Research proposes Weight-Space Programming method
Understand this story
Researchers have proposed a method called "Weight-Space Programming" that lets language models directly read and rewrite the raw weights of other models without the usual training pipeline of step-by-step mathematical optimization. The Decoder reports that GPT-6 Astra adopted this method. Researchers compare the approach to a mechanic turning knobs one by one, replacing expensive retraining. Experiment logs have been published on GitHub, and the rewritten Qwen model has been released on Hugging Face. The researchers also acknowledge that capabilities remain limited for now, and the next step is to build benchmarks for systematic measurement.
AI overview based on source reports · Updated 2 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- The DecoderFeaturedGPT-6 Astra sees the matrix and reprograms other AI models directly through their weights
Researchers describe "Weight-Space Programming," in which a language model called Astra reads and edits the raw weights of other networks without step-by-step training: it identified which of four games a network was trained to play, generated untrained game-playing weights that beat a standard baseline in Breakout (517 vs. 25 points), and changed 0.016 percent of Qwen3.5-0.8B's weights to raise MATH-500 accuracy from 7.9 to 48.1 percent. It could not measurably improve Gemma 3 270M, and Jürgen Schmidhuber called the approach "old hat," while researcher Omar Khattab said the novelty lies in how well today's language models translate between human language and a network's numerical values. Experiment logs are on GitHub and the edited Qwen model is on Hugging Face.
Activity for this story
- Current comparable activity
- 3
- Peak comparable activity
- 10Nov 5 01:00
- Change in the past 24 hours
- -50%
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.