重點導讀
summary_zh: GPT‑6 Astra 能直接讀取模型權重,最新論文《Sparks of Weight‑Space Programming》證明此功能,作者亦測試 Qwen 3.8 27B 可在 45 分鐘內達成相同效果。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
S
Simon Willison (@simonw)@simonw
·19 天前
* There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing simonwillison.net Qwen 3.8 27B on my laptop can interpret model weights, too 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 …
推文來源 · X (Twitter)在 X 參與討論
來源: Simon Willison (@simonw) (x.com)
在 X 查看推文