Qwen 3.8 27B reads and edits model weights on a laptop
Original title: Qwen 3.8 27B on my laptop can interpret model weights, too
Simon Willison says a local Qwen 3.8 27B reproduced a paper example previously shown with GPT-6 Astra.
The paper Sparks of Weight-Space Programming reports that frontier models, especially GPT-6 Astra, can interpret and modify raw neural-network weights supplied as text. Willison writes that Qwen 3.8 27B on his laptop handled one of the paper's examples. The note is a short personal experiment, and the abstract excerpt is cut off.
Unconfirmed: The laptop result is the author's own experiment and is not independently verified here.
- Named authors are Rikiya Takehi, Jayden Teoh, Alex Zhang, and Omar Khattab.
- The paper coins the term weight-space programming.
- Models receive a text listing of floating-point weights.
- Willison says he tried one example from the paper.
From the original article
Yesterday's hot paper was Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights by Rikiya Takehi, Jayden Teoh, Alex Zhang, and Omar Khattab. The paper found that frontier models, in particular GPT-6 Astra, can both interpret and modify the raw weights of a machine learning model - literally a text description of the model containing a bunch of floating point numbers.
simonwillison.net
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.