly it's a matter of increasing human attention leading t
227.
by sdenton4 1y ago
Curiously, humans don't seem to require reading the entire internet in order to perform at human level on a wide variety of tasks... Nature suggests that there's a lot of headroom in algorithms for learning on existing sources. In
228.
by sdenton4 1y ago
Here you go: https://github.com/google-research/perch-hoplite It's geared towards bioacoustics in particular. It's pretty easy to provide a wrapper for any given model though. Feel free to send a ping if you
229.
by sdenton4 1y ago
To be clear, I'm not the author of the post. But I do maintain a library for folks working with large audio datasets, built on a combination of SQLite and usearch. :)
230.
by sdenton4 1y ago
Bridge builders mostly don't have to design for adversarial attacks. And the ones who do focus on portability and speed of redeployment, rather than armor - it's cheaper and faster to throw down another temporary bridge than to bu
231.
by sdenton4 1y ago
They say int he post that they're doing optimized brute-force search, which honestly makes a lot of sense for the local-scaled context. Vector databases are often over-optimized for getting results into papers (where the be-all-end-all
232.
by sdenton4 1y ago
There is/was an AI Dungeon app running in the early/pre-ChatGPT days using GPT 3-ish, I think? Long term context was a real problem - story arcs were very.... drifty. A more modern agentic approach might help with this, doing mult
233.
by sdenton4 1y ago
You should talk to your representatives, then.
234.
by sdenton4 1y ago
I avoid buying 2d printers because they are so maintenance heavy... I use a 3D printer in a shared makerspace, where whoever has the most avialability takes on maintenance issues when they arise. The Prusa mk4's we use are extremely re
235.
by sdenton4 1y ago
Our makerspace at the office has a few prusa mk4's, and they're really great machines.
236.
by sdenton4 1y ago
The spare autoencoder work is /exactly/ premised on the kind of near-orthogonality that this article talks about. It's called the 'superposition hypothesis' originally: https://transformer-circuits.pub&#x
237.
by sdenton4 1y ago
IME: most of the reviewers in the big ML conferences are second-year phd students sent into the breach against the overwhelming tide of 10k submissions... Their review comments are often somewhere between useless and actively promoting scie
238.
by sdenton4 1y ago
Spherical codes are kinda of obscure: I haven't heard of them before, and Wikipedia seems to have barely heard of them. And most of the Google results seem to be about playing golf in small dimensions (ie, how many can we optimally pac
239.
by sdenton4 1y ago
The "willing to maintain your work" forever is the crux here - the open-source contributor is doing some amount of work to bring their contribution in line with the expectations of the repository, but it's up to the repositor
240.
by sdenton4 1y ago
You are free to fork and make whatever mess of features you wish for!

More ›