
Hacker Newsroom for 03 November: Astra Edits Model Weights, Ubuntu Patch Delay, OccupySaaS Software Ownership
- News bulletin
- Technology
Hacker Newsroom for 03 November recaps major Hacker News stories, moving through astra edits model weights, ubuntu patch delay, occupysaas software ownership. Chapters 00:00:00 — Intro 00:00:14 — Astra Edits Model Weights 00:02:15 — Ubuntu Patch Delay 00:04:04 — OccupySaaS Software Ownership 00:06:05 — Closing 1. Astra Edits Model Weights Astra is editing other neural networks directly, and a new research post calls the technique weight-space programming. The experiments ask the model to inspect the numbers inside a network, identify what it does, and sometimes change those numbers without gradient descent. Story link Hacker News discussion 2. Ub...
Opens the player: the full transcript, every word clickable, with translation and flashcards.
Is this episode right for you?
C1 · Advanced
Fast, dense and idiomatic
- 76% of the words are B1 or easier.
- 147 words a minute: a natural pace.
- 78 harder words (B2 and up) in the whole episode.
Try it: tap any word
Real sentences from the start of this episode. Every word is marked with its level.
Let's get started.
,
a new research post calls the technique weight space programming.
Tap a word to see its meaning in your language.
Do you know these words?
8 harder words from this episode. Tap one to see what it means, then mark it.
Mark a few words and we'll tell you how this episode fits your level.
Is this episode right for you?
C1 · Advanced
Fast, dense and idiomatic
- 76% of the words are B1 or easier.
- 147 words a minute: a natural pace.
- 78 harder words (B2 and up) in the whole episode.
Transcript
The opening of the episode. The whole transcript is in the player.
harder words (B2+)
Welcome back to Hacker Newsroom. Today we're covering Astra can read and write weights to improve QN, WTF Ubuntu, and occupy sass the stories that rose to the front page and what people made of them.
Let's get started. Astra is editing other neural networks directly, and a new research post calls the technique weight space programming. The experiments ask the model to inspect the numbers inside a network,
identify what it does, and sometimes change those numbers without gradient descent. In the headline result, the researchers report that editing just 0.016% of a small QN model's weights raised its math 500 score from 7.9 to 48.1% with the same sampling settings.
That was the opening. The rest of the episode (5 min) is in the player: every word clickable, with translation, saved words and flashcards.
Continue in the playerAutomatically generated from the audio. May contain errors. Audio belongs to its publisher and is played from their feed. Rights holders can request removal — copyright & takedown policy