@seekjourney: I made an interesting hand-drawn illustration to summarize what the paper is about, with the complete prompt: Create a …

I made an interesting hand-drawn illustration to summarize what the paper is about, with the complete prompt:

Create a header image for this article. Translate the article’s content into a “hand-drawn technical research notes” style poster: off-white paper texture, black pencil line drawings, minimal blue and red accent colors. Center it around a large core technical diagram, incorporating flowcharts, axes, formulas, code structures, decision trees, data charts, and handwritten annotations to convey ideas. Overall, it should feel like a researcher deriving a complex system on paper: rational, restrained, with ample white space, high information density, and a slight irregular hand-drawn vibe. Prohibit 3D, gradients, cartoons, commercial ad vibes, or PPT style. Automatically extract the content and decide on the most suitable visual structure.

极客杰尼 (@seekjourney): 一个很强的 RSI 新信号可以关注一波：

Google Research 开始研究 Agent Harness 的 Recursive Self-Improvement。

不重新训练模型，直接让 prompt、工具、memory、control flow 自动迭代。

这条路线我很看好。

因为相比之前 glm 发布的 model-level RSI 更清晰了 ：