I think this is the best post I've ever made.

dude !! love interactive articles and guides. i was searching for some inspiration last night for a piece i am writing. this is great.

Thank you! ❤️ what're you writing about?

creative vibe coding, prompting methods & my overall building process.

ngrok are hiring someone to come take ownership of our video content. I want to change the way the industry does developer content. Come change the game with me.

quantization is genuinely underrated. ran some quants locally on ollama and the quality delta from full precision is so small you barely notice it. 4x smaller for almost same output is kinda insane tbh

Right??? I was genuinely surprised by that.

What!? How did I miss this on my tl until tonight. Great work!

damn outdid my iframes

@CatAstro_Piyush beautiful

@simonw Beautiful visualizations! Well done @samwhoo!

the real win isn't speed or size it's that your MacBook just became an inference machine

Man, you are just consistently on another level 👑

All about putting the work in. Appreciate you, man.

Sincerely, thank you for taking the time to do this. Very insightful.

You’re welcome 🫶

stealing this 🔥 style for my future posts i'm never gonna write

Write them! Send them to me.

Well done! Would be great if everyone who has written Quantization on Twitter would read 😂

It’s truly web-native knowledge sharing.

@threepointone wow this is beautiful. i've been building my own visdev tools for exactly this purpose but you knocked it out the park yet again. like you and Catherine Yeh are the only people I know going this deep on ML visdev and I genuinely think we'd make breakthroughs faster w/ more of it.

@threepointone Catherine’s work is out of this world. Huge honour that you’d mention mine alongside hers 🙏

wow, it's really good!! Do you do a lot of AI-related posts? Would love to support

Thanks so much! We've also done one on prompt caching: We'll likely do more in future, too.

@SamCRichard would love to talk to you about this :)

@julien_c I've been emailing you for years Julien ;) Anyways, would love to chat ai gateways x huggingface if that's your remit

@samwhoo Hey! We’ve been paying for ngrok for a long time and are a happy customer But these days we are trying to upsell rather than being upsold 😂

That's really cool and a very nice post. It reminds me of learning all the integration methods like Runge-Kutta and Adams when I was studying optimization in grad school. The most interesting thing to me is how the tradeoff of accuracy and usefulness. A lot of times in academia you publish a paper about a 2% improvement in some accuracy or performance metric, but in the real world you can be severely limited by things like available memory or the latency of the computation itself.

Damn this is good.

So freaking good.

🫶

I’ve come across plenty of posts on this topic, but nothing at this level. Thank you for creating and sharing this! It’s genuinely insightful. And thanks to the algorithm for leading me to this gem.

hella cool. thanks for sharing

Too good

This is really great. Thank you for this!! I would love to read more from you

Thank you, that means the world to me. I have another post like this on prompt caching: I have also written 9 more of these on my personal site:

Incredible post, thank you.

I love that! Hope for more posts!

Looks freaking gorgeous!

Incredible blog, nice work

Better than my attempt a while back. Nice job.

Sam, this is gem See the rhyme?

Best quantization article I've come across in a while. What did you use for interactive widgets? quite clean look.

The design comes from ngrok’s The rest is a mixture of SVGs, canvases, DOM manipulation.

Charts for concepts was great. Thanks!

@steveruizok I understood why this works in like 60 seconds because of your post. Amazing.

any post you made is the best post

Wow, mind blowing elegance. Great blog post!!

@CatAstro_Piyush man love those visual interactions

this is fantastic