dadoomer

I'm also interested in how they figured out how to get the CMA-ES baseline to be this fast (less than 35 samples in Figure 5).

Maybe they are warm-starting CMA-ES with a pre-trained controller (although at a glance it doesn't seem they do). Haven't read the paper yet though.

> neovim + claude + tmux is the goated combo

It absolutely is. I don't really understand why people still confine themselves to a full IDE when doing "agentic programming".

Especially with neovim with something like fzf or ctrl-p, which lets just quickly navigate codebases. This is about as far as you need to go. So to me, being terminal-first and being able to do multiple things at once makes more sense (hence tmux).

Isn't all the "knowledge" just text files? I've transitioned between services easily by simply copying the text files.

Progress towards what? Do you think "progress" as conceived by sillicon valley billionaires aligns with the type of life you would like?

There's a non-trivial chance "working hard" as defined by the modern ethos is doing more damage than good.

Why would self-hosting for privacy reasons be unethical just because the query would be subject to subpoena in principle?

I've used Kdenlive for years. I'm someone who only needs video editing every once in a while, but even then I definitely recommend learning it.

What's the playlist? Curious as a bossa enjoyer.

This has great UX, congrats!

There's quite a few jazz things that are missing. E.g., altered and bebop scales, 7#9 and 7b9 chords. Plus, for any given chord, there are a lot of voicings (drop 2, rootless, etc.).

This suite could be a really cool resource for new/intermediate jazz guitarists if you add some of the jazz stuff in. Any decent jazz guitar book could give you the most used scales and chord voicings.

Since the UX is already quite good, maybe if you open source it and if these things are easy to add, you could rely on the community to do so to some extent.

> what's wrong with a usb-c to 3.5mm adapter

In my experience the connection is much easier to accidentally break through movement (e.g., walking) with a USB-C adapter than straight-through 3.5mm.

I really miss having a 3.5mm output on my phone...

You don't need every programmer to leverage the architecture for the market to accept it, just a few that hyper-optimize an implementation to a highly valuable problem (e.g., ads or something like that).

I was running i3 on Arch before. Then, I moved to Gnome Shell and have been daily driving COSMIC for a couple months. I think you will like it. At least on Alpha there were a couple rough edges here and there, but no deal breakers for me.

> the jump would be similar to the one from hand-written punchcards/assembly to higher level compilers

I wouldn't. Compilers are not stochastic text models and they can be verified and reasoned about to a great extent.

AntennaPod is so great. I have hundreds of hours.

It's even available on F-Droid.

Some of mine include Bach's violin partita 3, Beethoven's 9, and Mahler 1.

What's the name of the tool where you can specify the structure of the song?

> > You have to be bad in order to discover what kind of good you want to be (or are able to be). > > Sounds like one of those pseudo-profound woo statements. > > Many people manage being perfect good, well-adjusted humans without going through a bout of "being bad".

The way I read it, that paragraph is not about whether anyone can be a well-adjusted human or not. Rather, it is about how specifically some people discover what kind of good they want to be by being bad.

In that sense, "you have to be bad" talks to those people specifically, and uses "have" to refer to that causal process.

(English is not my first language, so my interpretation may be very wrong.)

They too could --and should-- have said no.

Take a look at language tool. It's pretty good in my experience. I'm not a native speaker, though.

(this post was not checked with languagetool)

In that case they wouldn't be a viable alternative, would they. A train can go 500+ km/h, and with good planning they don't have to cost 10x, so you're premise does not stand.

> there were too many comments

I didn't even knew this was a thing. I write tons of comments thinking they could be appreciated in case somebody has to glance at the code.

This looks very cool! Happy to see genetic programming receiving more attention in the last few years.

> using derived words with the listener expected to do live backwards derivation

This sounds very interesting. Any examples?

Yes, this is (I assume) easy enough that it must be cheaper than outsourcing.

> concept that they strongly oppose, the commune

This makes no sense. Why would people that want to survive these end-of-the-world scenarios be opposed to the one thing that makes survival more likely?

I'm not doubting you, but it makes no sense to me.

A counter-intuitive part of the reluctance of employers to keep work-from-home, is that offices costs them money. So, if efficiency stays the same (which seems to be the case), then they should want to keep work-from-home. So why do they want to go back? Is it all about control?

It's great to see so many apps for Lemmy. RedReader for Lemmy would be neat.

This a bit of a false dichotomy: genetic algorithms need not forego of directed or local search (for example, memetic and hybrid algorithms mix gradient search with populations).

Also, you have to keep in mind, it is often very hard to frame problems in a way that makes linear regression or gradient descent practical.

According to the article 38% see a correction in the near-ish future.

Also,

> Unlike during the dot-com bubble of the 2000s, AI isn’t entirely based on speculation.

I'd say the dot-com bubble was backed by a revolutionary product: the Internet. That doesn't change that expectations were too high.

The answer is that the traces are not super easy to find, depending on how we characterize what traces are in the first place. The question posed is known as the "Silurian Hypothesis":

https://www.scientificamerican.com/article/could-an-industri...

Very interesting, thanks for sharing. I would be interested in reading more on gradient-free optimization applied to large problems (like LLMs).