News for hackers. Stuff that matters.

7777777phil

Re:Astra can read and write weights to improve Qwen on Monday November 02, 2026 @03:15PM (#50343039)
A similar experiments found that Claude models can write entire, functional neural net weights from scratch: https://csinva.io/weight-space-programming-autoresearch/anal...
Ratio to average wages is 0.9 last with Switzerland.
I love using hugo. Also wrote about some time ago how I got a perfect 100/100 Lighthouse score with Hugo and plain CSS. Feel free to have a look: https://philippdubach.com/posts/hugo-lighthouse-100/
I been part of many discussions on how access to the internet should become a fundamental human right, like access to clean water. Maybe at some point we will have a discussion like that about access to AI?
Re:Voxtral 3 on Tuesday October 20, 2026 @12:44PM (#50169104)
I have been using different Parakeet models for STT over the past months and been pretty impressed with the local performance.
One of my AI predictions for 2030 was that AI agents will talk to agents [1] and this seems to be the perfect example of that.

> Customers can generate detailed requests and complaints continuously, at very little cost.

[1] https://philippdubach.com/posts/ai-in-2030/

Re:The most life-changing books, statistically on Friday October 16, 2026 @01:54PM (#50121859)
I thought the idea of scraping the book reviews and looking at them from a statistical pov was neat.

edit: also apparently the full list is here for free: https://docs.google.com/spreadsheets/u/0/d/1a3bkcdypddcxhtYW...

Older ยป