Advertisement ยท 728 ร— 90

Posts by Mark Gaiser

Sticky business, a name that sticks around.

5 months ago 0 0 0 0
https://www.visualcapitalist.com/the-worlds-biggest-empires-of-history-on-one-epic-visual-timeline/

Empires rise. And fall. Interesting to see that we don't have big empires anymore. Not on the scale of the Roman or ottoman empires.

https://www.visualcapitalist.com/the-worlds-biggest-empires-of-history-on-one-epic-visual-timeline/ Empires rise. And fall. Interesting to see that we don't have big empires anymore. Not on the scale of the Roman or ottoman empires.

Greece had an empire? ๐Ÿคฃ Yeah that's hard to imagine now. While this depicts empires, Europe's imperialism wasn't an empire so that's a bit weird in the picture.

5 months ago 3 0 0 0

4/4 So AGI? No. Not anytime soon. This bubble is gonna burst and that is before AI becomes AGI. But we might get much faster and much more efficient AI and that might be the groundwork for new inventions.

5 months ago 0 0 0 0

3/4 And then there is the computational cost. AI is exponential in big o time complexity. If it can go to quadratic or even log linear (that's quicksort kind of complexity) then you at least have a massive computational win that opens doors.

5 months ago 1 0 0 0

2/4 These two questions are a massive NO right now. It can't. And we need very fundamental new inventions to make these possible. We're not there by far thus we're not at AGI levels by far either.

5 months ago 0 0 0 0

1/4 oke, AI. AGI specifically. Ask yourself the following questions: can ai learn and retain that knowledge at inference time (that's when you type)? Can ai be curious and think, quite literally, out of the box?

5 months ago 0 0 3 0

Isn't all that extortion grounds to invoke article 25? All this stuff is short term strongman play but long term... Boy, that might hurt the US a lot! Ugh, after that 25th amendment you get the JD idiot... Can't you have snap elections? ๐Ÿ˜‚

5 months ago 1 0 0 0

I'll happily hear an independent party echos this sentiment. Proton is obviously biased in this. One might call it marketing ๐Ÿ˜‰ (For what it's worth, I expect this from Google and it's time for me to start ditching it)

5 months ago 0 0 1 0

. @bsky.app underneath each post there is the chat icon. It shows the number of replies. Why... WHY... does clicking that icon NOT show me the replies? I keep falling for that. Annoying ui choice... (Or a bug?)

5 months ago 1 0 0 0
Advertisement

I'll take the liberty of using a selection of your words. "If I were a real person I wouldn't be using Microsoft" ๐Ÿ˜‰๐Ÿ’ฏ

5 months ago 1 0 0 0
Azores, Sao Miguel. Nov 3rd 2025 and it's about 20ยฐc, not bad for fall/early winter. These are rare days though. It is the rainy and stormy season.

Azores, Sao Miguel. Nov 3rd 2025 and it's about 20ยฐc, not bad for fall/early winter. These are rare days though. It is the rainy and stormy season.

Just going for a walk on an average day in the middle of the atlantic ocean ๐Ÿ˜Ž too bad this is gonna end when I get a new job (jan 1st 2026). Work from home: ok. Digital nomad lifestyle? Yeah, that's gonna end.

5 months ago 3 0 0 0

That sounds like a wonderful optimization! Hope you get it, fingers crossed ๐Ÿคž๐Ÿ’ฐ

5 months ago 1 0 0 0

"aliens are real" along with the "up his shorts" comment did it for me ๐Ÿ˜‚

5 months ago 1 0 0 0
Preview
GitHub - johannschopplich/toon: ๐ŸŽ’ Token-Oriented Object Notation (TOON) โ€“ JSON for LLM prompts at half the tokens. Official spec & TypeScript implementation. ๐ŸŽ’ Token-Oriented Object Notation (TOON) โ€“ JSON for LLM prompts at half the tokens. Official spec & TypeScript implementation. - johannschopplich/toon

github.com/johannschopp...

Interesting little project to have a more condensed LLM friendly "json" format. This is great, new more condensed but still readable data formats!

5 months ago 1 0 0 0

2/2 and once you have that conversion engine you might be able to make an ai engine that converts markdown back into full fledged websites ๐Ÿ˜ฎ this creating a sort of web engine proxy that is simpler. Just rambling here, but that seems plausible to me.

5 months ago 1 0 0 0

๐Ÿค” I'm not so sure. What you see is ai and content being converted to markdown and from there parsed by ai. An "engine" might well be a conversion between the vanilla web to an "ai friendly" web? That would also be a lot more beneficial in terms of using web content, no?
1/2

5 months ago 2 0 1 0

Nice! Thank you ๐Ÿ‘ these events seem to go completely under the (web3) events radar. Looking into those ๐Ÿ˜

1 year ago 1 0 0 0

Ohh @bmann.ca you might know? Do you know of any decentralized conferences this year that dive into tech? Like IPFS thing in recent years? I can't find any...

1 year ago 1 0 1 0
Advertisement

It looks delicious ๐Ÿ˜‹

1 year ago 1 0 0 0

In a couple months you'll say "I wish it was only down by 13%" ๐Ÿ˜‰ we are due for a recession in normal market conditions. But this accelerated economic destruction is probably gonna hurt. A lot.

1 year ago 1 0 1 0

Oh my! We're going from flat earth to hollow earth. Where do these hollow brainless idiots come from?....

1 year ago 0 0 0 0

Isn't this just PL dev guild? Or filecoin grants? Or IPFS Devgrants? Or to a lesser extend Funding the commons? You see where i'm going? Sure, they all have their own details that all slightly deviate. Sticking to one program is not an option?

1 year ago 1 0 1 0

They are in room 42

1 year ago 0 0 0 0

Ha, you can see his sleep pattern (kinda) in these dots.

1 year ago 3 0 0 0

It's because of your nice humorous educational videos! Well done and definitely keep it up! ๐Ÿ‘

1 year ago 1 0 0 0

Ha nice! Congrats and well deserved! ๐ŸŽ‰

1 year ago 3 0 0 0
Advertisement
AMD Technical Information Portal

And here's how you can enable it in pytorch docs.amd.com/r/en-US/5730...

1 year ago 1 0 0 0
Preview
AMD ZenDNN 5.0 Software For AI Delivers "400% Performance Uplift" Released last November following the AMD 5th Gen EPYC 'Turin' server processor launch was ZenDNN 5.0 as their deep neural network library optimized for EPYC/Ryzen processors

You seem to be in the perfect spot to use zendnn (zentorch) which just a couple days ago got a massive performance improvement (around 400%). www.phoronix.com/news/AMD-Zen... no clue how you'd hook it into ollama though.

1 year ago 1 0 1 0

#bluesky is there a feed that combines all the feeds I'm subscribed to? Or is that the "for me" feed?

1 year ago 1 0 0 0

As a Linux user, this bug definitely makes me grin ๐Ÿ˜ hope your flight to nowhere lands somewhere ๐Ÿ˜‚ (joking aside, I'm amazed at how centralized the world still is! We can't even easily change it with better alternatives... Ouch!)

1 year ago 2 0 1 0