hadjian

Thank you bertails for the post. I work in IoT and we are also moving in that direction. I was of course hesitant, because of the criticism of semantic web on the ...web ;-), but this approach is used very successfully in our market for many years:

- https://haxall.io/ basically a reinvented copy of sem-web, but so much nicer from a dev perspective

- Used by https://skyfoundry.com/product

- https://brickschema.org/ the new star, built on sem-web

- Used by https://www.mapped.com/ where one of the founders worked on Brick

- https://docs.open223.info also a model that has caught interest in the market

- https://qudt.org/ de-facto standard for units and conversions

We started just a couple of months ago, but I have not regretted the decision, yet.

The downside of a lesser known tech-stack is always the missing OSS tools, e.g. for authoring, ownership management, review workflows etc. Counting on you guys to open source some good stuff ;-)

When we are further down the road, I am interested to get in touch with you guys, if you are up for it?

edit: forgot the ontology part of the haxall project.

- https://www.project-haystack.org/doc/index

- Kitty terminal. My daily driver.

- Wikipedia

- Used to donate to visidata and will probably again, if I need it for work again

This article lists likely candidates for three categories and only one even contains Prevost.

I think the value lies in the mapping of heterogenous sources to one network that can be queried. How you arrive at conclusions derived from the information is human interpretation.

I’d be interested in how the facts were validated against the sources. Like debugging the mapping process.

This post is a treasure. I have too many itches to scratch and would want to know at least a ballpark figure for the time investment.

One suggestion for the already complete and detailed post: what is the personal/professional gain you get for your time investment? In this case maybe: „program design became clearer after having worked through SICO.“ or „I can discuss design more formally now“ or just „It was interesting and everyone talked about it, so I needed to know“

I was referring to aipapersacademy and not the arxiv link.

Also I didn’t mean the occurrence of the word „coconut“ in the paper, but thanks for counting.

I meant their publication site: https://ai.meta.com/results/?content_types%5B0%5D=publicatio...

The video is something I’d expect from AI.

If this site didn’t appear here, I’d think it’s a scam:

- site claims to simplify papers, but movies AI generated

- full of ads

- Can’t find „Coconut“ on the official Meta FAIR page

Is this the best site to link to?

I respect you for that joke.

All the prices for DIY robots seem to be defined by the servo motor prices.

I will start by building the Koch robot arm with servos harvested from my kids toys. Let’s see how good that works.

Yes. I was really happy, when I discovered GPT3, because I could ask infinit „but why“s without being worried about annoying the tutor. Would be great, if the LLMs wouldn’t be such push-overs. I’d like to hear more „you’re wrong!“.

I agree, but I also hope for several revolutions in this space. Everything around educational offerings is horrible beyond compare:

- Classroom learning doesn’t cater to individuals

- Corporate learning programs that I came across were 100% useless

- Personal learning is unbelievably hard to organize (persistence, learning path, real world application…)

I had basically five wonderful experiences:

- Karpathy on YT ;-)

- Dash and Dot from Wonder Workshop

- Crafting Interpreters

- Gilbert Strang

Hope eureka labs becomes the sixth. Good luck Andrej!

Potentially stupid question: did she work with them and watched them live or is there some kind of recording?

Andreas worked professionally on WebKit at Apple and succeeded in writing SerenityOS from scratch.

So he is in a good position to decide on the approach and has proven he can finish things. I’d just trust him on this, if my opinion was even relevant.

But as it’s a for fun project, he can write a browser in Fortran and abandon it half-way ;-)

Did you really watch all videos in the playlist? I am at video 4 and had no background in PyTorch or numpy.

In my opinion he covers everything needed to understand his lectures. Even broadcasting and multidimensional indexing with numpy.

Also in the first lecture you will implement your own python class for building expressions including backprop with an API modeled after PyTorch.

IMHO it is the second lecture I can recommend without hesitation. The other is Gilbert Strang on linear algebra.

I thought the same thing. Most of them from Sharif, which is Iran’s equivalent of Stanford.

Congrats Saul! Hope it’s ok to solve this during Christmas, because of vacation ;-)

His jokes are hilarious!

Yes, that’s basically the algorithm, but it took me a while to understand why it works.

The subtraction from base^n is the radix complement, the subtraction from base^n-1 is the diminished radix complement.

Example base 9: with a fixed number width of say four, e.g. 4781, you take the radix complement by subtracting from 9^4, which is 10000 in base 9. But to take the complement, you again need subtractions with nasty borrows. But you wanted to reuse your adder and not have a subtractor to take the complement.

So you observe, that the formula is the same as 10000-1-4781+1 (subtract one and add it back). Subtracting the one gives you 8888. this solves your problem, because every digit is the highest one in its base, so you will never have to borrow. This is the diminished radix complement, to which you add one to get back to the radix complement.

Back to base 2: you take the diminished radix complement to avoid borrows (which is conveniently just a bit flip in base 2) and add one to arrive at the radix complement. Thus the familiar algorithm: „flip all bits and add one for two‘s complement“.

I hope I did a better job explaining it in my post linked above.

Had the two‘s complement journey a couple of month ago:

http://www.hadjian.com/articles/20221229_twos_complement.htm...

If you can answer why two‘s complement has an apostrophe and ones complement doesn’t, you understood it ;-)

Wow. Your mom is a wizard.

Thanks for the hint, that Geron has books on machine learning. I know him from a YouTube video, where he explains how they use the google knowledge graph for YouTube: https://youtu.be/ywfYMQD6yYk

He explains it so well. Then his books are worth a read.

That’s great to hear. I will take you up on your offer.

Also I’ll check out your product in more depth.

BTW, what’s your mom‘s background? Sounds like awesome parenting

Congratulations on the launch!

I work at Siemens and we need to integrate lots of brownfield devices. We use the W3C Web of Things standard for representing devices: https://www.w3.org/TR/wot-thing-description11/

Would love to hear your thoughts on this.

An example from Deutsche Telekom: https://github.com/w3c/wot-cg/blob/main/Presentations/2022/2...

I think the author of OpenHAB is on that team.

Ah, the release notes still say: „ and experimental support for Debug Adapter Protocol (DAP).“

They’ll get there eventually.

Does debugging work? I think the dap support was experimental.

Can somebody share his/her experience?

Oh yes, 3b1b is my favorite. All of them, but especially Fourier series.

That sounds about right. I read the book and watched the lectures after I graduated. And it was just fun.

The thing is that the ideas stuck and I was very grateful for that.

But we can be lucky that there are so many approaches out there to pick from.

So strange. It was the best book I’ve read about the topic. It’s been a while, but I don’t recall anything not presented in the right order. Going from linear equations to a geometric interpretation of the rows, then to linear combination of the columns. Then Gauss-Seidel to LRU.

I liked his approach of “ideas first, rigor later”. I think after reading this book, you can easily grab a book with more formalism, if you feel lacking rigor.

I’m interested to understand where you felt the order was wrong?

This was also my first visit in trying to fill my gaps. He has a very nice practical approach, but the difference between the ones’ complement and the two’s complement was missing.

Also the encoding was just given, so the “subtraction by addition” part was missing.

I wonder if I will find the time to build his CPU. One of the best resources on YouTube, really.

I might come off a bit naive, but I didn’t picture the data collection being _this_ precise.

Now I’m interested in my own data. Are those companies required to offer a means to download your own data?

Did anyone do that and if so, was it readable?

Couldn‘t agree more with the author and in fact, I started to code on the C64 with my son around three years ago, for exactly that reason.

There are educational coding environments, but you will never understand the underlying computer model, just by playing with them.

And no setup whatsoever. Python was supposed to be that simple, but we’re past that, obviously.

Hm, strange. As I remember, eglot’s design goal was zero configuration. Could you elaborate a bit? For context, I use go, had gopls installed and just had to start eglot. All xref shortcuts worked immediately.