lagrange77

I can also recommend Steve Brunton's playlist [0] on the topic of physics informed machine learning, as well as the book 'Data-driven Science & Engineering' [1] by him and Nathan Kutz.

[0] https://www.youtube.com/playlist?list=PLMrJAkhIeNNQ0BaKuBKY4...

[1] https://databookuw.com

Neural ODEs are also interesting.

Mathematica (in Python via WASM in Jupyter) in the Browser?

Original URL: https://www.heise.de/en/news/Meta-s-head-of-AI-Yann-LeCun-do...

Yes, but it is rare to see \times used for scalar multiplication, isn't it?

> Put a hook on the x to distinguish it from a times sign [...] In 3rd-semester calculus and onward you’ll be using the times sign quite often.

I wonder if he's talking about the cross product.

> But I think a reasonably compentent Python/JAX programmer can roll out whatever they need relatively easily

You mean in terms of the ODE stuff, Julia provides?

Nice, thanks!

> Where do you get that?

From "when you get low enough but still riding on an external abstraction".

> He is saying implement only the parts you use.

Thanks.

Very cool!

I love the texture of the enclosure. I guess that's not possible with an FDM printer, right?

> but still riding on an external abstraction, you rewrite that, but ONLY to support the layers above you.

i don't get it. Why do i stop before stripping all abstractions?

> doesn't seem invalidate that more compute wouldn't translate to even _higher_ capabilities?

That's how i understand it.

And since their current goal seems to be 'AGI' and their current plan for achieving it seems to be scaling LLMs (network depth wise and at inference time prompt wise), i don't see why it wouldn't hold.

I hate having to rely on these drip-feed vague statements to gauge the fate of the planet.

Thanks for your perspective!

> Yes, lifelong learning is an enabler for that, which is why I am doing a part-time PhD.

I'm thinking about that, too!

> I am fairly upset about the bullshitting and general cheating and dehumanising that the current hyped 'AI' is enabling.

Oh man, i am, too. And i have the fear that enshittification will only increase.

> we don't need to be using it to destroy the planet and people's sense of self-worth

We absolutely do not, but i guess they will, because it promises profits.

Are e-ink displays especially bad for the environment, or are you talking about electronics in general?

While i'm with you in principle, over the years i've learned that we should not talk down good UI/UX, if that's what the wrapper adds. It's a crucial component to the value of a piece of software for the end user.

> Given more compute and enough recursiveness we should be able to reach any kind of result from the LLM.

That was always true for NNs in general, yet it took a very specific structure to get to where we are now. (..with a certain amount of time and resources.)

> thinking that computations in vector space are easy to make turing complete if we just have an eigenvector representation figured out

Sounds interesting, would you elaborate?

Wow, you just solved the ethics of technology in a one liner. Impressive.

You're safe, i guess, at least if you team up with the trailer voice guy.

If their mission is the wellbeing of their peoples, they should take any action that ensures that.

Besides regulating the technology, they could try to protect people and society from the effects of the technology. UBI for example could be an attempt to protect people from the effects of mass unemployment, as i understood it.

Actually i'm afraid even more fundamental shifts are necessary.

That implies that human intelligence is equivalent to AGI.

> You'll know AGI is here when the exercise of creating tasks that are easy for regular humans but hard for AI becomes simply impossible.

That's the most plausible definition of AGI i've read so far.

> These models are tools to help engineers, not replacements. Models cannot, on their own, build novel new things no matter how much the hype suggests otherwise.

But maybe models + managers/non technical people can?

I hope governments will finally take action.

Which implies that a few years later they will not become senior SWEs either.

What's wrong with that statement? Can you access the filesystem without a dialog?

Nice, congrats!

Interesting decision, to put the self hostable version behind the lifetime plan. The strategy i see most of the time for self hosting is a free and open source 'community edition'.

May i ask how you prevent piracy?

What makes it better than any other rotating LED matrix POV display?

That's a really interesting project!

In the docs you say > Currently, there is support for GSL, forward Euler, and exact integration

Does 'GSL' refer to GNU Scientific Library, which you use as a backend?

Oh, like the gas stations in TaleSpin.